Objective
Develop a reporting system that integrates multimodal patient data (EEG, questionnaires, cognitive tests, histories) and systematically compares LLM-based and rule-based approaches to generate clinically interpretable, auditable treatment recommendations.
Duration: M7-M18 (August 2026-June 2027) | Effort: 11.1 person-months
Tasks
T2.1: Data Integration Pipelines
Implement pipelines for preprocessing and integrating multimodal patient data, ensuring reproducibility and data protection compliance.
T2.2: LLM Benchmarking
Apply LLMs to process unstructured data and benchmark them against rule-based template systems, with Human-in-the-Loop validation of all treatment recommendations.
T2.3: Reporting Interface
Develop and pilot-test a reporting interface that produces clear, personalised, and auditable treatment recommendations for clinicians.
Deliverable
- D2.1 Prototype of report generator (M17, June 2027)
Milestone
- MS2 Clinician feedback integration (M17, June 2027)