Talk:Body roundness index

Body_roundness_index_graph.svg

Some thoughts on :

1. I love the graph. It is simple, elegant, pretty. Well done!
2. Solved: Why only cm? Why not also have feet and inches, like on ?
3. Solved: Why show BRI for 0.5, 1.5, 2.5 but not for 11, 13, 15? Alternative: Show BRI lines for 0, 1, 2, 3, ... ... 18, 19, 20, show numbers only for 0, 5, 10, 15, 20.
4. Solved: The colours are pretty, yet don't tell anything at the moment. Could the areas be colour coded, green for normal, yellow for alert, red for alarming, similar to ?
5. Solved: Why do axes start at 0? Why not limit height and waists to realistic values?
6. The c and h are a bit mysterious. Spell them out to circumference and height?

Uwappa (talk) 08:44, 1 October 2024 (UTC)

1. Good critique, Uwappa. Note BMI in your comments should be BRI.

Attention for Cmglee, with thanks for the excellent design! Minor issue for the graph notes - the article uses American English spelling.

2. For ease of introduction and use by the general user, I would prefer the box indicating Part of a series on be above the graph at the top right.

3. Under the calculator box, Doc James, should the source be given in small font, such as Source: Wikimedia Medicine? Zefr (talk) 14:41, 1 October 2024 (UTC)

Corrected BMI to BRI, thanks. Uwappa (talk) 14:44, 1 October 2024 (UTC)

Response from illustrator:

1. Thanks! Please update the description page as needed.
2. As it's a ratio of two lengths, they can be in any units, as long as both are the same. How can I make it clear in the graph?
3. As the lines get closer for higher BRI while the number of digits get higher, I had to skip values as BRI increases.
4. The colours make it easy to tell the values apart. If there's an official guidelines of which values are healthy etc, I can change that.
5. Starting from 0 shows that all lines intercept the origin.
6. I wanted to make the diagram language-agnostic. Their meanings are in the caption.

cmɢʟee⎆τaʟκ 16:45, 1 October 2024 (UTC)

2. See for two units on one Y axes. cm to the left, feet and inches to the right. Same story can apply to x-axes.

3. Yes, I do see how you run out of space. Yet, the graph now fails to graphically communicate that the lower values lines are far apart and high value lines are closer together. The proposed alternative is to show lines for all integers yet go low on the number labels. shows only numbers for 5-folds. You could to the same.

I like that. As there's a large gap between lower integers, should I use dotted lines for the x.5 values? Without labelling every line, I can continue the dotted lines through the whole range too. cmɢʟee⎆τaʟκ 21:43, 2 October 2024 (UTC)

4. So this is a question to pass to the medical experts: What values define BRI categories, similar to Body_mass_index#Categories? I would like to see those BRI categories in the article so the numbers gain meaning. What is healthy? What is not?

Uwappa - Given in Table 3 of the original Thomas ref, the range found in the database studied was 1-18, with "normal" as 3 to about 6 (median around 5.4). We state this in the lede. Under 3 and over 7 increased the risk of all-cause mortality; shown in blue).

Cmglee - would a newly-drawn ellipse (or 3 overlapped showing low-middle-high roundness), similar to the one shown in Fig. 3, green, here be feasible? Zefr (talk) 21:25, 1 October 2024 (UTC)

@Zefr: Showing the ellipses overlaid on a body silhouette is a smart idea. We could either have two diagrams, or the BRI lines of my chart could be labelled with them. I'm unsure how to render realistic silhouettes with different waist circumferences without simply horizontally stretching a generic silhouette, which would also e.g. stretch the head. Any ideas? cmɢʟee⎆τaʟκ 21:28, 2 October 2024 (UTC)

cmglee - see the silhouettes and corresponding ellipses of body shapes in dashed green (normal/healthy), solid blue (lean, relatively tall), and dotted red (too round, same height as "normal") at the bottom of Fig. 2. Similar body outlines seem a good place to start. Zefr (talk) 21:54, 2 October 2024 (UTC)

Yes, but I should not trace the non-free images. Would you know of an image set similar to File:Obesity_&_BMI.png with frontal and more views? Even better if I can parametise the waist. cmɢʟee⎆τaʟκ 00:27, 3 October 2024 (UTC)

P.S. File:Body_Mass_Index.jpg is better. cmɢʟee⎆τaʟκ 00:29, 3 October 2024 (UTC)

Cmglee - This second image for underweight vs. healthy vs. obese would give a good impression of roundness as a silhouette. Using the article calculator at an equal height for all three of 178 cm (5'10"), one can give waist circumferences for each classification:

• underweight, 85 cm (34"), BRI 2.95 (< 3 category)
• healthy, 105 cm (42"), BRI 5.20 (3-7 category; "green zone" roundness in Fig. 3 of Thomas)
• obese, 130 cm (52"), BRI 8.71 (> 7 category; waist circumference giving impression of greater roundness)

Zefr (talk) 01:58, 3 October 2024 (UTC)

Other silhouettes from the Obesity article to consider for comparing roundness. The drawing was provided by an FDA editor, who stated for the image file (data from a 2000 FDA public advisory): "the "healthy" man has a 33 inch (84 cm) waist, the "overweight" man a 45 inch (114 cm) waist, and the "obese" man a 60 inch (152 cm) waist." These waist circumferences seem more realistic for the graph parameters. Zefr (talk) 15:23, 3 October 2024 (UTC)

Good find. That's better as it's already SVG. I'll just stretch the figures for intermediate values. cmɢʟee⎆τaʟκ 15:40, 3 October 2024 (UTC)

5. Yes, the lines do intercept at zero, how interesting for mathematicians. Now: How many of us have a height and waist of 0cm?

If you want work done, that attitude is not going to win you any favours. cmɢʟee⎆τaʟκ 23:50, 1 October 2024 (UTC)

Wow, that sounds like 'mathematician' would be an insult. Really? Sorry, it was not meant to be insulting. What I would like is to have a chart that is tailored to real life. A chart that tells a large group of people: What is a healthy waist size for your height? That could even be life saving. So I want that chart not just to be good, I'd like it to be excellent. Having seen some of your other work, you seem well qualified to take the chart to a higher level.

Would you like me to step in and work on the chart? Uwappa (talk) 09:33, 2 October 2024 (UTC)

@Uwappa: No worries and thanks for explaining. It's the tone of the second part: if you had specified ranges rather than implying that my choice of 0 was absurd, I would have taken it better. I'll use the heights (and style) in the BMI chart. I can update it this weekend. Cheers, cmɢʟee⎆τaʟκ 21:25, 2 October 2024 (UTC)

Heights of the BMI chart will be great. Thanks! Uwappa (talk) 05:06, 3 October 2024 (UTC)

6. Yes, the chart is good and certainly has potential to be of use in many languages. Shortening to one character works only for some languages. E.g. 'c' will work for French 'circonférence' and Spanish 'circunferencia' but will fall short for German 'Umfang', the Greek 'περιφέρεια', the Chinese '圆周'. For now, just go for English, polish the chart to perfection first and worry about proper localisation later. SVG files can be multi-lingual, no worries.

Uwappa (talk) 17:54, 1 October 2024 (UTC)

It is just the equation in question... Wiki Project Med does not own the source to that. Doc James (talk · contribs · email) 18:45, 1 October 2024 (UTC)

From experience with popular images like File:Moon_names.svg, it becomes a mess when some editors add translations in the SVG file (preferred) while others upload new files (any bug-fixes or updates to original do not propagate). I tried to remedy it by using as few language-specific labels as possible. Of course, h and c are initials in English, but that's a compromise used in much of STEM e.g. E = m c² or V = I R even if the terms don't start with these letters. Cheers, cmɢʟee⎆τaʟκ 21:37, 2 October 2024 (UTC)

Would annotations work with an SVG image? It probably will as it generates a PNG when used on a page. That would keep the language specific words away from the image. An English page could show 'height', a French page 'hauteur' and a Chinese one '高度'. Uwappa (talk) 18:15, 5 October 2024 (UTC)

An odd idea that might or might not work: Rotate the whole graph a bit anti clockwise, so 'normal' is straight up, too lean is left of the center and too round is to the right of the center. A bit like a tachometer or a VU meter , but with a red background both for too lean as well as for too round and a green center area for healthy.

Uwappa (talk) 08:26, 5 October 2024 (UTC)

I think having axis at oblique angles might confuse readers. cmɢʟee⎆τaʟκ 15:08, 7 October 2024 (UTC)

OK. Having seen your new version of the graph, rotation won't add much. And yes, it may make the graph hard to read, confusing. So I let this odd idea go. Uwappa (talk) 16:27, 7 October 2024 (UTC)

Update

Graphs of body roundness index vs height (vertical axis) vs waist circumference (horizontal axis)

I've updated the image roughly to the style of File:BMI_chart.png as on the right.

There was enough space to label every line, making it easier to tell which is which without colours.

To work around the i8n issue, I omitted the axis labels. The units should hopefully be language-independent enough. The caption can be as in the thumbnail but it should be obvious which is which.

I've yet to add silhouettes for selected values.

Let me know when you've finalised the tint ranges (note that SVG doesn't do conical gradients so please use solid hues).

Cheers, cmɢʟee⎆τaʟκ 15:06, 7 October 2024 (UTC)

It's a nice, informative graph as it is, Cmglee - thanks. Subject to further thoughts of Uwappa, Doc James or others, and following Tables 2-3 and Fig. 3 of Thomas, the "green zone" of healthy BRI should be the brightest green at 5 and fade lighter in both directions to 3 and about 6.6. From 6.6 to 16, the red hue should increasingly become brighter.

As the Zhang report indicated health concerns for low BRI, yellow could progress from light at 3 to dense at 1. Maybe this could be the starting ranges to allow other editors consider the graph's impression for visual effect. Zefr (talk) 15:59, 7 October 2024 (UTC)

I love it! Well done Cmglee, you just nailed it. Thank you!

The ranges for the axes worked out very well. The chart now offers plenty of space for all the lines and labels up to 20. Great!

The double axes labels for for cm and feet&inches are great too.

Could you have a go with Template:Annotated_image for axes labels? How about simple labels: waist and height? A ready-for-translation English example would be just too easy for translators.

I think it would be OK to be bold and colour code the lines now, with colours suggested by Zefr. My own preference would be red-green-red, with simple logic being: red = not healthy, green = healthy. By colour coding the lines (and not backgrounds) you won't need conical gradients and avoid the discussion of the precise impact of other variables. And if colour coding the lines might turn out to be too bold, well that will probably get the discussion started on what the colours should be. Colouring lines is easy to undo, no worries.

It is hard for me to follow a grid line from axes to the point of interest. Could the cm lines be thinner or a lighter shade of grey than the dm lines?

You may want to delete the black lines at the axes as they don't add anything. How about a grey background for the chart-area and have white grid lines as in ? A grey background may work very well with coloured BRI lines. Uwappa (talk) 16:17, 7 October 2024 (UTC)

For the color representing a BRI of 3 and below, consider lean athletes who would have BMI of <18.5 and % fat of 10-16% (gender differences, Table 3, Thomas). This is a reason for not representing BRI of <3 as red, but rather for the general population as a yellow for caution. Zefr (talk) 16:42, 7 October 2024 (UTC)

I am still a bit worried about 'just yellow' for too lean. Anorexia_nervosa should be a red flag.

Let's discuss line colours in detail. Feel free to update the table below.

The Body mass index graph offers good colors; Cmglee's superimposed line numbers would work well (right).

