AI / Healthcare
AI-Powered Mental Health Risk Assessment
A screening-support model that helps identify elevated mental-health risk indicators from structured and text-based inputs.
Overview
A research-driven project exploring how machine learning can support early identification of mental-health risk, intended as a decision-support aid alongside — never a replacement for — clinical judgement.
Problem
Mental-health risk factors are often under-identified until a crisis point, particularly where access to screening professionals is limited. Existing screening tools are largely manual, static, and hard to scale.
Approach
We worked with structured questionnaire data and free-text responses, applying feature engineering and NLP techniques to model risk indicators. The system was designed with clear thresholds, explainability, and a strong emphasis on responsible use — flagging risk for human review rather than issuing automated diagnoses.
Architecture
- 1Data ingestion & validation layer for questionnaire and text inputs
- 2Feature engineering & NLP preprocessing pipeline
- 3Model training & evaluation pipeline with cross-validation
- 4FastAPI inference service behind an authenticated internal API
- 5Human-in-the-loop review dashboard for flagged cases
Technology
Results
- Model accuracy
- Indicative — see note
- Inference latency
- Indicative — see note
- Dataset size
- Indicative — see note
This case study uses indicative placeholders. Once we publish verified evaluation results, we will replace these figures — we don't publish numbers we haven't validated.
Have a similar problem?
Talk to our team — we'll tell you honestly how close this is to what you need.