← Blog/Clinical Insights

Digital Health Had a Patient Effort Problem. Passive Monitoring Makes Workflow Integration Optional.

Tanel Petelot·December 15, 2025·9 min read

Every previous generation of digital health required patient effort — and every one of them needed workflow integration to compensate for the adherence problem that followed. Passive monitoring removes the first link in that chain, and the whole chain collapses.

Most serious evaluations of Emobot begin with the same technical question from the clinic's operations lead: "How does this integrate with our EHR? Which workflow does my staff need to run?"

It is a reasonable question — because for the last fifteen years, the answer from every digital health vendor has been some version of: "It lives inside your EHR, your MA has to nudge the patient, and here's the training plan for your clinical staff." That is what "clinically deployable" has meant since the first generation of patient-reported outcome tools shipped.

Our answer is different. Emobot does not require EHR integration. It does not require the MA to do anything. It does not require a clinical workflow at all. The patient installs it once, and the data shows up in a dashboard next time the physician opens it.

That answer almost always prompts the second question: "Then how do you get patients to actually use it?" — which is the question this post exists to answer. The short version: we do not have to, because we removed the thing digital health was asking them to do.

Why traditional digital health needed workflow integration in the first place

This is not a knock on the previous generation of digital health tools. The people who built PHQ-9-by-email platforms, symptom tracking apps, journaling tools, and patient-reported outcome dashboards were solving a real problem with the constraints they had. Understanding those constraints is the key to understanding why passive monitoring is a genuinely different paradigm — not just a faster horse.

The constraint was simple: every previous generation of digital health required patient effort. A survey to fill out. An app to open. A mood rating to record. A journal entry to write. A daily check-in to acknowledge.

And the moment a product requires effort from a psychiatric patient population, two things happen:

CONSEQUENCE 1

Adherence craters.

Industry baseline for PHQ-9 by email sits at 30–40% completion. Mood-tracking apps typically show 90-day active user rates below 20%. The patients most likely to drop out are the ones whose symptoms are worsening.

CONSEQUENCE 2

The data biases toward stability.

Patients who complete surveys are over-represented in stable windows. The patients who most need monitoring are under-represented — because they are not filling things out. The tool fails exactly where it is most needed.

Digital health vendors responded to this problem the only way they could given the architecture: they pushed the burden onto the care team. Workflow integration was not a feature — it was a workaround for poor patient adherence.

If the patient will not fill out the survey on their own, the MA reminds them. If they still will not fill it out, the nurse calls. If the data is not flowing, the care coordinator chases. Integration into the EHR was the mechanism that made that chasing structurally possible: alerts, task queues, dashboards that prompted staff action.

That works — up to a point. It also consumes a significant fraction of clinical staff time, creates alert fatigue, and puts the product's success rate in direct competition with every other clinical task the MA and nurse have to do that day. Above a certain patient volume, it stops scaling.

THE CAUSAL CHAIN OF TRADITIONAL DIGITAL HEALTH

Patient effort
required
Poor adherence
(~30% completion)
EHR integration
+ workflow nudges
Care team
burden

Every link was necessary given the starting constraint. Remove the starting constraint, and the whole chain goes away.

What changes when the patient has nothing to do

Emobot's passive architecture removes the first link. The patient installs the app once — a three-minute process the physician demonstrates in the exam room. After that, the app runs in the background. There is no daily check-in. There is no weekly survey. There is no mood rating to log. There is no prompt the patient has to respond to.

What the AI does instead: four signal streams process entirely on the patient's device — facial microexpressions during the screen time they already spend, vocal acoustics during the calls they already make, actigraphy from the motion sensors already in the phone, and digital behavior patterns from the device usage they already have. All of it on-device. None of it content-level. What leaves the phone is a two-number mood index, nothing else.

That architectural change has a downstream consequence most operators do not anticipate on first pass: the adherence problem disappears, and with it the rationale for workflow integration.