Uwappa - with ^ in the line numbers, I've offered brighter selections for the transition and key colors. 10-15 colors would transition to the brightest red. Zefr (talk) 19:48, 7 October 2024 (UTC)

Body mass index colors

• I am not sure about your colour 16, #FF0000. It is a very bright, 'happy' red. The alternative is something 'unhappy', 'unhealthy', 'gloomy', a darker shade of red like the colour of congealed blood, signalling something unhealthy to avoid, like #8E000E.
• For colour 5, the 'center of health' I would prefer a colour that is lively, a green similar to new life, the colour of new plants in spring. #32cd32 is a bit too dark for me, not 'happy' enough.
• I've moved your alternatives to an extra column and added transition colours you can select from. For me the shades of red between ff6347 to ff0000 are too hard to tell apart. Would it be OK for you to finetune the colours in your column and pick transition colours?

Uwappa (talk) 20:55, 7 October 2024 (UTC)

I suggest we use "traffic light" colors of red-yellow-green to be universally recognizable. For example, traffic light red is #FF0000, and traffic light green ("healthy body roundness") is #00FF00. Zefr (talk) 22:41, 7 October 2024 (UTC)

I think those bright colours would have been OK as line colours, but are way to bright to fill background of areas. I think the green yellow red idea is there already, albeit in pastel background colours. I am happy to let go of the line colour idea, am happy with the pastel background colours. Uwappa (talk) 23:13, 7 October 2024 (UTC)

BRI	Line Color	Zefr's alternative
1	#FFCC00	#FFCC00
2	#FFE97F	#FFE97F
3	#EBFF7F	#EBFF7F
4	#BFFF7F	#BFFF7F
5 ^	#80FF7F	#00FF00
6	#BFFF7F	transition via EBFF7F and ffCC00?
7	#EBFF7F
8	#FFE97F
9 ^	#ffCC00	#ff6347
10	#E9A503	transition ff6347 to ff0000
11	#DA8A05
12	#CB6D06
13	#BB5108
14	#AC370A
15	#9F1F0C
16 ^	#8E000E	#ff0000

Uwappa (talk) 18:00, 7 October 2024 (UTC)

Thanks for the compliments. I've added tints as above but reduced the opacity to 0.1 as they were rather saturated.

I've also been thinking that the silhouettes could be in a different file to not clutter up the graph.

What do you think? cmɢʟee⎆τaʟκ 21:06, 7 October 2024 (UTC)

P.S. I extended the scale to 20 and extrapolated the colours. cmɢʟee⎆τaʟκ 21:13, 7 October 2024 (UTC)

Instead of a lot of back-and-forth, can you try editing the file and opening it on a web browser? In the latest version, on line 26, remove

fill-opacity="0.1"

and on lines 36 to 56, amend the colours.

When you're satisfied, either upload the amended file, or paste here the updated colours. cmɢʟee⎆τaʟκ 22:46, 7 October 2024 (UTC)

• Ha ha ha, I think you should take it easy mate, slow down, sit back and relax as you're outrunning us. We were still discussing colours...
• BRI 17-20 are back in, good for people in the danger zone!
• I do not understand what "silhouettes could be in a different file" means. What silhouettes?
• I am not sure if the colours should fill the areas between lines. Filling areas seems a bit too much for me. I actually liked your idea of coloured lines in your first version.
• But... yes, the reduced opacity works well, generating pastel colours that work well for areas. The colours look like a gradient. The chart looks very nice now. I think you should go and improve . It now looks outdated, ha ha ha.

So... Zefr, I am actually quite happy with this result, how about you?

Uwappa (talk) 22:13, 7 October 2024 (UTC)

I do like this graph above, feeling only a brighter green would highlight Thomas' "green healthy zone" of roundness.

For the graph colors, is it possible to highlight the green in a brighter tone as used by Thomas? Zefr (talk) 22:11, 7 October 2024 (UTC)

 Done cmɢʟee⎆τaʟκ 20:00, 9 October 2024 (UTC)

Silhouettes development

Silhouettes of roundness from left to right: healthy, overweight, obese

Cmglee and Uwappa - Thanks for the work - I feel the silhouettes should be a separate file.

This is the silhouette graph. It's possible it could be used in the article as it is, with an informative caption for BRI. Cmglee felt it can be improved. Zefr (talk) 22:21, 7 October 2024 (UTC)

Silhouette caption options for consistency with the article presentation of roundness as <3 lean roundness, 3-6.6 healthy roundness, >7 excessive roundness (obese).

Caption 1: Silhouettes of roundness from left to right: healthy, overweight, obese

Caption 2: Silhouettes of roundness from left to right: lean, healthy, overweight

Caption 3: Silhouettes of roundness from left to right: lean - BRI <3, healthy - BRI 3-6.6, obese - BRI >7. Zefr (talk) 17:07, 8 October 2024 (UTC)

For now, the silhouette is not in the article yet. And... it might never be in this form. So I think the caption is not something to worry about now.

My understanding is that the image will serve as just the source for tailored images based on a BRI computed for a specific individual. A generated image for lean will probably be leaner than the leftmost healthy silhouette.

BRI categories still puzzle me. See #BRI_categories below. Uwappa (talk) 17:38, 8 October 2024 (UTC)

I filled in the regions to be more like the BMI graph. Happy to have coloured lines too.

P.S. I originally wanted to have cartoons of human figures of different BRI to accompany the lines' labels. Thinking more about it, I think it just messes up the graph. We can have instead a separate drawing with superimposed ellipses to show how BRI is calculated for a few representative BRIs. cmɢʟee⎆τaʟκ 22:49, 7 October 2024 (UTC)

Yes, that is definitely a separate graph.

Some thoughts, probably impossible due to javascript security reasons, but still:

• Make it one graph, preferably one that reacts to the calculator input, so it gives a personalised graphic.
• Stretch the graph vertically when the reader increases height in the calculator.
• React to a change in waist too. Always show the green silhouette for the green range. For a BRI inside the healthy range, there will be just the green silhouette.
• Show any excess as a red silhouette, just outside the green silhouette. The bigger the waist, the more red around green.
• Similar story for too lean: red inside the green silhouette.

It will be playful, yet conveying a personal message that could save a life. Uwappa (talk) 22:51, 7 October 2024 (UTC)

I don't think a calculator (with input fields) is viable on Wikipedia. You could create a Toolforge app like this one but it's beyond my skillset. Try asking at Wikipedia:Reference_desk/Computing. cmɢʟee⎆τaʟκ 22:58, 7 October 2024 (UTC)

The calculator is there already. See Body_roundness_index#Calculation Uwappa (talk) 23:02, 7 October 2024 (UTC)

By "calculator", do you mean the formula? I envision a calculator as a web page with text-entry boxes where one can type in (or at least select) values and some results appear, such as http://htmlpreview.github.io/?http://github.com/cmglee/web/blob/main/scale_solar_system.htm cmɢʟee⎆τaʟκ 23:18, 7 October 2024 (UTC)

No, I mean a working calculator that is in the article right now.

You may have to check "Enable on-screen javascript calculator" in your preferences, almost at the bottom of tab Gadgets in your user preferences.

See Template_talk:Body_roundness_index#test_of_calculator_with_one_height,_many_waists,_each_with_BRI for a version with just height as input, outputting many waists and BRIs. Uwappa (talk) 23:28, 7 October 2024 (UTC)

Wow, that's something I've wanted WP to do for ages. Thanks for letting me know. It only appears for people who have enabled the setting, though. Do many (know how to) do that? cmɢʟee⎆τaʟκ 23:43, 7 October 2024 (UTC)

Unfortunately only a few people know about it. It was a pleasant surprise for myself a few weeks ago. I stumbled upon it when it was requested to have that setting checked by default.

The calculator triggered the idea of having a 'zero input' calculator, showing many heights and waists. That is where you jumped in and created . I assumed you knew about the calculator, when you started working on the chart.

My hope: the calculator outputs its result to a variable (or even many variables). If you can pickup such a variable value and use it in an SVG, a personalised silhouette becomes possible. Uwappa (talk) 00:11, 8 October 2024 (UTC)

I saw the message about it on the village pump and thought a graph would work.

If you find out how to pipe the calculator output to an image or even a template parameter, do let me know. cmɢʟee⎆τaʟκ 09:13, 8 October 2024 (UTC)

The calculator already has a radio button metric/imperial that shows and hides input fields, based on input values 0 or 1.

That gives me hope it is also possible to show to show only 1 of 21 silhouettes based on an output value, a computed, rounded, min/maxed BRI. Uwappa (talk) 09:03, 8 October 2024 (UTC)

The Shard

Transamerica Pyramid

Great Pyramid of Giza

If you can dynamically set the lang parameter in an image thumbnail, you can use my frowned-upon hack above. cmɢʟee⎆τaʟκ 09:33, 8 October 2024 (UTC)

Sorry, that is beyond my level of calculator knowledge. user:Bawolff and user:Doc James have worked on the BRI calculator and can probably answer your questions. Uwappa (talk) 09:52, 8 October 2024 (UTC)

The output appears as an input tag: 100

px

If it can be plain text, there's a chance it could affect the thumbnail. cmɢʟee⎆τaʟκ 10:19, 8 October 2024 (UTC)

To clarify - the calculator cannot adjust wikitext. At most you can do some stuff with css and checkboxes, which allows you to make some images visible or invisible, do some adjustments to images (stretch, change colours, etc) Bawolff (talk) 22:50, 8 October 2024 (UTC)

 Done cmɢʟee⎆τaʟκ 19:59, 9 October 2024 (UTC)

I am not sure I understand you correctly. Sorry, I am not an expert on the calculator and my javascript knowledge is limited. Are you trying to assemble a string value in stead of an integer, something like '|30px'? Would something work like formula='|'+(num*2)+'px'?

Yes, and it didn't work. cmɢʟee⎆τaʟκ 12:00, 8 October 2024 (UTC)

OK, too bad. How about a less beautiful alternative, technically similar to hiding/showing cm and inches input fields, reacting to a {0, 1} value from radio buttons:

• prepare 20 silhouette images, from lean to wide
• show only one based on a hidden, computed, rounded, min/maxed BRI integer value {1, 2, 3, ..., 18, 19, 20}?

Uwappa (talk) 12:13, 8 October 2024 (UTC)

Using the pyramid-image hack, one needs only one image. The lang parameter determines which variant is shown. cmɢʟee⎆τaʟκ 13:05, 8 October 2024 (UTC)

Hmm, yes. So what you gain is one single file name for the many versions of the image. But you loose on coding comprehensibility.

Alternative to keep administration of versions doable and coding simple for the rest of us:

• Use BRI numbers in the filenames, e.g. silhouette_for_bri_01.svg, silhouette_for_bri_02.svg, ...
• Use "other versions=" in the wikimedia file infobox, to see all versions when looking at any one version. See 'Other versions' for File:Caricature_climate_change_Gerhard_Mester_english.png#fileinfotpl_ver an example.

And yes, it will be a tedious job to set up 19 other versions for 20 files. So be it. I'll be happy to assist you with that. Uwappa (talk) 13:57, 8 October 2024 (UTC)

The main issue is keeping all 20 files in sync: when one needs to be modified, all must be modified accordingly. Keeping it in one file lets common features be changed all at once.

Another solution is to put all images side-by-side in one file such as File:Earth_orthographic_projection_Earthmap1000x500.jpg and use template:CSS_image_crop to extract each image.

Will that work? cmɢʟee⎆τaʟκ 01:09, 9 October 2024 (UTC)

Cropping sounds good. It is simple to understand for the rest of us, keeps the language parameter available for local texts and keeps all versions in one easy file. It could be just one row of images, where you only need to crop horizontally to get the right part, which is then hopefully easy to link to the calculator.

