Body fat scale with public API? Withings vs. Fitbit? Or something else?

LNP · August 18, 2020, 1:00pm

Currently, I’ve been using my normal weight and the U.S. Navy seals method for calculating body fat, but I would like to “automate” the process, so I only have to stand on a scale.

This is what I want:

Weight, Lean body mass & Lean fat mass
Relative measure of progress (It doesn’t need to be absolutely correct)
A good Public API, so I can automate the export

But I was afraid of buying the withings, because some people say they have a really bad API:

gist.github.com

https://gist.github.com/katemonkeys/e17580777b57915f5068

gistfile1.md

# Top Six Things You Need To Know About The Withings API*

##### *where “you” is probably a developer, or at least a strange user

*I should preface this by saying that I got a Withings Smart Body Analyzer for Christmas last year and I’ve been generally happy with it. It purports to be able to take my heart rate through my bare feet and that seems not to work for my physiology, but overall I’m a fan. If if their [Wikipedia page](http://en.wikipedia.org/wiki/Withings) is to be believed they are having a pretty rad impact on making the Quantified Self movement more for normal people and they only have 20 full time employees. Also they try hard to use SI units, which I can get behind. Anyway, on to the rant.* 

I originally called this post “Everything wrong with the [Withings API](http://oauth.withings.com/api/doc)” and I meant it. For every useful field I can extract from their “award winning” app, I have spent an hour screaming at the inconsistencies in their implementation or inexplicable holes in their data coverage. This is as much a post for developers looking for reasons why they are tearing their hair out trying to access an unruly metric as it is for anyone who provides a third party API that is intended to be consumed by anyone not themselves, and what us poor data consumers have to deal with. 

### It’s inconsistent about formatting 
Sometimes we assemble a query for startdate (which needs to be formatted as a time in epoch seconds, for sleep detail measures) and sometimes for startdateymd (which returns a starttime in epoch milliseconds for sleep summary measures). Some measures return device identifiers: a model 16, for example, is their Pulse watch, and a model 32 is their Aura bedmat. What then am I to make of a sleep data object reporting a model 61??

This file has been truncated. show original

What do you think is the best choice for me?
Withings? Fitbit? Something else?

ejain · August 20, 2020, 4:25am

I have been using an older model of their body composition weight scale for almost 8 years now, and it’s given me no trouble. Their API isn’t going to win any awards, but it’s been stable, and you shouldn’t have any trouble getting body mass + fat mass.

As with any such device, you’d have to see how closely their fat mass estimates track your manually measured values.

dreeds · August 20, 2020, 6:46am

I use a cheap Renpho Bluetooth weight scale. https://www.amazon.ca/RENPHO-Bluetooth-Body-Fat-Scale/dp/B01N1UX8RW/
The included app easily syncs with Google Fit. Google’s API requires proper Oauth2 authentication but once you get over that hurdle, the daily data from the scale is solid.

Other benefits of the scale is long battery life and it’s rechargeable via a micro-usb cable. Also, it seems to be relatively accurate… if I weigh myself 3 times in 55 minutes, it’s usually within .2 of a pound or less.

I’m not associated with Renpho, I’ve just used the scale for more than a year and been happy with it and it’s easy linking to Google Fit.

xman2012 · August 20, 2020, 11:13am

I’ve also used a Renpho scale, and I like it as well. It also connects to Apple Healthkit.

xman2012 · August 20, 2020, 11:19am

@dreeds Do you have any sense how accurate the body composition measurement is? Just curious.

LNP · August 20, 2020, 2:24pm

I did also consider buying a cheaper scale that syncs with Google fit or similar, because then I can just get the data from the Google fit API There wouldn’t be any downsides to that? Right?

dreeds · August 23, 2020, 5:11am

I don’t have any other scales and I haven’t done any other body composition tests (dexa scan, bod pod, hydro-static, calipers, etc) so I can’t say. But I do know the scale is better than previous scales I’ve owned… I just put it in three different rooms within a few minutes and got the exact same readings. So, the variance is small, though I don’t know how to test for accuracy.

But for me, I focus more on the trends than the absolute numbers. For example, if my weight, bmi and subcutaneous fat percentage are all going in the wrong direction, it’s a pretty clear indicator that I need to take action.

The scale I use has almost 40,000 ratings on Amazon and it’s sitting at 4 1/2 stars out of five, so it’s meeting most people’s needs.

The other thing I didn’t mention is that it works for multiple people… I added the app to my partner Christine’s phone and linked it to Google Fit. She uses it almost daily. EventLoggers pulls her weight from Google Fit and all she has to do is stand on the scale, so she’s happy too. YMMV

LNP · August 24, 2020, 3:10pm

What kind of data can you get from the Google fit API from the Renpho scale?
Can you both get weight, Lean body mass & fat mass?

dreeds · August 25, 2020, 6:00am

My apologies - I link Renpho data to my Fitbit account (just weight) and not Google Fit, so I can’t comment on what metrics are available to Google Fit. For Fibit, it looks like it’s only weight. Sorry for the mistake… so many APIs!

SimonP · September 5, 2020, 1:08pm

This might be of interest: https://github.com/oliexdev/openScale

Open source app and a list of scales with which it works.

I’ve not used it myself, but stumbled across the page while looking for cooking scales with an open API.

LNP · September 6, 2020, 9:47am

Are you sure they have an API? It doesn’t say anything about API on the wiki.

Also they support Renpho ES-CS20M, but I think I want to buy Renpho ES-CS20M-B

kristenpe · April 16, 2022, 9:38pm

There are many special methods used to measure body fat. A method involving equations developed at the Naval Health Research Center by Hodgdon and Beckett in 1984. Please note that the results of these calculations are approximate, as they are based on a variety of different assumptions to make them as applicable to as many people as possible. For more accurate measurement of body fat, it is necessary to use tools such as bioelectric impedance analysis or hydrostatic density measurement. In addition, you can use the bathroom scale.

Kyriakos · April 22, 2022, 11:46am

I’ve been using Withings for years, but to be fair the API is not good enough. Also the body fat measurement is never accurate. Having said that, it really depends on the ecosystem - eg if you use a garmin watch, i’d also use a garmin scale

maith · May 11, 2022, 7:42pm

There are a few like fitbit, garmin, withings, the cheapest is Renpho! We’ve built an integration for it, let me know if you want access.