We have been skeptical of digital therapeutics for most of the last five years, and we have been wrong about some individual companies while being essentially right about the category-level dynamics. The DTx hype cycle ran faster than the evidence, the FDA breakthrough device designation was applied to products that did not warrant it in retrospect, and the reimbursement story that was supposed to drive commercial scale has been much slower to materialize than the pitch decks of 2020-2022 implied.
That said: some digital therapeutics for metabolic disease are producing clinical evidence that is credible, reproducible, and in some cases comparable to what you would expect from a modestly effective pharmacological intervention. The job now is separating those from the majority that are not.
What "Evidence" Should Mean
The word evidence in the digital health context has been so thoroughly abused that it requires definition. We mean randomized controlled trial data showing statistically significant improvement in a validated clinical endpoint — not an app engagement metric, not a patient-reported outcome measure with no established correlation to clinical outcomes, and not a surrogate endpoint chosen for feasibility rather than clinical relevance.
For metabolic disease specifically, the relevant clinical endpoints are established: HbA1c reduction in type 2 diabetes, body weight reduction and maintenance in obesity, blood pressure control in hypertension. These have known clinical significance and payer-accepted thresholds. Any company claiming clinical evidence for a metabolic digital therapeutic that cannot point to randomized trial data on one of these endpoints is making a different claim — possibly a valid one, but a different one.
The companies that have done this correctly — and a few have — ran trials that looked like pharma trials, not digital health pilots. Pre-specified primary endpoints, independent data monitoring committees, per-protocol and intention-to-treat analyses. It is expensive and slow. It is also the only way to know if the product actually works.
The GLP-1 Question
The emergence of GLP-1 receptor agonists as the dominant pharmacological approach to obesity and type 2 diabetes has changed the competitive context for metabolic digital therapeutics in ways that have not been fully worked through.
The bull case for digital therapeutics before GLP-1s was that they could produce meaningful weight loss and glycemic improvement through behavioral mechanisms, at low cost and without the side effect profile of existing drugs. That value proposition is harder to articulate now that a drug class exists that produces larger effect sizes on the same endpoints with strong clinical outcome data.
The more defensible position for digital therapeutics in the GLP-1 era is as combination or adjuvant therapy — digital tools that support medication adherence, help patients manage the behavioral components of weight management, or extend the effects of pharmacological treatment into domains the drug does not reach. That is a different product positioning and a different commercial model than the standalone DTx pitch, but it may be more sustainable.
Companies that have started to reframe their value proposition around GLP-1 combination, or that have proactively partnered with payers to test hybrid drug-plus-digital programs, are thinking about this correctly. Those that are still pitching standalone digital therapeutics for obesity against the current standard of care have a difficult evidence gap to close.
What Good Looks Like
The metabolic digital therapeutics that have generated credible evidence share a few characteristics.
First, they target a specific mechanistic pathway — usually cognitive behavioral approaches for eating behavior modification, continuous glucose monitoring with closed-loop intervention, or structured physical activity programming — rather than generic "wellness" or "lifestyle" features. The therapeutic mechanism is described with enough specificity that a clinical trial can be designed to test it.
Second, they have run trials with adequate sample sizes. Most of the underpowered 50-100 person pilot studies that dominate digital health evidence do not tell you whether the intervention works — they tell you whether it might be worth running a real study. Real studies need to be powered to detect clinically meaningful effect sizes and the companies that have done this have typically run trials of 300-1000+ participants over 12 months or longer.
Third, they have worked through the regulatory pathway with FDA early and gotten feedback on their endpoint strategy before investing heavily in trials. Several digital therapeutic companies have lost years on underpowered studies with endpoints that FDA did not consider clinically valid for a marketing claim. That is an expensive mistake that disciplined early engagement with the agency avoids.
Our Current View
We are not actively investing in standalone digital therapeutics for metabolic disease at this stage of the market cycle. The evidence bar has risen, the competitive context from pharmacological alternatives is tougher, and the reimbursement landscape has not evolved at the pace the 2020-era projections assumed.
Where we remain interested is in the intersection of digital tools and pharmacological programs — specifically, companies building digital components that are co-developed with drug programs as combination regimens or clinical trial infrastructure. The clinical operations angle is also interesting: digital endpoints and digital data collection platforms that enable more efficient and more patient-centered clinical trials for metabolic and other disease areas have a business model that does not depend on the standalone DTx reimbursement story working out.
The companies that have genuine evidence can raise from us. The ones that are still defining clinical endpoints after Series B cannot.