So... for a simplified version with only 3 images, you could almost use the existing if the silhouettes would be equally spaced? Uwappa (talk) 06:43, 9 October 2024 (UTC)

The tailored silhouette could use the same colours as the chart to colour the 'shortage' or the 'excess'.

• A healthy silhouette could have a green line, as thick as a BRI 3-5.
• When a person increases waist size in the calculator, the 'fat layer' would grow and change colour. An obese silhouette would have a red layer of fat around a healthy green silhouette.

Overweight readers could play with the waist size and discover: how much waist do I need to loose to be just green, no excess, healthy? The colours for the chart are still subject of discussion, so please wait till those colours are stable. Uwappa (talk) 07:57, 9 October 2024 (UTC)

A "fat layer" doesn't work as the silhouettes are simple shapes which are stretched, so there will be e.g. a "fat layer" around the person's hair and the arms will be in different places.

You can update the color parameter on lines 20 to 39 of http://upload.wikimedia.org/wikipedia/commons/d/de/Body_roundness_index_silhouettes.svg as you see fit.

Cheers, cmɢʟee⎆τaʟκ 21:59, 9 October 2024 (UTC)

Yes, agree. It seemed like a nice idea at the time, but I'll strike it.

I am not sure about the ellipses. They are something for mathematicians that wish to understand the formula.

The silhouettes do a pretty good job on their own. I would remove the ellipses, just go for the silhouettes.

An alternative that may or may not work to show the delta with healthy in another way:

colour fill the entire silhouettes and paste a 'healthy' semi transparent green ellipse on top of each silhouette.

• Wide dark red silhouettes will be wider than the healthy green ellipse.
• Healthy green silhouettes will just fit the green healthy ellipse.
• Lean yellow silhouettes will have spare space inside the green healthy ellipse.
• When the use plays with waist in the calculator, the green healthy ellipse will seem constant while the silhouette grows/shrinks and changes colour.

The scenario I am hoping for:

1. The reader sees the colourful chart at the top of the page, but does not yet pay much attention to it the first time. The reader scans the page, sees the calculator and starts playing with some random height and waist values, sees the silhouette changing. Hey, what is going on here? (note: I do see the calculator when logged out, so the vast majority of readers may actually see the calculator) The changing silhouette colour may ring a bell, are these colours similar to the colourful chart at the top of the page?
2. The reader enters own real height. That is more or less a constant value for adults.
3. The reader enters own real waist and sees how much the silhouette deviates from the green healthy ellipse.
4. The reader plays with other waist sizes so the silhouette fits the green healthy ellipse. And bingo dear reader: you have just discovered the healthy waist size that you should aim for!
5. The reader starts reading the text on the page and recognises the colours and BRI values in the chart. And... may take action in real life towards a healthy waist size.

Wikipedia may have saved a life! And the reader may possible recommend the page to friends and family.

Uwappa (talk) 23:25, 9 October 2024 (UTC)

Another odd idea that may or may not work: The current BRI numbers on the chest of the silhouettes are redundant as the calculator shows a computed BRI with two decimals.

Could you remove the BRI numbers from the silhouettes chests and replace them with the BRI computed by the calculator?

Make the SVG the background image of the BRI output cell of the calculator? Tweak the computed BRI so it appears at the chest position?

That would give the impression that the silhoutte even responds to minor changes, e.g. when the BMI increases from 4.24 to 4.36. Uwappa (talk) 00:37, 10 October 2024 (UTC)

I've made the background transparent and removed the numbers. You can use Template:Annotated_image to crop the image, and position and overlay text on it. cmɢʟee⎆τaʟκ 22:37, 10 October 2024 (UTC)

Thank you.

Oops, annotated image seems to have a minimum width. width=50 gives same result as 100, where 150 does produce a wider image.

BRI 1	BRI 1	BRI 1
1 width 50	1 width 100	1 width 150

So annotated silhouettes would work only with large silhouettes?

1

width 50

Uwappa (talk) 08:00, 11 October 2024 (UTC)

Oh, I haven't encountered that issue as I never needed such a small crop. Could you ping its creators/maintainers? cmɢʟee⎆τaʟκ 15:02, 11 October 2024 (UTC)

Thanks. Posted at: Template_talk:Annotated_image#Minimum_width_for_crop? Uwappa (talk) 16:22, 11 October 2024 (UTC)

Update

Generic profiles for body roundness index values from 1 to 20

I've uploaded File:body_roundness_index_silhouettes.svg with ellipses coloured as in Uwappa's earlier table. I added black dots as some colours were hard to see. The left silhouette in File:Obesity-waist_circumference.svg was rescaled for BRI 1 to 5, the middle for 6 to 10 and the right for 11 to 20.

Here are some examples of use with Template:CSS image crop, the oLeft parameter being (BRI − 1) × cWidth, where BRI is an integer in [1, 20], cWidth is the intended image width, and bSize is cWidth × 20:

BRI 1	BRI 5	BRI 10	BRI 16	BRI 20

Can someone please check if I got the proportions right? I calculated the values with this spreadsheet. Feel free to update the SVG file as needed.

Cheers, cmɢʟee⎆τaʟκ 21:51, 9 October 2024 (UTC)

Your math skills outclass mine, so sorry, but someone else will have to check your calculations.

Could the silhouette background colour be transparent instead of white so the silhouettes could blend in with any cell background colour? Uwappa (talk) 11:19, 10 October 2024 (UTC)

I've made the background transparent. cmɢʟee⎆τaʟκ 22:38, 10 October 2024 (UTC)

BRI categories

Split off of reply by Zefr, 21:25 1 Oct 2024 above:

Uwappa - Given in Table 3 of the original Thomas ref, the range found in the database studied was 1-18, with "normal" as 3 to about 6 (median around 5.4). We state this in the lede. Under 3 and over 7 increased the risk of all-cause mortality; shown in blue).

Thank you. Is the following interpretation of Table 3 correct:

Classification	Color	Average BRI
lean		< 2
healthy		3 - 5
overweight		> 6.5
obese		> 9

Do ranges really overlap that much? So a BRI of 4 could be underweight, normal, overweight or obese? That can't be true, can it?

Uwappa - I modified the above table using averages from Table 3 of the Thomas paper. To make the table visually understandable for the casual user, I feel the averages and colors for each classification should be as simple as possible. Addressing extra-lean roundness conditions such as AIDS or anorexia seems outside the scope of a general article on BRI. If the table had a circumference silhouette beside it, this would seem to fulfill the goal of a straightforward visual presentation. Zefr (talk) 21:01, 8 October 2024 (UTC)

• What is the classification of a BRI between 5 and 6.5?
• Overweight is 6.5 to 9?
• When the silhouette will be tailored to calculator results, it could show the classification of the computed BRI. That would be only one, tailored classification at a time, not an explanation of all classifications.
• To explain all classifications, we could have a legend for the colours in (a non overlapping simplified version):

  -2 Lean

  3-5 healthy (we must retweak the graph colours to have green for healthy)

  6.5-9 overweight

  9- obese

See also idea above, in Silhouettes development.

• I agree that simple BRI classification would be nice. But... Does such a simple classification exist in a reliable source? The ranges in Table 3 just do overlap. And they overlap a lot. How about:

BRI classifications

1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18
underweight
normal
	overweight
		obese
					extremely obese

Uwappa (talk) 12:50, 9 October 2024 (UTC)

Although these overlapping ranges are accurate from Thomas, I believe they would only cause confusion for casual users seeking a quick review of BRI. Let's try a simplified version of the table focused on the Thomas averages. Note: highly trained professional and Olympic athletes in sports such as gymnastics, distance running, tennis, F1 racing, horse racing et al. are commonly small in stature and especially lean, not underweight. The term healthy for BRI is from the Thomas classification. My colors for the table below were guesstimates, but you'll see the idea. Zefr (talk) 17:30, 9 October 2024 (UTC)

Alternate table based on Thomas averages

BRI classifications

1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18
lean
		healthy
						overweight
								obese
											extremely obese

Good to see the design of the BRI classifications table is well understood. Finetuned colours in your alternate table, hope that is OK. Agree, the highly trained athletes are an exception, out of scope.

I am in doubt about the classifications lean, healthy, overweight, obese and extremely obese.

• Yes, I agree that your classifications are superior, little overlap, easy to understand.
• But... the nasty Wikipedian question: where to WP:Verify these Thomas classifications? Where in the document are these terms linked to those BRI ranges? The text above Table 3 is the source for my terms underweight, normal, ..., extremely obese.

Uwappa (talk) 18:33, 9 October 2024 (UTC)

Thanks for the table color adjustments. Thomas' Table 3 gives a mean BRI, BMI range and % fat mass, which together are represented in the Table 3 header as "underweight, normal weight, overweight, obese, and extremely obese in the NHANES database" - these are common terms in describing body shapes (aka "body types"). Perhaps, if a table like this were used in the article, we could have an explanatory caption with general sources to clarify the body types.

Other language in the Thomas article includes "lean" as a description of low BRI in Fig. 2.

We haven't discussed enough whether a modification of the 3 ellipses of Thomas' Fig. 2 could be used as a visual summary. Zefr (talk) 19:03, 9 October 2024 (UTC)

Please have a look at the text below the chart of Body_mass_index. It says those BMI categories are defined by WHO. I'd say that is a world standard we should stick to. Uwappa (talk) 18:50, 9 October 2024 (UTC)

Saw that, and the 6 categories are replicated in the range of BRI values in the graph. Thomas seems to display a simpler visual of using just 3 categories of "lean-healthy-too round", which I felt was useful for the casual encyclopedia user. Zefr (talk) 19:12, 9 October 2024 (UTC)

Verified 'lean' in figure 2. Hmm yes, it is there, but not linked to a range of BRI values.

Yes I agree, lean-healthy-too round are easy terms for a casual reader. Still, I think we should use the same terminology as the BMI article, the WHO and the text above table 3. And: underweight, normal and overweight are also quite easy to understand, just other words.

Suggestions:

1. Avoid confusion, use WHO terminology as the golden standard and stay compatible with the terminology in the BMI article.
2. Go for your range definitions and refer to the numbers in table 3 as a verifiable source. Extend the classification texts in your table with the ranges from table 3. Remove the vertical lines around empty cells.
3. Link to table 3 in the classification table header as a reference so the classifications are verifiable.
4. Do not copy the ellipse from figure 2. The ellipse is nice to explain the math to scientists, but we will have something better: the growing/shrinking in the calculator. Thomas had different audience and could not do an animation on a printed paper. Link to figure 2 as a reference when the calculator is ready.

Source used for table below

BRI classifications

1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18
underweight average: 1.86
		normal average: 3.17
						overweight average: 4.81
								obese average: 6.67
											extremely obese average: 9.93

Could you fill in the ranges? Uwappa (talk) 20:13, 9 October 2024 (UTC)

Uwappa - Creating clear ranges for this table from the Thomas Table 3 is unsatisfying because the ranges in the NHANES data and the standard deviations per category are large (there were 6,439 people included in Thomas' use of the NHANES database). I entered female-male averages as a default, but this is not useful because the average number is mismatched with the column headings of BRI.

In the "overweight" category where the Thomas average is 4.81, the Zhang analysis included nearly 33,000 people (the larger number of subjects presumably reducing variation), where the average "normal" BRI was about 5. Not sure how to best deal with this discrepancy.

Column widths at the top of the table became irregular in this new table; a result of colspan settings?

