Digital Therapeutics (DTx)
Validating AI that delivers treatment through software
Industry Challenge
Digital therapeutics — software-delivered, evidence-based interventions for chronic disease, mental health, and behavioural conditions — are subject to increasing regulatory scrutiny. Their AI components need both high-quality training data and clinical validation of model outputs. The data is often sensitive (mental health records, addiction treatment logs, behavioural assessments) and requires annotators with genuine clinical empathy and domain expertise.
How SCILabel Serves This Industry
Data Collection
We source anonymised behavioural health assessment records, digital CBT session logs, mood tracking data, and adherence records from DTx clinical trial and research partners. All sensitive data is handled under enhanced privacy protocols with explicit participant consent.
Data Annotation & Labeling
Our clinical workforce annotates behavioural health data with PHQ-9/GAD-7 severity classifications, CBT thought record labels, engagement and dropout risk indicators, symptom trajectory annotations, and natural language sentiment and distress labels on patient-authored text.
Data & Model Evaluation
Clinical psychologists and trained evaluators assess DTx AI model outputs for clinical safety (does the intervention recommend a safe response to suicidal ideation?), therapeutic fidelity (does it follow CBT principles correctly?), and effectiveness signal detection.
Annotation Types & Formats
- PHQ-9, GAD-7, and PCL-5 severity classification
- CBT thought record annotation: automatic thought, cognitive distortion, reframe
- Sentiment and distress labeling on patient-authored text
- Engagement and dropout risk indicator annotation
- Safety flag annotation: expressions of suicidal ideation or self-harm
Specialist Workforce Tracks
Track 1 (Medical NLP), Track 3 (RLHF), and Track 5 (AI Safety): Psychiatric Nurses, Clinical Psychologists-in-training, Public Health Professionals, Clinical Researchers.