Episode Details

This is the Weight and Healthcare newsletter! If you appreciate the content here, please consider supporting the newsletter by subscribing and/or sharing!

I have had several readers ask about the Lark program. In each case, it was being actively marketed to them, without their consent, by their insurance company and/or employer.

Lark is, predominantly, a weight loss program that also bills itself as a “diabetes prevention” program and “mind body” health coaching all offered by an Artificial Intelligence powered app. (Note that while it has various health programs, the big font headline at the top of the website says “Lose weight and be your healthiest self with Lark.”) It is covered by some insurance, and people can “earn”  things like a scale and a fitbit by meeting “minimum program engagement requirements, such as weighing in, completing missions with your digital coach, and logging activity or meals.”

The graphic on their website that they are calling “clinically proven results” says:

5.3% average weight loss in year one (1)

40% achieving greater than or equal to 5% weight loss in year one (1)

9.6lbs average weight loss by Lark members beyond 2 months in the program (2)

(1) Based on Lark program data for members meeting CDC qualification criteria(2) For those with available weight data

These results are from the study Weight Loss in a Digital App-Based Diabetes Prevention Program Powered by Artificial Intelligence,  Graham SA, Pitter V, Hori JH, Stein N, Branch OH. Accepted, Digital Health, vol 8. 2022. DOI:10.1177/20552076221130619 (as always, I don’t link to studies that pathologize higher-weight bodies, but I provide enough information for those who are interested to find them on Google.)

The study’s lead author is Sara Graham who, according to LinkedIn is the “Clinical Research Manager at Lark Health”, in fact, per the conflict of interest statement, every single study author is employed by Lark.

Let’s dig in, shall we?

Let’s start with the study’s stated conclusion:

[Lark] facilitated weight loss and maintenance commensurate with outcomes of other digital and in-person programs not powered by AI. Beyond CDC lesson completion, engaging with AI coaching and frequent weighing increased the likelihood of achieving ≥5% weight loss. An AI-powered program is an effective method to deliver the DPP in a scalable, resource-efficient manner to keep pace with the prediabetes epidemic.

When they say “commensurate with outcomes of other digital and in-person programs” what I think you’ll find they mean is – people didn’t lose much weight, just like with every other program as we have seen in research for the last hundred years.

And when they say it’s an effective way to deliver the DPP [Diabetes Prevention Program] it’s important to know that they did not, in any way, measure blood sugar, or anything other than weight and weight loss.

But let’s really dig into the numbers:The one-year study started with 3,933 Lark Members, but at its 12 month conclusion only 414 members had provided beginning and ending weights.

Those 414 were divided into two groups for comparison:

Group 1: CDC qualifiers: Completed ≥4 educational lessons over 9 months (n = 191)

Group 2: CDC non-qualifiers: did not complete the required CDC lessons but provided weigh-ins at 12 months (n = 223)

So, did you catch that? They started with 3,933 members and only 191 actually managed to stay with the program for a year. (The other 223 “group 2” didn’t complete the program but did provide their weight at the beginning and at 12 months.)  

My first question is: Do the results matte