Thinking ahead to modifying the article, is a color-coded table with the BRI ranges needed in light of Cmglee's template of silhouettes? The silhouettes give an immediate visual definition, which may be sufficient with text added for clarification. Zefr (talk) 02:41, 10 October 2024 (UTC)

• The column widths became irregular because even a small, 3 column, range had to accommodate a large text like (female-mage average = 1.86). Band-aid solution: made all column heading equal width using silhouette figures. Shortened '(female-mage average = 1.86)' to 'average: 1.86' so it fits inside the width of 3 column headings.
• The average for overweight is 4.81 within a range 7 to 9? That can't be right. Similar question for obese and extremely obese.
• My first reaction to the categories: One silhouette for one BRI at the time in the calculator won't be enough. I think we need a chapter like Body_mass_index#Categories to link BRI numbers to categories. What are you suggesting? Show all 20 silhouettes?
• Something like table 3 will only be clear to people that understand what standard deviations are. If category boundaries are fuzzy and overlapping, we really need an easy to understand graph. If the table above is not suitable for that, well so be it, I am happy to trash it and design something better.
• First thought for an alternative: something like but with bell curved lines as in , something similar to . That could show overlapping categories, with a top of a bell curve at 1.86 for underweight, at 3.17 for normal, 4.81 for overweight, etc.
• Second thought: have 10 of those bell curves, 5 for female, 5 for male. The top of each bell curve will determine its line colour, taken from . The left most underweight line will be yellow, the rightmost will be dark red for extremely obese. Don't fill those curves as it will be an overlapping mess (or won't it?), just draw coloured lines, like in .
• Third thought: have 2 charts, one for females with just 5 lines and another chart for males. That will reduce the overlapping chaos and will provide a chart tailored for most readers. I have not seen any numbers for Intersex people.
• Fourth thought: I have some doubts about table 3. Does that table really map BMI categories and BMI range values against BRI numbers? Duh... Am I trying to stick apple labels on oranges? Should BRI have its own categories, based on its own ranges? And are those BRI categories something that have yet to be invented? So, is table 3 the wrong source for BRI categories? Is that why you suggested to just go for the silhouettes?
• Fifth thought: I have based colours on this figure 2, Association Between Body Roundness Index and All-Cause Mortality Risk. How about mapping BRI numbers to silhouettes and mortality risk? That should be doable and sidesteps the whole question of categories.

Uwappa (talk) 17:49, 10 October 2024 (UTC)

Waist to height classifications = BRI classifications?

BRI ranges can use WhtR classifcations, without any value overlap.

WHtR Classification	height in cm	waist in cm	WHtr	BRI
healthy central adiposity	180	72	0.4	1.67
	180	89.9	0.499	3.35
increased central adiposity	180	90	0.5	3.36
	180	107.9	0.599	5.42
high central adiposity	180	108	0.6 or more	5.43 or more

Looking at the formula, there a waist-to-height ratio can be converted to a BRI and vice versa, as the only variables it uses are the same as in Waist-to-height ratio.

$${\text{WHtR}}={\frac {\text{waist circumference in cm}}{\;\;\;\;\;\,{\text{height in cm}}}}$$

$${\text{BRI}}=364.2-365.5{\sqrt {1-{\Big (}{\frac {\text{waist circumference in cm}}{\;\;\;\;\pi \;\;\;\;\;\times \;\;\;\;\;\,{\text{height in cm}}}}{\Big )}^{2}}}$$

constants

364.2, 365.5, 1, pi, 2

variables

waist circumference in cm, height in cm

So, the good news: Waist-to-height ratio and BRI are basically two different scales for one and the same concept. BRI is not as new as it seems. It is old wine in a new bottle.

Any research, any publications about waist-to-height ratio are also valid for BRI. Just convert the waist-to-height values to BRI. Uwappa (talk) 08:01, 14 October 2024 (UTC)

Uwappa - thanks, but we would need a WP:BMI source that explains this in a rigorous review, rather than doing our own waist-to-height/BRI scale (WP:OR). We also need to observe WP:WFTWA, i.e., keep the article definition simple and BRI-specific for the non-scientific user. Zefr (talk) 18:12, 14 October 2024 (UTC)

No, you do not need a source for simple calculations, WP:CALC. Most medics are smart enough to check computations themselves. To check calculations you need formulas, values and a calculator. I've added the WHtR formula, so you can see the obvious: WHtR and BRI have waist and height as the only variables.

No, medics do not need a WP:BMI source to see: WHtR and BRI formulas both use waist/height as the only variable part. It is really simple, obvious:

• Meters and yards are both units for length.
• WHtR and BRI are both units for body roundness.

And yes I fully support WP:WFTWA. Experts don't need Wikipedia for their own field of expertise. Wikipedia is for the general public.

• WP:NOR Converting jargon to plain English is required by policy. Medical jargon like "increased central adiposity" is OK on a talk page, but should be translated to plain English in an article. See Mortality risk, 11:01 14 Oct below for suggested plain English terms.
• The BRI formula is a bit repellent, but OK as it can be safely ignored with hiding all computations and the calculator will make computations just too easy.

I hope that you realise that BRI is not really a new kid on the block. It is actually just a different scale for body roundness. That is good news. You do not need to wait for more BRI research, you can reuse old WHtR research, just convert WHtR values to BRI, as shown in the table above.

We can have a BRI graph (and a WHtR one) with crisp, clear boundaries. Just step away from troublesome BMI->BRI mapping as converting weight to roundness is troublesome. Embrace a conversion from one body roundness scale to another. To me this is so obvious. Resistance is futile.

Bold suggestion: Combine articles on WHtR and BRI to one, called "Body Roundness". See Length for an example, explaining both km and miles.

Show WHtR and BRI in a chart:

Uwappa (talk) 11:29, 15 October 2024 (UTC)

That chart could show colours for classifications, horizontal from Y-axes to line for BRI and vertical from line to X-axes for WHtR. Uwappa (talk) 06:55, 19 October 2024 (UTC)

BRI of famous people

An odd idea that might just work to explain BRI to a large audience:

Show a table with the height, waist and BRI of famous people, ranging from via to .

This might make the reader curious. How about my own BRI? It might trigger the use of the BRI calculator to compute own BRI and take action towards a healthy waist size.

Idea inspired by:

• A remark by about athletes by user:Zefr
• WP:MTAU (Make technical articles understandable), chapter Add a concrete example
• the weekly WP top 25. Week after week, readers flock to Wikipedia mainly for information about famous people: actors, athletes, musicians, politicians.
• the silhouettes
• Show, don't tell

Uwappa (talk) 14:06, 15 October 2024 (UTC)

Proposed graphic for article

Uwappa - thanks, your combination of the silhouettes with the BRI ranges impresses as a good solution, although I don't feel all the silhouettes are needed. Both in Thomas and in Zhang, the highest BRI was 16, and the point of extreme obesity is made with BRI 13-16.

The BMI and WHO terms for body shape are 50+ years in use, and are being criticized, while BRI is a relatively new concept of 10 years and under active research. Contrary to your good advice, I feel we should help break new ground and follow the Thomas terms of "lean" and "healthy" BRI, keeping in mind many athletes are in the lower BRI range and are certainly not "underweight" or "normal" compared to the general population. Medical editors who visit the article can challenge/change this terminology, as needed.

A new article subsection called "Range of BRI values" can contain the modified graphic and a description with the Thomas Table 3 as the source. Feel free to modify/discuss.

Cmglee: can the ellipse colors for each silhouette be made brighter? Zefr (talk) 18:26, 10 October 2024 (UTC)

Feel free to edit the colours on lines 19 to 38 of http://upload.wikimedia.org/wikipedia/commons/d/de/Body_roundness_index_silhouettes.svg cmɢʟee⎆τaʟκ 22:39, 10 October 2024 (UTC)

Cmglee - thanks, but this file link produces only images. Zefr (talk) 22:49, 10 October 2024 (UTC)

Exactly: it links to the actual SVG file. Save it on your computer, open it with a text editor, edit the rrggbb values on the color="#rrggbb" lines, save it, and then upload the new version. cmɢʟee⎆τaʟκ 23:01, 10 October 2024 (UTC)

Alternative: colour the background, which is now possible as cmglee changed the background colour of the silhouettes to transparent. That would be inline with use of backgrounds in . See BRI 1 below for an example of CSS.

The ellipse could serve a new function: Show the size of healthy, fixed size, green in colour. Too skinny will have excess space inside the ellipse, to round will be wider than the ellipse. Each silhouette could show an ellipse for a healthy waist size, similar to

• For a too lean silhouette that ellipse will be too big, indicating waist shortage.
• For a healthy silhouette, that ellipse will fit perfect around the waist.
• For a too wide silhouette that waist will be too small, indicating excess fat.

In the calculator this could work like a mini game. Play with waist size till 'it fits the belt' and the reader has found a personal goal for a healthy waist size.

Uwappa (talk) 13:27, 11 October 2024 (UTC)

1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
lean 1.86
		healthy 3.17
						overweight 4.81
								obese 6.67
												extremely obese 9.93

Possible subsection text: The graphic displays body roundness shapes in relation to BRI values across a range from lean to extremely obese. The numerical value of BRI shown for each classification was determined using the average BRI between both female and male American adults.[Thomas ref] According to research, some overlap of BRI occurs with the adjacent classification of body roundness.[Thomas ref][Zhang ref]

1. Agree: We should have a new section that explains BRI values. Without it the only explanation will be the silhouettes.
2. Agree: The high end long tail of BRI values is quite small and irrelevant. But still... this figure 2 goes all the way up to 20. I think it is OK to keep the 20 silhouettes. No harm done.
3. The extremely obese bar now starts at 13 where it started at 12 in the previous version. Would you like me to fix that?
4. You have lost me with the number crunching. Overweight 4.81 in a range from 7 to 9? Similar question for obese and extremely obese. So... I don't know what I am looking at.
5. Sorry, bad news: Introducing new terms remains a no no for me on Wikipedia. Sorry, but I fail to see the link between BRI ranges and the terms in the Thomas doc. My doubt about the WHO terms is that they are related to BMI. So... go for it on your own if that is what you really want. Simple alternative: BRI quintiles Q1, Q2, Q3, Q4 and Q5 as in BRI and All-Cause Mortality? I don't really like it though. Q1 is confusing with the first quarter of a year. Also it is confusing to have numbers to classify numbers. Uwappa (talk) 20:04, 10 October 2024 (UTC)

About 4, the number crunching In Thomas Table 3, the 5 categories and averages of female-male BRIs read up from BMI 18.5 (combined with increasing % fat mass) as lean to normal ("healthy") to etc. ending with BMI 40 (extreme obesity). 4.81 is the f-m average for the 3rd category, "overweight", at 25 ≤BMI <30. I'm open to a better definition from the Thomas paper, if there is one. Also, the Zhang mortality study defined the middle quantile of BRI as 4.5 - 5.5, and showed in Fig. 1 the well-established rise of body weight, BMI, fat mass, and BRI from around 5 in American adults over the period 1999-2018. Zefr (talk) 22:31, 10 October 2024 (UTC)

The 4.81 overweight average is clear, male female average: (4.66+4.96)/2 = 4.81.

Where I get lost:

How do you get to an overweight range of [7 9] from male [2.03, 7.33] and female [2.34 11.53]? Why not [2.03 11.53]? How does 4.81 fit into [7 9]? Where do you see overweight has a [7 9] BRI range?

Uwappa (talk) 10:17, 11 October 2024 (UTC)

About 5, terminology I'll try to raise an assessment about this by other medical editors. Zefr (talk) 22:31, 10 October 2024 (UTC)

I get the feeling the Thomas Table 3 is torn between two thoughts: old BMI classifications versus new BRI ranges, which results into a mess of overlapping BRI ranges as shown earlier. One way to deal with that mess is to just accept it for the time being.

