Case studies · Healthcare

Mental health data platform: 30× more patient data, every day

Daily patient data instead of memories from a monthly visit, in a secure and anonymised data platform.

Client: Digital mental health company · Industry: Healthcare

from one visit a month to daily tracking

daily active compliance

from 30-day recall to real-time signals

The challenge

Psychiatrists rely on what patients remember from the weeks between visits, which is often biased.

The goal was to capture daily, reliable data on sleep, anxiety and mood, so treatment can be based on facts and later analysed with machine learning.

How we built it

Phase 01

Foundation and data generation

Goal

Build a secure pipeline that turns how patients feel into longitudinal clinical data.

Define the clinical gap
Psychiatrists rely on episodic patient memory, which is often biased. The strategy: capture daily metrics on sleep, anxiety and mood to build a ground-truth dataset.
Build the data collection first
No data existed, so we built a mobile app with scientifically validated questionnaires (PHQ-9, GAD-7) and a secure, anonymised data lake that joins the daily logs with clinical records such as brain CT scans and prescription history.
Prove clinical value
A therapist dashboard maps patient-reported symptoms over time. The digital logs were validated against standard clinical assessments, producing the first verified patient trajectory graphs.

Quality gate: security audit and adherence review, with more than 80% daily active logging required.

Phase 02

Engineering and clinical intelligence

Goal

Analyse treatment effects across app, clinical and prescription data, with the doctor in control.

Keep the data flowing
The app tracks patient engagement, watches for survey fatigue and adapts question frequency, so the quality of the data stays high.
Multi-modal analysis
A machine learning model correlates treatment history with CT scan features and patient-reported sleep and mood. It was tested so that it highlights correlations without making autonomous diagnostic claims.
In use in hospitals
The system is deployed in mental health hospitals in Warsaw. The therapist panel shows data-based insights that help doctors adjust treatment, while the decision stays with them.

Outcome

Daily, validated patient data and insights that support, not replace, the doctor's decision.

Results

87% of patients logging daily, enough for valid ML datasets.
Daily tracking instead of one visit a month.
Deployed in mental health hospitals in Warsaw.

Start with

Fixed scope, fixed price

AI readiness assessment

Which AI use cases pay off for you, and is your data ready for them? A ranked shortlist, an architecture and a roadmap.

From

€6,000net

Duration: 3 weeks

Who leads it: A senior data architect together with an AI engineer who has deployed models in production. You talk to the people who would build the solution.

Order the assessment

Services:Data foundation for AI Data & AI teams