This is Optimizer, a weekly newsletter sent from Verge senior reviewerVictoria Songthat dissects and discusses the latest gizmos and potions that swear they’re going to change your life. Opt in for Optimizer here.
The Concept of Metabolic Twins
In return for feeding a digital twin my metabolic data, it predicts how I'll react to certain foods and, one day, maybe medicines.
For the past 100 days or so, my mornings have had a new rhythm. When I wake up, the first thing I have to do is feed my digital twin. It consumes my metabolic data, and its first meal of the day is my body composition data, sent straight from a smart scale. Once that's logged in the app, I scoot over to the kitchen to fill an Owala with lemon water to take my medications with. That, too, gets logged in the app. On Mondays, I'll follow that up with several calming breaths, measure my blood pressure, and offer that to the digital twin as well. Then it's time to consider breakfast. I consult my digital twin.
My digital twin isn't so much a robotic clone as it is an AI-powered, color-coded food database that predicts how my glucose levels will react to everything I eat. Red is not so good. Orange is so-so. Green is ideal. It's built based on the data I feed it, as well as data collected from a continuous glucose monitor.
A bowl of protein Cheerios and milk? My digital twin tells me this is orange if eaten alone, but green if I take my daily walk after. So that's what I decide to eat and do. When that's all done, I log it in the Twin Health app. In a few hours, the app will generate an insight based on how my glucose reacted, and my twin will become more accurate. For now, my twin is fed.
Rinse and repeat for roughly another 99 days.
Digital twins are a relatively new concept in physical health, both in medicine and sports. It's a virtual model of a person built from various data points — labs, wearable data, images, genetic profiles, etc. — that is continuously updated over time. Once built, clinicians can then use AI to see how a patient might react to specific medicines or treatments by first testing them out on their digital twin. Theoretically, this could also help healthcare providers assess injury risk, predict disease progression, and implement preventive care while minimizing adverse effects for the actual patient. The thought is that this could one day help realize the dream of personalized medicine. Metabolic coaching is perhaps the lowest-stakes, highest-reward version of this technology that can be tested in the real world now.
The white scale is my Twin Health scale. When not building a digital twin, I sometimes use it to help me keep Pablo’s weight in check.
Real-World Implementation
My metabolic twin was born in June with a giant box from Twin Health delivered to my doorstep. Twin Health is a digital twin program to help people heal their metabolisms, such as losing weight or managing diabetes. It's not available for widespread download just yet. Digital twins are still experimental, and currently, this service is only accessible through corporate or private health plans. It's typically aimed at people with chronic metabolic issues. Twin Health provides everything you need to build a digital twin. Building mine would require 90 days worth of Dexcom G7 CGMs, a smart scale, a smart blood pressure cuff, three bottles of multivitamins and magnesium glycinate, and my Apple Watch. That was followed up with a trip to Quest Diagnostics for further analysis.
A few days later, I hopped on Zoom with my Twin Health coach Amanda and a doctor to review my lab results. Several numbers and formulas were thrown up on a shared screen. As it turns out, my liver was inflamed, consistent with a moderate case of nonalcoholic fatty liver. Insulin resistance was present despite ideal HbA1c and fasting glucose tests. My triglycerides were also high, though the rest of my lipids looked normal.
I knew going in that I inherited a bunk metabolism. A year of testing CGMs as a non-diabetic made that abundantly clear. But after the doctor logged off and Amanda asked me to define my goals, I launched into a spiel about my vague desire to reduce the number of medications I needed to keep my middle-aged metabolism functioning. Could a digital twin help me do that?
Maybe. For now, Amanda logged my goal, set up a follow-up call in two weeks, and encouraged me to try logging as many of my meals as possible for the first 30-day period. Once I hit the 90-day mark, I'd be scheduled for another blood draw.
In many ways, I'm relieved my digital twin doesn't have its own avatar. At least, not one that I can see. I'm glad for the abstract color labels. There's novelty in going to see what color a cup of pineapple or a turkey hot dog with ketchup and relish might be. At times, I am reminded of Kazuo Ishiguro's dystopian novel Never Let Me Go, in which clones are raised as organ donors. The scenario isn't a one-to-one comparison. An AI-powered digital twin can't actually eat, die, or feel pain.
Data Fatigue and Behavioral Shifts
Still, as the months ticked by, I couldn't help noticing how hungry my digital twin was. Building this thing takes an immense amount of personal data. In my bimonthly check-ins with Amanda, she noted that tracking fatigue was frequent feedback from Twin Health users.
My behaviors did change in both unexpected and expected ways. Anytime I have to log food, I end up avoiding unhealthy treats more often than not. That didn't result in any meaningful weight loss, however, and my goal of tapering off medications did not go as smoothly as hoped.
I also didn't learn too many surprises from my digital twin's metabolic predictions. Though it was validating to see that a bowl of soondubu jjigae and white rice is a solidly green or orange meal depending on physical activity, I've tested far too many CGMs and AI nutrition apps to feel surprised by these insights.
Twin Health's bimonthly check-ins did motivate me to get back to training. The extra accountability of having to state my progress aloud to another human kicked my inner overachiever into gear. After months of feeling too weak from medication side effects, I recently felt strong enough to run an uninterrupted 5K and return to strength training.
This is an orange food because protein bagels are still bagels.
Results and Medical Improvements
Most importantly, after 100 days, my follow-up labs came back with marked improvements. Amanda says my metabolic health score has jumped by nine points since joining. A year after my doctor's eyes bugged out at my liver enzymes, all but one are back to normal. My triglycerides dropped by 70mg/dL, and there has been a significant drop in hs-CRP, a biomarker for overall inflammation. I'm still insulin resistant, but less so.
I can't really argue with results like that, can I?
The Burden of Trust and Privacy
I can, actually. The concept of a digital twin is founded on trust. As I said, metabolic use cases are high-reward and low-stakes. If you feed a twin an immense amount of data for losing weight or better energy, failure isn't immediately life-threatening.
Personalized medicine is a different beast. Researchers are looking at digital twins for things beyond metabolism, such as cancer care and research. There's massive potential there, particularly when you consider how grueling chemotherapy can be, but also huge challenges like the burden of trust for data privacy. Digital twins are hungry for data, and their accuracy depends on a continuous stream of sensitive information.
The CGM experiments continue, as my digital twin hungers for this data.
As for me, I've decided my digital twin will get to live for another three months. Despite my lack of success so far in tapering off medications, I could use all the help I can get in getting my liver labs all the way back to normal. Amanda also says I don't have to be as militant with the data for my twin now that I have a solid foundation.
So, today, I'm asking my digital twin what it thinks about a protein waffle with a tablespoon of peanut butter and some honey, with a side of pineapples. Orange, but again, I can turn it green with a walk.