The Thomas document does not define new BRI classifications, with clear BRI ranges. That is a pity, but reality. It is the current state of science for this new kid on the block called BRI.

Alternative WP:BOLD, yes, just boldly go for it and see if other editors object. I think it is important that the BRI numbers are properly categorised, more important than my own reluctance. Suggestion: Define a whole new set of terms, avoid all BMI terms. Where BMI terms as underweight and overweight are related to mass, weight, BRI terms should be related to roundness. How about:

• emaciated
• lean
• healthy

(this one stands out as it does not indicate roundness.)

• curvy
• rotund
• spherical

Uwappa (talk) 10:17, 11 October 2024 (UTC)

Subsection text

Sorry I did not reply earlier to your text proposal, I was focussing on the graphic.

The current text gives me a headache, way too academic. It overloads my short term memory with redundant fluff. Please trim it down for the rest of us. See 'suggested' column below.

current text	comment	suggested alternative
The graphic displays body roundness shapes in relation to BRI values across a range from lean (BRI less than 3) to extremely obese (rotund; BRI more than 12).	The graphic is pretty strong on its own. No need to announce the obvious. A lot of the text is redundant. Recommended exercise: Remove fluff from your writing Yes graphics display things, that is what graphics do. Yes I do see roundness shapes and BRI values. Yes I do see lean is for BRI 1 to 3 I do not see 'Extremely obese' in the graph. The limits of 3 and 12 are unclear. Does 3 mean 3.0 or 3.99? What would be the category for 3.5? And how about 12.5? Is 16 really the max? There is a long tail up to 20 in figure 2. Yes I know, numbers above 16 are exceptional. Alternative: Change 16 in the column header to '16, 17, ...'?	BRI values range from lean to rotund. Or even bolder: no text at all, the graphic speaks for itself. The text may serve the blind, as their vocal screen reader will not 'show' the graphic. The text serves many purposes, such as retrieving specific information from Google and AI chatbot searches, and is part of the description for readers of the article who may not be familiar with science. In WP:MEDMOS, we have WP:WFTWA, which is intended as a guide for keeping content at a level high school students could follow. I try to follow WP:BETTER, but improvements can be made by anyone, so please edit accordingly. "Extremely obese" and "rotund" are equated in the text. Zefr (talk) 18:52, 12 October 2024 (UTC)
According to development of the BRI and subsequent research, some overlap occurs with the adjacent BRI classification of body roundness.	Keep references out of the main text. Yes, development and research do yield conclusions. That is what they do. Any conclusions that are not based on research should not be in Wikipedia. Now, just give me the outcomes and I'll look at the refs for the underlying research only if I am really interested. Yes, I do see that the overlapping BRI classifications are adjacent. And that non adjacent classifications do not overlap. It would be messy if they did. "BRI classifications of body roundness": Really? So that is Body Roundness Index classifications of body roundness?	Most BRI classifications overlap. Or again: no text at all, the graphic speaks for itself, except for the blind. Move simplified text to a table caption? A rule of thumb in MEDMOS and WP:MEDRS is that every statement should have a source. On a relatively new and narrow topic as BRI currently is, we have only 11 references and two reviews (Thomas, Rico-Martin). I don't feel the article is burdened by citations. Zefr (talk) 18:52, 12 October 2024 (UTC)

References

Bold proposal: Move the graphic to the start of chapter Calculation. Rename that chapter to 'BRI values'.

Second thought: You may want to take the text to a higher level. The graph is very strong and does not need much explanation for a person with good vision. The 2 dimensional graph is stronger than a 1 dimensional string of words can be. So the target audience of the text may be mainly blind people.

I once visited an exhibition completely in the dark, organised to make people experience how life is for blind people. It was a fascinating experience.

• Colours do not exist, but shapes do as long are they within reach. Anything out of reach is 'beyond the horizon, out of sight', does not exist. Coins were easy to tell apart, but banknotes all felt the same. It was hard to buy things that were wrapped in 'transparent' plastic. The whole concept of transparent lost its meaning.
• Health levels do exist, some people in the group moved easier, faster than others.
• Lean and fat persons do exist, but I could not tell the difference until you touch them.
• Colours are an unknown concept. Races did not exist, as I can't hear race. Accent was the only clue to distinguish an Italian from a Chinese. Genders were easy as I could distinguish a male voice from a female one.
• Numbers do exist as long as I could feel things to count them.
• Distances and directions exist. Sounds gave me an indication of direction and distance.
• Parallel input was hard. Having a conversation was difficult when many people were talking nearby.
• Any text that is out of touch, out of reach, out of hearing distance does not exist. Listening to a stream of words does not allow scanning. Fluff is very annoying, slows down. I can't skip ahead to the next sentence, the next paragraph, the next chapter, as I would be clueless where that would start.

So... giving this 'blind world' setting, what story would the graph tell to a blind person?

Bold suggestion for a new text, using terminology that makes sense in a 'blind world':

BRI values range from 1 meaning lean to 16 meaning rotund. BMI classifications overlap:

• 1 to 3 means lean, with 1 being a serious health risk
• 3 to 7 is the optimum, with a peak at 5
• 7 to 9 means too fat
• 9 to 12 means obese
• 13 to 16 means rotund, a serious health risk

And... please have a look at an idea to improve the graph even further, transforming it into a health arch.

Uwappa (talk) 12:22, 12 October 2024 (UTC)

Please just go ahead and edit to improve clarity. Zefr (talk) 18:52, 12 October 2024 (UTC)

It is unclear to me to which part of my suggestions you agree with and would like to see in the article.

I do agree that we should avoid WP:WFTWA. That is the main point here. To me this looks like something a student would write trying to impress a professor. It is a lot of fluff to wade through searching for a little bit of content.

Also, please keep in mind that for a large part of the audience English is a second language. I do not have any figures at hand, but it would not surprise me if the foreign language speakers are the majority. Reading another language is tricky. Chaque mot demande un effort mental pour être traduit. Plus il y a de mots, plus le lecteur a de travail. Les mots superflus font mal. (Every word requires a mental effort to translate. The more words, the more work for the reader. Fluff hurts.)

My question is: who would the audience be for the text?

To me, the graph is already very good, needs very little or even no explanation at all.

• People that can read charts do not need the text, so boldly trash it? Just move the references to the graph?
• People that have difficulty to read charts? Blind people? And yes, I would be happy to address those audiences, hence my proposed text.

Please could you implement the suggestions that you agree with? Uwappa (talk) 19:51, 12 October 2024 (UTC)

Age ajusted boundary values

Waist-to-height_ratio#Age-adjusted_boundary_values lists age specific limits with significant reduction in life expectancy if exceeded.

The BRI article could show the BRI equivalents:

Age	WHtR limit	BRI limit
under 40 years	0.5	3.36
40–50 years	0.5 to 0.6	3.36 to 5.43
over 50 years	0.6	5.43

We could have

• 3 sets of colours for health level,
• 3 versions of , same BRI lines, but different background colours
• age category as input for the BRI calculator, resulting in age specific silhouette background colours.

Uwappa (talk) 08:56, 15 October 2024 (UTC)

Other table considerations

And for a correct classification, you would need gender, race and age? That sounds like opening a can of worms, destroying the charming simplicity of BRI.

Could we use figure 2 for a simple classification:

Classification	Color	BRI
too lean		< 3
healthy		3 - 7
too round		> 7

Uwappa (talk) 10:25, 2 October 2024 (UTC)

Uwappa - good concepts, thanks. Have simplified the bottom color table for discussion. An alternate display could be as 3 ellipses of different shapes and colors, like the "green zone" of Fig. 3, with corresponding BRI values connected to the respective ellipse (< 3 vs. 3-7 vs. > 7). Zefr (talk) 19:11, 2 October 2024 (UTC)

Please ping me when you've decided on the appropriate ranges. I'll update it in the weekend. Cheers, cmɢʟee⎆τaʟκ 21:40, 2 October 2024 (UTC)

Just 3 colors is nice and simple. 2 colors would be even simpler, where too lean is dangerous, just as too round. In addition, cyan has no 'traffic light meaning' as green, yellow red do. So for simplicity and clarity I'd prefer just green and red, for good and bad:

Classification	Color	BRI
too lean		< 3
healthy		3 - 7
too round		> 7

Yet this is a bit too simple for me. 2.99 is too lean, and 3.01 is healthy? 6.99 is healthy and 7.01 is too round?

Two possible solutions:

1. Complex: The graph shows areas between lines in gradients of green, via yellow to red.
2. Easy: The colors of the lines are shades of green for healthy (3-7), shades of yellow, orange and red for risky to dangerous.

That would show there is benefit in slimming down from 10 to 9, from 9 to 8. More colours can encourage people to make a step in the right direction, no matter how small. Uwappa (talk) 10:29, 3 October 2024 (UTC)

The classifications are for mortality predictions, which are likely too multifactorial for BRI to apply clinically as a general guide. What does Doc James - a practicing physician - think about the value of such a 3-classification chart? Zefr (talk) 16:36, 3 October 2024 (UTC)

Mortality risk

What I do like to see: a graph that shows BRI versus mortality risk, the blue line and area in this figure 2. What does a computed BRI value mean for you, your health, your life? How risky is lean or obese? What BRI range is the most healthy? How much waist should you gain or lose to be healthy? The Template:Graph:Chart is still down, but I think a tweak of a stacked bar as an alternative. Will come back on that one. Uwappa (talk) 20:04, 10 October 2024 (UTC)

Starting this new discussion introduced by Uwappa. Zefr (talk) 22:30, 10 October 2024 (UTC)

How about no new graphic, but... enhance the current graphic. Use the y-axes for health, the inverse of mortality risk. That Y-axes is currently wasted, unused.

Put healthy on top of a 'health arch'. Classifications with a similar health risk will be at the same height (the same row in the table).

• The health arch will be like a vertically flipped blue mortality risk line from figure 2.
• That blue mortality line goes steep down from 1 to 3, so flipped vertically here it should go steep up. BRI 1 (emaciated, skinny, gaunt?) should be as dangerous as BRI 14, same colour, same very bad health level.
• The 'y-axes' could have labels related to health, from top to bottom terms like: healty, risky, dangerous, morbid.
• The BRI classifications will be all related to roundness, e.g. 'normal' will replace the current 'healthy' and that normal will be at the healthy health level.

	Body Roundness Index
	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16, ...
healthy			normal
risky		lean					curvy
dangerous	thin?								obese
morbid	gaunt?												rotund

And yes, I should have thought of this earlier, when I based the colours on the blue mortality risk line.

Uwappa (talk) 00:22, 12 October 2024 (UTC)

This presents a good visual summary of the Zhang study, but for the article on BRI, it represents only one cohort analysis of mortality risk, making it primary research and relatively low in WP:WEIGHT on the WP:MEDASSESS quality pyramid. I think the Zhang report is adequately used in the article among other examples of risks and disorders studied with the BRI. Zefr (talk) 18:03, 12 October 2024 (UTC)

Well, what is your alternative? How about sticking to the current graph, current terminology, current references and just

• add a general header explaining the 1-16 values.
• move 'lean' two rows down
• move rotund one row up
• remove the colours around silhouettes as the center values are clear.

Body Roundness Index
1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
		healthy
lean						overweight
								obese				rotund

That will put 'healthy' nicely at the top, marking it clearly as the best. It will make the graph smaller, easier to look at. Lean and overweight will be at the same level as counterparts, share colours, though horizontally flipped. Obese and rotund will be the long tail, adjacent, in line with each other without any overlap. The whole will still resemble a vertically flipped blue line from figure 2, but without explicitly telling so. Readers do get a hint of a hidden health y-axes, but it is not explicitly there.