Look at the retention numbers side by side:

METRIC PHQ-9 BY EMAIL SYMPTOM APPS EMOBOT
Initial activation ~50% 30–50% 80%
30-day active rate 15–25% 20–30% >90%
90-day active rate <15% <20% >90%
Care team nudges required Yes Yes No
EHR integration required Typically yes Often yes No

These are not marginal improvements on an old model. They are the output of a different model entirely. A 90%+ monthly retention rate in psychiatric outpatients is not achievable with any active tool. It is achievable here because the retention question is architectural: once the app is installed, staying active requires zero effort. Uninstalling would actually take more effort than doing nothing.

Why the "integration conversation" is different with passive monitoring

When we tell an IP clinic operator that Emobot does not need to be integrated into their EHR, the typical reaction is skepticism — because the last ten vendors told them the same thing and then quietly asked for read access to Epic three weeks into the rollout. That skepticism is fair. Here is why the claim actually holds in our case.

Integration serves two functions in traditional digital health:

FUNCTION 1 · PATIENT ACCESS

Getting the tool in front of the patient at the right moment in the visit.

Passive monitoring: not needed. The patient already has it installed. It is already running. The physician opens a separate web dashboard before the appointment — 15 seconds.

FUNCTION 2 · ADHERENCE NUDGING

Prompting the care team when the patient is not adhering.

Passive monitoring: not needed. There is no adherence to chase. If the patient has their phone, the data is flowing. There is no survey to remind them to complete.

That second point is the one operators usually need a full call to internalize. It sounds too good. So we walk through it concretely: in a 200-patient TMS+Spravato clinic using Emobot, the operational burden on the care team is opening the dashboard before each appointment. That is the workflow. There is no task queue. There are no alerts to triage. There is no patient-chase activity. The product creates zero new work for staff because the product requires zero new work from patients.

The clinics where we have deployed have effectively the same staffing model after Emobot as before. That is the point. The product does not succeed by embedding into the care workflow — it succeeds by not needing to.

When integration is still useful — and when it is not

To be precise: integration is not impossible with Emobot. It is optional. A clinic that wants the dashboard link surfaced inside their EHR patient chart can add a simple launcher. A network that wants billing tags flowing back to their RCM can pipe the participation data. A research program that wants de-identified mood trend exports for outcomes analysis can have them.

But none of these are prerequisites for clinical value. The product works before any of them are built. That is the structural difference: in traditional digital health, integration is the thing that makes the product viable. In passive monitoring, integration is a convenience that some clinics choose and others skip.

TYPICAL DEPLOYMENT PROFILE

TRADITIONAL DIGITAL HEALTH

~12 months

IT integration sprint · API compliance · staff training curriculum

EMOBOT

~1 quarter

Physician demo in exam room · patient installs · dashboard live at next visit

For a mid-sized IP network weighing a rollout timeline, this changes the project profile materially. There is no six-month IT integration sprint. There is no vendor-side compliance review of EHR API access. There is no training curriculum for the MA population. At network scale, this collapses a typical twelve-month digital health rollout into a rolling quarter.

The underlying principle

Good clinical tools should feel lighter, not heavier, as you scale them. The test is whether adding the 200th patient takes more staff effort than adding the 10th. With active monitoring tools, the answer is yes — because the patient-chase volume scales linearly, and workflow integration was always the scaffolding that made that chase tolerable. With passive monitoring, the answer is no — because the care team does not participate in the monitoring loop at all.

The reframe for operators is: "how does this integrate into our workflow?" was the right question for the last generation. It is the wrong question for this one. The right question is "what does my staff have to do?" — and the answer is "nothing you are not already doing."

That is the difference between a digital health tool built as a workaround for patient non-adherence, and one built from the assumption that the patient should not have to do anything in the first place.

TP

Tanel Petelot

CEO & Co-founder, Emobot

Want to see it in your practice?

Book a 30-minute demo

We’ll walk through a real patient case and show you exactly how the dashboard works.

Book a Demo