60% of TRD Relapses Go Undetected. Here's Why — and What Fixes It.

There's a number that keeps coming up in conversations with IP clinic directors: 60%.

Sixty percent of relapses in TRD patients go undetected by the treating clinic. The patient deteriorates. They stop responding to messages. They miss appointments. Eventually, they disappear — and the clinic only finds out weeks later, when it's far harder to re-engage them.

This is not a failure of clinical care. It's a structural problem with how psychiatric monitoring works between visits.

The monitoring gap

Consider the timeline: a TRD patient completes acute TMS treatment and enters maintenance. They come in once a week, then every two weeks. In between those visits, there are 166 hours per week — and in standard-of-care, those hours are essentially invisible to the clinical team.

The most common monitoring tool between visits is the PHQ-9 by email. Completion rates average 30–40%. That means 60–70% of your patients have no monitoring at all between visits. Of those who do complete it, many do so during a relatively stable window. The PHQ-9 they fill out on a Tuesday afternoon doesn't capture what they experienced Saturday night.

And here's the insidious part: the patients most likely to miss that PHQ-9 email are the ones who are relapsing. Depression reduces motivation, executive function, and ability to engage with administrative tasks. The monitoring system fails precisely where it matters most.

What the data shows

Across our clinical studies — conducted with partners at Yale, Harvard, Johns Hopkins, UCSD, McGill, Charité Berlin, and GHU Paris — we see a consistent pattern. Passive monitoring detects objective changes in facial expression, vocal biomarkers, physical activity, and digital behavior approximately 48 hours before a patient would endorse clinically significant worsening on a standardized scale.

That 48-hour window is the difference between a proactive call from a clinical coordinator and a no-show at the next appointment.

The fix is passive, not active

The failure of active monitoring tools (surveys, check-ins, journaling apps) is predictable: they require a depressed patient to do something. Passive monitoring removes that dependency entirely. The EmoDTx app processes four signal streams — facial expressions, voice, actigraphy, and digital behavior — entirely on the patient's device. No surveys. No check-ins. No daily prompts.

The result: 100% monitoring coverage, regardless of patient symptom severity. The patient who's too depressed to open an email is still being monitored.

And the clinical outcome: the 4× net ROI documented in interventional psychiatry clinics using Emobot is driven almost entirely by one mechanism — preventing the patients who would have silently deteriorated and dropped out from doing so.

What this means for your practice

If you're running a TMS or Spravato program, the question isn't whether to monitor between visits — it's whether your current monitoring approach reaches the patients who need it most. If your PHQ-9 completion rate is below 50%, it doesn't.

The 60% undetected relapse rate isn't inevitable. It's the product of a monitoring approach that was designed for a different era. Passive monitoring changes the equation.