The current classifications are like comparing apples with oranges and eh... a tomato?

1. 'healthy' is on some health scale.
2. lean and rotund are on a scale of roundness
3. overweight is on a scale of weight.
4. not sure where obese fits in, could be any or all of health, roundness, weight.

My suggestion: keep looking for terms that all are on the same scale, roundness, the R in BRI.

And please think about where unhealthy Emaciated/gaunt would fit here, as the counterpart of rotund. I think it should be at the same level, same colours as rotund, being equally dangerous.

Also please have a look at #Waist_to_height_classifications_=_BRI_classifications? for some alternative qualifications based on October 2022 NICE guidelines, without overlapping ranges.

Uwappa (talk) 19:20, 12 October 2024 (UTC)

Thanks for the work and concepts, but I will follow consensus among editors on this one, WP:CON. To me, providing a graphic like this overemphasizes a primary source, which would be WP:UNDUE. We should wait for a WP:MEDRS review of other reputable studies on mortality risk. Zefr (talk) 19:40, 12 October 2024 (UTC)

O, I had the impression we almost reached WP:CON here and that I did not do anything to overemphasize a primary source. But well, OK, so be it. I'll sit back and wait. Uwappa (talk) 19:57, 12 October 2024 (UTC)

I have had a long look around and I cannot find any WP:MEDRS review on mortality risk on this. A few textbooks are starting to mention BRI but it is still early days. They are currently conducting cohort studies but it will be a few years until we see a good review on BRI and chronic disease risk. The AMA is currently looking into this topic [1], [2]. The article is well written with good sourcing, the only sources I would dispute would be the Frontiers journal studies which is primary source material. Psychologist Guy (talk) 15:32, 13 October 2024 (UTC)

Statement: Waist-to-height_ratio categories also apply for BRI, as BRI is just a new scale for waist-to-height ratio. A ratio can be converted to an index and vice versa.

Similar to: sizes in meters can be converted to yards and vice versa.

Suggestion: Go and look for reliable sources on Waist-to-height ratio. Use waist-to-heigth categories for BRI, with ratio values converted to BRI values.

In the meantime, think of a meaningfull y-axes for the graph.

• At the moment the y-axes is based on table 3, BMI categories, see first design, 12:50 9 Oct, which just shows the correlation of BMI and BRI values, with a lot of overlap in ranges.
• Using a 'health level' for y-axes seems so logic to me. Put healthy on top, morbid at the bottom. What is the alternative, stay stuck with the y-axes of the first design?

Future publications won't change the design concepts of the health arch. The top of the health arch will remain at the top, may be with a different name and some fine-tuned range values. The bottom of the arch will remain the most dangerous.

How about using waist-to-height related categories on the y-axes? Terms like "healthy central adiposity" are incomprehensible for the general reader. How about a translation to understandable terms:

• WHtR below 0.4 -> ??? -> ??? -> risky to dangerous (BRI below 1.67)
• WHtR 0.4 to 0.49 -> healthy central adiposity -> no increased health risks -> healthy (BRI 1.67 to 3.35)
• WHtr 0.5 to 0.59 -> increased central adiposity -> increased health risks -> risky (BRI 3.36 to 5.42)
• WHtR 0.6 or more -> high central adiposity -> further increased health risks -> dangerous (BRI 5.43 or more)

These BRI ranges support your earlier statement: athletes not underweight. The current graph shows WHtR 0.6 (high central adiposity) as healthy, BRI 5.43 in the range 3 to 7.

Please check my WHtR to BRI mapping of ranges. Are my computations correct? Uwappa (talk) 11:01, 14 October 2024 (UTC)

Section on Range of body roundness

I added the graphic and description of BRI ranges with this edit.

Please select in your preferences: Enables javascript Calculator template to see a working calculator.

Body roundness calculator

Units	Metric		Imperial
Height	178 cm		5 ft 10 in
Waist	80 cm		31 in
WHtR	0.44
BRI	2.46

Somehow, the edit interfered with the calculator display, as it is no longer present. Zefr (talk) 17:26, 11 October 2024 (UTC)

The calculator is working fine for me. I see it right here and it works in the article too.

It needs a checkbox checked in your settings which used to be off by default.

1. Go to preferences, tab gadgets
2. Scroll down a lot to Template Gadgets
3. Check "Enables javascript Calculator template.".

and retry. Uwappa (talk) 17:36, 11 October 2024 (UTC)

Somehow, the edit interfered with the calculator display, as it is no longer present. Zefr (talk) 17:26, 11 October 2024 (UTC)

Calculator shows fine now.

In the silhouette displays, leaving one silhouette to the left of lean (BRI 2) and right of rotund (BRI 15) allowed the ellipse outline to be visible, which helps the viewer's eye on shape.

The ellipse colors do not show well from BRI 1-8 - I wasn't able to edit the svg file using Google Docs. Cmglee or Uwappa - is this something you can fix? Thanks. Zefr (talk) 17:36, 11 October 2024 (UTC)

Think old skool, low tech, use a very basic ASCII editor like Windows_Notepad. Work locally, see Cmglee's advice above.

I think the problem is the line width of the ellipse. For tiny silhouettes it becomes very thin for low BRI values, too thin to see. After zooming in using CTRL+ the ellipses do become visible.

Suggestion: Slow down. Let us agree first on what to do with the ellipse in #Proposed_graphic_for_article above. If we go for a 'healthy sized belt ellipse' it will be one ellipse, same size, same colour for all silhouettes.

Suggestion: Swap the sections 'Calculation' and 'Range of body roundness'.

1. Explain the values first so the reader understands what values 1-16 mean. This will hopefully trigger the question: What would my own personal BRI value be?
2. The calculator allows readers to compute a personal BRI value.

Uwappa (talk) 18:25, 11 October 2024 (UTC)

I'm not sure what you mean by editing it in Google Docs. It's meant to be edited as a local file.

If you can prepare a table of colours for each BRI value (don't merely write "transition from X to Y", I'll update the SVG. cmɢʟee⎆τaʟκ 13:39, 12 October 2024 (UTC)

P.S. The ellipses will be more visible if they're filled with a solid colour instead of an outline. Will that work? cmɢʟee⎆τaʟκ 13:40, 12 October 2024 (UTC)

Yes, that would work, but... would draw attention away from the main object, the silhouette.

At the moment the silhouette and ellipse are two visualisations of one and the same variable: BRI. The ellipse was OK in a static documents to explain the formula for academics, but the silhouette is superior to show roundness to the general public online, in an interactive calculator.

Alternatives, my most favorite first:

1. Replace the ellipse by a visualisation of a constant, the healthy-waist-belt ellipse, same (healthy) size for all, not impacted by stretching of the silhouette.
2. Trash the ellipse, just show silhouettes. One visualisation for BRI will do. No need for 2 visualisations of 1 variable.
3. Swap: Colour the silhouettes, fill all ellipses grey.
4. Give ellipses the same, fixed line width, not impacted by a horizontal stretch of the silhouette.

Uwappa (talk) 16:25, 12 October 2024 (UTC)

BRI raison d'être?

Why does BRI exist?

It is very similar to WHtR which also yields a value for body roundness, albeit with a different formula, a different value range.

The WHtR and BRI formulas can be merged. Combine

$${\text{BRI}}=364.2-365.5{\sqrt {1-{\Big (}{\frac {\text{waist}}{\;\;\pi \;\;\times \;\,{\text{height}}}}{\Big )}^{2}}}$$

with $${\text{WHtR}}={\frac {\text{waist}}{\text{height}}}$$

to get $${\text{BRI}}=364.2-365.5{\sqrt {1-{\Big (}{\frac {\text{WHtR}}{\;\;\pi \;\;\,}}{\Big )}^{2}}}$$

Given the last formula, it should not be surprising that a WHtR value can be converted to its BRI equivalent.

Please select in your preferences: Enables javascript Calculator template to see a working calculator.

WHtR to BRI converter

pi = π	3.141592653589793
${\text{BRI}}=364.2-365.5{\sqrt {1-{\Big (}{\frac {\text{waist}}{\;\;\pi \;\;\times \;\,{\text{height}}}}{\Big )}^{2}}}$	3.3588109427631707
${\text{WHtR}}={\frac {\text{waist}}{\text{height}}}$	0.5
Waist	80
Waist diameter = Waist / pi	25.464790894703256
Height	160

• Why would a Physician use BRI and not WHtR? What would be the advantage of BRI?
• WHtR is very easy to apply, is independent of a unit such as cm, m, inch, feet.
• The same does apply to BRI, if both height and waist are measured in the same unit. However, in the UK body heights are usually expressed as a combination of feet and inches.

Please select in your preferences: Enables javascript Calculator template to see a working calculator.

Body roundness calculator

Units	Metric		Imperial
Height	178 cm		5 ft 10 in
Waist	80 cm		31 in
WHtR	0.44
BRI	2.46

• The WHtR formula is easy to apply, where most people would require a calculator for the BRI formula.
• WHtR allows a simple check of the NICE guideline boundary level value of 0.5 for increased central adiposity, increased health risks . No required, as a simple piece of string would suffice: Mark a persons height with a string, try to wrap that length around the waist twice.

Too short

increased central adiposity, increased health risks.

just right

just within the health limit

too long

no increased central adiposity. A healthy central adiposity, no increased health risks for most people, but the WHtR could be below 0.4, the minimum value of a healthy central adiposity

Could the simplicity of WHtR be the reason why it is more popular than BRI, more research available, more reliable secondary sources?

And a tricky question for Wikipedians related to the BRI raison d'être?

Does the Body roundness index article have a raison d'être? Uwappa (talk) 04:27, 25 October 2024 (UTC)

It is not really a tricky question, since the term appears to have significant currency in the US and is therefore WP:notable. But perhaps the real question is whether it merits an independent article. It seems obvious to me that it merits no more than a section at the end of Waist-to-height ratio and a short one at that, no longer than the lead of the BRI article as it stands, with a note saying that it is a mathematical rephrasing of WHtR as an integer rather than a decimal fraction. BRI would continue to exist as a redirect to that section.

Do we need a formal WP:merge process first? --𝕁𝕄𝔽 (talk) 08:04, 25 October 2024 (UTC)

Please sit back and relax for now. Choosing a direction is easiest when at rest. We need to get a clear view first.

My suggestion:

1. Have a new article about the concept of body roundness. Just like articles on concepts like distance, mass, time. Explain things that WHtR and BRI have in common, such as being based on a proportion of height and waist, a quite reliable indication of body fat, difference with BMI, independent of unit, male/female differences. And... its 2 units being WHtR and BRI.
2. Keep the WHtR article as an article on a unit, just like meter, inch, mile, lightyear are for distance, Kilogram, Pound_ for mass and second, minute, hour, century for time. Add the simple WHtR formula.
3. Move paragraphs about the concept from WHtR to the new body roundness article.
4. Same story for BRI, the concept stuff goes to the new body roundness article. And... surprise, surprise, the BRI concepts will overlap with those moved from WHtR.
5. Create links from body roundness to WHtR, BRI, and vice versa. Just like distance links to kilometers, miles and vice versa.
6. Keep the article for BRI. It is an existing concept. WHtR will probably be the general favorite, fine, just like meter is more popular worldwide than yards. And ha ha ha, don't tell the Americans and the British here at the English Wikipedia.
7. Think about graphics that show how various units relate to each other, like and well..
8. Think about calculators for conversions of units, like the one at Inch#Equivalents and Centimetre#Equivalence_to_other_units_of_length and well... Template:WaistHeightRatio_to_BodyRoundnessIndex_converter
9. Think about where to go with Template:Body_roundness_index

Please select in your preferences: Enables javascript Calculator template to see a working calculator.

