Digital therapeutics (DTx)
Software as the treatment itself, not the wrapper around it. The category is real, the evidence bar is a device trial — and the business model has been the hard part, not the science.
In one line
A digital therapeutic is software that is the treatment — delivering a clinical intervention directly to the patient, with evidence to back the claim. Not a tracker, not a reminder, not a wellness app: the software is the therapy.
The distinction that defines the category
Most health apps help you manage a condition. A DTx treats it.
| Wellness app | DTx | |
|---|---|---|
| Claim | "Feel better", "track your steps" | "Treats insomnia", "reduces HbA1c" |
| Evidence | Testimonials, engagement stats | Randomised controlled trials |
| Regulation | Largely none | SaMD — a regulated device |
| Route | App store | Often prescribed |
The clearest examples are digital CBT — cognitive behavioural therapy for insomnia is the canonical one, because CBT-I is a well-evidenced protocol that turns out to deliver perfectly well through software. There's no pharmacology to replicate; the therapy is structured interaction, and software does structured interaction well.
That's the honest boundary of the category: DTx works where the therapy was always information and behaviour. It works less well where the therapy is a molecule.
The evidence bar is real
This is the part that separates DTx from the app store, and it's why the category deserves respect. A genuine DTx runs clinical trials with clinical endpoints — not engagement, not downloads, not "users report feeling better." Reduced HbA1c. Reduced insomnia severity index. The same currency as a drug.
NICE's evidence standards framework is worth knowing here: it grades digital health technologies by the claim they make, and it asks for evidence proportionate to the risk. That's the right shape — and it exists because the field was flooded with products claiming outcomes they'd never measured.
Why the hard part hasn't been the science
The instructive story of DTx is commercial, not clinical. Several high-profile companies built products with genuine trial evidence — and struggled or failed anyway. The reasons are worth understanding, because they're structural:
- Nobody knew who pays. A drug has a reimbursement pathway built over a century. A prescribed app fits nowhere in it. Germany's DiGA scheme is the notable attempt to build one.
- Prescribing friction. A clinician has thirty seconds. Prescribing an app means explaining it, and it competes for that time with a drug they can write in five seconds.
- Engagement decay. The trial had support staff and motivated volunteers. Real patients stop using it in week three, and the effect size in the trial assumed they didn't.
- The comparator got free. Your evidenced CBT-I app competes against a mediocre free app, and the patient cannot tell the difference from the store listing. Evidence is invisible at the point of choice — which is the deepest problem in the category.
That last point is the one that should interest anyone building here: a market where quality is unobservable to the buyer doesn't reward quality. It rewards marketing.
India
Largely an open field. There is no DiGA-equivalent reimbursement pathway, CDSCO's device framework technically applies, and the practical market is out-of-pocket or employer-paid.
The opportunity is real and the caution is the same as everywhere: without a payer, a DTx is a consumer app with a trial attached, and it will compete on the terms consumer apps compete on. The scientific work is necessary and nowhere near sufficient.