Body roundness calculator

Units	Metric		Imperial
Height	178 cm		5 ft 10 in
Waist	80 cm		31 in
WHtR	0.44
BRI	2.46

.

My suggestion:

• at the new body roundness page as shows the concept of roundness and it will show health risks again when the nasty bug has been fixed.
• at WHtR as it shows computed WHtR values, will show health risks, next to the NICE health risk levels
• at BRI as it shows the computed BRI values, near the formula, where it is already. With the BRI formula being the most complex, that is the place where the calculations of the calculator will be most helpful.

And... please get calculator 4.0 going again. One of the blocking points now is a new selection of colours. Pick the key colours, like you did last time. And the good news: you can now pick colours that match the NICE guidelines health risk ranges and have them linked to clear and crisp WHtR values.

That calculator may later be extended with a female silhouette. I would also love it to have the BRI shown on as an annotation. Not sure yet, but I think that will be possible without any javascript security issues. Yes, I really love Template:Calculator. Please have a look at the wikicode for a simple example at Inch#Equivalents. I think most Wikipedians should be able to create those type of conversion calculators.

Enough to go for, but for now... One step at the time is difficult enough. Could you propose WHtR based colours at Template_talk:Body_roundness_index#WHtR_based_colours,_silhouettes_and_text_for_Calculator_4.0. Similar to what you did last time? Uwappa (talk) 09:59, 25 October 2024 (UTC)

• Hold the front page! I've found a citation that suggests we may have an error somewhere, that hip circumference is also a factor, making the vertical elipses significant. For now, I'll take it to Uwappa's talk page lest it is a misunderstanding. --𝕁𝕄𝔽 (talk) 10:05, 25 October 2024 (UTC)

  Keep the discussion here please, not on my talk page.

  I am curious to read the latest breaking news! Uwappa (talk) 10:49, 25 October 2024 (UTC)

  Breaking news moved to #The_vertical_elipses. Uwappa (talk) 12:20, 25 October 2024 (UTC)

The meaning of pi?

What is the meaning of pi in the formula $${\text{BRI}}=364.2-365.5{\sqrt {1-{\Big (}{\frac {\text{waist circumference in cm}}{\;\;\;\;\pi \;\;\;\;\;\times \;\;\;\;\;\,{\text{height in cm}}}}{\Big )}^{2}}}$$ at Body_roundness_index#Calculation?

Does the BRI formula assume that each waist has the shape of a perfect circle, not an ellipse or an egg?

• a hypothetical totally flat waist would be a 'flat' ellipse, just width, no depth
• such an ellipse would have an area of 0
• a zero area can not contain any fat. Such a hypothetical waist would be totally flat, extremely emaciated
• but while the depth is zero, the waist width could be 80 cm, giving a waist circumference of 160cm
• a waist circumference of 160 would mean a BRI of 13.98 for a 1.78 height
• a GP could have a look at File:Body_roundness_index_graph.svg

13.98

Height 1.78, waist 160, BRI 13.98 in the red zone for a totally flat, emaciated, hypothetical person

for BRI 13.98 at vertical height 1.78, horizontal waist 160 and reach the false conclusion that the totally flat hypothetical person is obese.

• Is the BRI formula only accurate for waists that are circle shaped and a bit off for real life waists, with an ellipse, egg shape?
• Are most waists close enough to a circle, so the BRI won't be far off, especially for obese individuals with a round waist? And obese people will have more health problems, which was the a reason to visit the GP in the first place?
• A GP won't even bother to measure waist if the person is not obese, not emaciated?
• So in real life, the BRI will be pretty accurate?

Uwappa (talk) 05:28, 25 October 2024 (UTC)

It is not our function to question the logic, merely report what RSs say. (But fwiw, the belly of an obese person is very likely to fold down, not stick straight out like a full-term pregnancy. So I imagine that measurement would have to be taken with the patient horizontal, resulting in a very eccentric ellipse.) --𝕁𝕄𝔽 (talk) 09:09, 25 October 2024 (UTC)

It may seem like a small details, not obvious to most text editors, but try to tell that to a silhouettes graphic and a roundness calculator. Still that is what Cmglee and I did, think about the meaning of pi for graphics and the roundness calculator:

1. Kudos to user:Cmglee for the computations to get the right. At the moments those silhouettes are based on perfect circle waist, just like the BRI formula does. So you could say that the silhouettes follow sources.
2. Cmglee's graphic to explain the BRI formula.
3. Note that pi is an absent constant in the WHtR formula. WHtR is not using diameter, it uses circumference, where pi is a rather irrelevant constant that separates those two. The so easy NICE 0.5 health range limit is actually a factor pi off and that is fine. The meaning of pi is... irrelevant.
4. Also note that the roundness calculator shows a silhouette based on WHtR and that is where the same pi story makes a comeback. For small silhouettes that are based on a rounded integer, the difference with non-source reality will probably be sub pixel, not noticeable. Some people even prefer to round WHtR values to just 2 decimals, yet panic about a difference some decimals further down when crossing a NICE health level boundary at 0.5. Very few people text editors realise what is happening under the hood. So be it.

So the Wikipedia sales story: Yes we do follow sources, even when formulas are actually wrong. And it does not matter much because the difference with reality is too small to be of importance. If those differences are acceptable for the whole medical world, they are OK for 20 silhouettes in Wikipedia. A subpixel difference in a silhouettes waist is not a matter of life and death. Uwappa (talk) 10:41, 25 October 2024 (UTC)

The vertical ellipses

Retrospectively added intro: The Thomas et al paper mentioned below uses the vertical ellipses that appeared in earlier versions of this article. It may be that we should not have discarded them as irrelevant. [added --𝕁𝕄𝔽 (talk) 14:52, 25 October 2024 (UTC) ]

I'm concerned that we have introduced an error somewhere? This paper suggests that hip circumference is important too. At one point they say prediction of total % FM in females in the St. Luke’s and Kiel database suggests that HC is a critical covariate for FM especially in females. [btw, they have removed their calculator from https://www.pbrc.edu/research-and-faculty/calculators/ (tbf, the paper dates from 2013), which may suggest subsequent reservations?]

Anyway, they introduce it by saying that

We applied waist and hip eccentricity as opposed to BRI in these formulas since the purpose of the BRI is to generate a visibly simple numerical value that has a wider range than eccentricity values. To develop the statistical models, we revert back to the original directly computed eccentricity.

and they define BRI as Body Roundness Index = 364.2 - 365.5 × Eccentricity NB Eccentricity, not WC.

Moving on, they conclude that

With no additional covariates, the combination of waist eccentricity and hip eccentricity outperformed predictions of % body fat for men and women that relied solely on BMI, WC combined with HC, WC, and HC. We note that this improvement over WC and HC is only slightly better. However, taken together, the best single predictor of both % body fat and % VAT is the combined waist and hip eccentricity.

well this is hardly rocket science, the limitations of BMI are widely known. (But, as the paper acknowledges elsewhere, it is far less intrusive to measure a patient's height and weight than to get the to go on their hands and knees to measure their true waist circumference. BMI is good enough for an indicative measure for most people at the central area of the bell curve.)

Elsewhere, I found this rather worrying statement, a classic description of pseudoscience (assertion first, now find and munge the evidence needed to support it):

Our approach advances several attempts to combine circumference measurements with height to identify body shape by fitting a pre-designated formula to data (11, 12). These existing approaches require conjecture of formula type and a restriction to a few variables that may be important.

Conclusion: I found this paper unsatisfying but the key conclusion that I draw from it is that we can't have an article called "Body roundness" in the abstract because there is no evidence of MEDRS consensus on what it is. There may also be an issue with your calculation given their use of "eccentricity" rather than WC [which aligns with your concern about the use of π, which I agree makes the horrible error of assuming a perfect circle].

Comments? --𝕁𝕄𝔽 (talk) 11:11, 25 October 2024 (UTC) Uwappa (talk) 11:36, 25 October 2024 (UTC)

• Please have a look at Special:NewPages. How many of those pages are based on scientific sources?
• What is the scientific source for length?
• What is for time?
• What is for mass?

Would you go and delete those pages if they are not based on a scientific source?

Now... Please sit back, relax and think:

• For which measure are these things units?

• Body_mass_index
• Body roundness index
• Body_shape_index
• Waist–hip_ratio
• Waist-to-height_ratio
• and ...
• and ...

• On which page would you compare the differences, advantages and disadvantages of BMI, BRI, WHtR, ...?
• On which page would you describe both Emaciation and Obesity?
• Which page would be good to move fat distribution to?

Its is .... ........ Uwappa (talk) 12:19, 25 October 2024 (UTC)

New pages that describe topics with significant medical implications must satisfy MEDRS; any that don't get deleted asap.

Mass, length and time are fundamental units. In "body roundness" we would have a very derived metric that we can't even define. The paper I cited even remarks One possible reason for this difference may be due to higher variability in measurements of WC and HC in standing individuals with higher amounts of body fat, is simply because "flab flops".

None of those topics describe units but they do describe notable topics that are extensively referenced in medical literature. Right now "body roundness" does not have that status.

Anyway, back to my question about your BRI calculation: can you go back to first premises and double check, please? Because this Thomas et al paper explicitly references eccentricity whereas your formula specifies a perfect circle. And if it includes (as they argue it should) hip circumference, then we need to reconsider. 𝕁𝕄𝔽 (talk) 14:35, 25 October 2024 (UTC)

• Please look at the second paragraph of body roundness index: "The BRI models the human body shape as an ellipse (an oval), with the intent to relate body girth with height to determine ***body roundness***." What does body roundness mean there? Is that an unsourced, undefined term? Could it be roundness of the body?
• Look at same paragraph, second sentence: "It is calculated simply using a tape measure to obtain waist circumference and height, which are then combined with an ellipse factor...". What is the ellips factor there, is that a factor for an average waist ellipse that we could use for the silhouettes? If so, where is that factor in the BRI formula?
• Where does the Thomas doc describe an ellipse ***for the horizontal waist***, not for the vertical body?
• Where in the BRI formula do you see variables for waist width and waist depth?

$${\text{BRI}}=364.2-365.5{\sqrt {1-{\Big (}{\frac {\text{waist circumference in cm}}{\;\;\;\;\pi \;\;\;\;\;\times \;\;\;\;\;\,{\text{height in cm}}}}{\Big )}^{2}}}$$

• Where in this graphic do you see waist width and depth?

Where in its version of the formula?

• Please have a look at formulas at Circumference#Circle and Circumference#Ellipse. Which of these 2 do you see in the BRI formula?

Uwappa (talk) 16:08, 25 October 2024 (UTC)

• What do the words "body roundness" mean in "body roundness index"? It means whatever whoever first proposed it, declared it to mean, because it is not a recognised or standard metric (see Anthropometry). I read Thomas et al as challenging that definition because (a) it fails to recognise that midriff-loading is predominantly a male trait whereas hip-loading predominates in women and (b) obese subjects cannot be measured reliably when standing.
• Where does the Thomas doc describe an ellipse ...etc I think what they are saying is that the human body can be enclosed in an ellipse that has height as its long axis and the widest part of the body [waist in overweight men, hips in women] as its short axis.
• Where in the BRI formula do you see variables for waist width and waist depth? I don't. They are assuming that the waist or hips can be enclosed in a circle of diameter = greatest dimension. In slim people that is certainly not true at the hips but may be good enough for their waists. I guess for the target population (overweight, obese), the distinction is not material.
  • But the key issue is that the formula as shown does not have a hip circumference. It should and in their (now deleted) calculator, it did.
• I don't have time now to study it but I'm a bit worried by your conversion of the formula, as it is not immediately obvious that it simplifies like that. Can you explain, please?

• Taking the standard formula for the eccentricity of an ellipse: $$e={\frac {c}{a}}={\sqrt {1-{\frac {b^{2}}{a^{2}}}}}.$$ In this case, b is the half the diameter of the (assumed circular) waist and a is half the height.

𝕁𝕄𝔽 (talk) 16:51, 25 October 2024 (UTC)

The BRI formula as I understand it applied in real life:

1. Get a to measure waist circumference. Don't worry about the shape of your waist, as ellipse, egg and circle shaped waists are all treated equal, as if they are all circles.
2. Yes, I fully agree: this will be good enough for round people and quite wrong for lean people, In real life, lean people are not obese, so their waists won't be measured much by doctors.
3. There is a very small range of WHtR values between 0.2215 and 0.4. The 0.2215 is for Cathie_Jung (smallest waist in the world. The 0.4 is the start of the healthy range. Cathy Jungs waist is not realistic, should be ignored.
4. Mortality goes steep from healty to deadly below WHTR 0.4 (BRI 1.674742). See Zangs figure 5.
5. Emaciated people will have the worst error margin, but they don't see doctors much.
6. I do not know what the smallest waists are in studies. I failed to find studies with WHtR values for emaciated people. There must be data for the lower end, emaciated people. Zefr, do you know any more data for the low end of the WHtR scale?
7. Mortality goes up below WHtR 0.4, see Zangs figure 5. Being too lean is deadly. We should express that will health level colours going rapidly from healthy green via amber and red to deadly darkred
8. My recommendation: define multiple small ranges with colours for WHTR values 0.35 to 0.4. Define colours for 0.35 (deep red), 0.36, 0.37, 0.38, 0.39, 0.4 (shade of green, healthy). Base the colours on Zang's mortality graph as that is the best available source.
9. For the body roundness calculator the leanest silhouette has a BRI of 1 (WHtR 0.35). I expect that to be on the long trail already. Emaciation is deadly.
10. So, the colours for the values below WHtR 0.4should go rapidly from healthy green to deadly dark red. The best source for the colours is Zangs figure 5, mortality data.
11. For the body roundness calculator this is all pretty irrelevant as they are just silhouettes 1 and 2 on the low end of the scale.
12. The NICE health categories have no definition for WHTR < 0.4 (BRI 1.674742). Given the high mortality, my recommendation would be to show a health warning there. Current risk text: "Unspecified for WHtR below 0.4"
13. So, back to the BRI formula. Yes, agree, the very lean have largest error factor, but in real life it is the obese that live to tell the tale and have health problems. That is the target group for the BRI formula and yes, agree those waists are closest to circles, so smallest error margin.
14. So OK, waist circumference measured. That is one variable for the formula. The waist is a perfect circle for the BRI formula, which is good enough for medics around the world in real life, so it should be good enough for Wikipedia. There is no variable to define an ellipse shaped waist, not even a constant for an average ellipse. A circle shaped waist is what the BRI formula is based on.
15. So... onto second input variable. That is height. That is just too easy. We have all the input we need, just 2 numbers. Nice and easy to apply in real life.
16. We could keep it simple here and go for WHtR = waist circumference / height. Too easy!
17. For the WHtR value, 2 decimals is not enough. Go and play and see that similar widths won't even show a difference with just 2 decimals. I would recommend to go back to 4 so each minimal change in waist size results in a different WHtR. People will be able to see gain in just 1 cm waist loss, which can be motivating to keep loosing weight, keep people on the right track.
18. BRI uses a pi to go from circumference to radius or diameter (factor 2 between those 2). And yes, the diameter is for a waist width a perfect circle shape.
19. waist diameter and height define the vertical ellipse. Done!
20. Eh... well almost, there is some further fiddling with numbers. I guess that is just to make BRI ranges easier, a scale of 1 to 18. And no those numbers are not just integers. You can have a BRI value of 2.71828182845904523536 or 3.14159265358979323846. Few people do.

And hurray, we are done with the formula, with just two input variables: height and waist.

• We have 20 silhouettes, 2 more than actually needed, fine. 20 silhouettes is good enough. Silhouettes 1 and 2 look pretty similar. It must be colours that make the difference on the low end of the scale.

• We may need to define more colours, especially between 0.35 to 0.4 and the best source I could find for those colours is Zang's mortality graph.

Uwappa (talk) 18:30, 25 October 2024 (UTC)

As I understand it, the input variable for male and female is the same, just waist and height, no hip in BRI.

It would be for a general article on body roundness listing various methods to measure body roundness. Yes, I know, that is a no no.

There is Waist–hip_ratio but that is another story, another article.

Another idea: How about a poor man's choice: a category body roundness?

That will be one way to group BMI, BRI, WHtR, waist-hip, ...

Would that be OK?

The difference is in the classification of the BRI value. See the online commercial BRI calculator. It uses a cyan ellipse for healthy range. Change the gender and see that cyan oval change for health level.

For the calculator the male/female difference is for a future version. Let us get going with background colours and health levels first. One step at the time. That is tough enough already. I am still waiting for some key colours to compute new WHtR based gradients.

Please assist and update Template_talk:Body_roundness_index#WHtR_based_colours,_silhouettes_and_text_for_Calculator_4.0 similar to what you did last time, but now for WHtR related numbers.

Yes agree. The only ellipse in the BRI formula is the vertical one, based on height and diameter.

Agree, the largest target group that will use the formula has a waist close to a circle. Good enough for medics around the world, good enough for us.

Do you mean the formula in the 3 steps BRI graphic? That is not my work. Go and ask User:Cmglee about the disappearing two's. I just focussed on the formula in the BRI article. Uwappa (talk) 18:54, 25 October 2024 (UTC)

For a design idea that:

• transforms data to knowledge
• in line with function psychology principles: computer does a lot of work for the human, interface is deceptively easy, requires minimal input, does not require thought
• WP:NOTOR
• WP:OI
• WP:PROOF
• WP:RS

see

Template_talk:Body_roundness_index#Information_hierarchy

Thank you for the thought provoking questions! Uwappa (talk) 01:41, 26 October 2024 (UTC)

Done:

• in information hierarchy, explaining the formulas for WHtR and BRI at the level one above basic input.
• moved WP:RS one level up, away from user input. Data from reliable sources is in formula's, invisible, but still reliable
• Added WP:NOMEDICAL to top level of hierarchy. Please check text by expanding the collapsed guidelines.

user:Cmglee please remove the formulas from the graphic so it can be used at many pages, with possible different versions of the formula. The graphic support all formula versions. Moved graphic from intro to Body_roundness_index#Calculation. That chapter now has two explanations of the formula. Uwappa (talk) 04:28, 28 October 2024 (UTC)

Zang's figure 5 for mortality

user:Zefr Would you like to add Zang's figure 2 to the BRI artile. Is is allowed to use that graph? OK with WP:IMAGEPOL?

Alternative: Do you know any public data that is available for an alternative graphic? Uwappa

What I would like to create: a deeplink from the future body roundness calculator documentation to those mortality figures, to explain the background. Or maybe a ref direct to Zang's figure 2? It feels a bit odd to add ref's to the calculator interface.

See Template_talk:Body_roundness_index#Information_hierarchy for how those mortality figures were the source for Template_talk:Body_roundness_index#Colours_for_Body_Roundness_4.0, which may going to be gradients.

Uwappa (talk) 10:00, 26 October 2024 (UTC)

calculation has 2 different versions of formula, WP:NOTHOWTO?

Moved formula graphic and its text from intro to chapter calculations.

That chapter now needs some fine-tuning:

• there is 2 explanations of the formula and they do not align, confusing
• doubt about WP:NOTHOWTO based on earlier comments (by JMF if I remember correctly). Should the bit about how to measure waist be removed?

Uwappa (talk) 04:32, 28 October 2024 (UTC)

ncbi new source?

Any new info in https://pmc.ncbi.nlm.nih.gov/articles/PMC8582804/ that could be included in BRI article? Or all covered already? Uwappa (talk) 10:27, 1 November 2024 (UTC)

We don't have an article about ABSI so it wouldn't be easy. The main value would be to add another MEDRS that recognises BRI, though it damns it with faint praise:

BRI and ABSI have discriminatory power for hypertension in adult women and men from different populations. Although, WHtR and WC provided the best performance when assessing hypertension, no significant differences were found for BRI. Finally, BRI was significantly better predictor of hypertension than ABSI.

In other words, don't waste your time using a complicated formula to work out a BRI when WHtR does the job easier and better.

But here's an interesting observation: since we know that BRI and WHtR are mathematically the same, how did they find any difference whatever? (even though it was not statistically significant). So maybe in a "further reading" section? --𝕁𝕄𝔽 (talk) 10:54, 1 November 2024 (UTC)

Sorry mate, the medical jargon is like 汉语 to me.

Had I known the answer, I would not have asked the question.

Agree, WHtR can not be better than WHtR mathematically.

It can be better for practical reasons

• easier, even too easy. A 3 year old in outback Australia could check the whole local community for > 0.5 value, while the white fellows are struggling because their calculator batteries have gone dead.
• more data, more research available.

That easiness shows in the calculator code,

dive in and see at the very bottom:

• whtr: formula=waist/height
• bri: formula=364.2-365.5*sqrt(1-pow((whtr/pi),2))

I love this simplicity in the code.

What do you think about the following idea:

1. show the WHtR formula and show a 'filled in version too', a copy of the hight and waist in the calculator.
2. That is too easy, because calculator fields can cooperate when in the same page, even if they come from different templates.
3. so it could show the math formula followed by: = 178 / 80 = 0.449438202247191 That would support WP:VERIFY, increase confidence n the calculator.
4. Same story for the BRI formula, show the simpified version, the WHtR just shown above.

That could read as:

math formula with WHtR as input

= 364.2-365.5*sqrt(1-pow((0.449438202247191 /3.141592653589793),2))

= 364.2-365.5*sqrt(1-pow((0.449438202247191 /3.141592653589793),2))

= 364.2-365.5*sqrt(1-pow(0.143060623,2))

= 364.2-365.5*sqrt(1-0.020466342)

= 364.2-365.5*sqrt(0.979533658)

= 364.2-365.5*0.989713927

= 364.2-361.740440319

= 2.459559681 (matching the number in the calculator)

Again verifiable, and understandable for the rest of us.

That will: increase confidence in the calculator, make it obvious that it really is just a calculator crunching numbers

• inviting people to play with different input, ha ha, look how all those numbers change.
• ha ha ha, Wikipedia is now showing the height and width of my cat! I did that!
• I'm going to show that to my mates in the pub tonigh. Look what I did...
• O bummer, it has no memory, next time it starts fresh.
• Well that's a fair go, so no privacy issues here.
• I can feed it my own real height and width.

Uwappa (talk) 12:37, 1 November 2024 (UTC)

The paper is about a statistical analysis, not about medical conditions, so no physiology to understand. It adds nothing new so it would be a "make weight" in this BRI article; if we had an article about ABSI, it would have value because it denounces it. It is even less relevant to the WHtR article.

As for the rest of your post, I'm afraid I have to see it as flogging the topic to death now. If you really think it worth doing, then only place for it would be the template documentation page. It definitely doesn't belong in the BRI or the WHtR articles. 𝕁𝕄𝔽 (talk) 15:51, 1 November 2024 (UTC)