DILIGENT Clinical Copilot

1. Project Overview

DILIGENT Clinical Copilot supports clinicians during Drug-Induced Liver Injury (DILI) evaluations with a FastAPI backend and an Angular + TypeScript frontend. It collects anamnesis, medications, and lab values, then coordinates clinical analysis with optional RAG support and session persistence for review.

Work in Progress: This project is under active development and may contain incomplete features or defects.

Conceptual flow from case intake through guided DILI analysis, session recording, and review.

2. Quick Start

2.1 Windows (Recommended)

Run:

start_on_windows.bat

The launcher prepares local runtimes/dependencies and starts backend plus frontend.

2.2 macOS / Linux (Manual)

Prerequisites:

• Python 3.14+
• Node.js 18+ and npm

Backend:

cd app/server
python -m pip install -e ".[test]"
uvicorn app:app --host 127.0.0.1 --port 8000

Frontend:

cd app/client
npm install
npm run build
npm run preview -- --host 127.0.0.1 --port 7861

3. Runtime Profiles

DILIGENT is configuration-first and uses one active runtime file: settings/.env.

Switch to local profile:

copy /Y settings\.env.local.example settings\.env

Switch to local Tauri profile:

copy /Y settings\.env.local.tauri.example settings\.env

See assets/docs/RUNTIME_MODES.md for full runtime and packaging details.

4. Using the Application

Typical workflow:

1. Enter anamnesis, exam notes, medications, and lab values.
2. Choose model/provider settings and optionally enable RAG/web search.
3. Run analysis and review the generated report.
4. Use Data Inspection to explore current knowledge base.
5. Explore past sessions to modify and revise them.

Detailed user journeys and feature guidance are documented in assets/docs/USER_MANUAL.md.

Screenshots

Dashboard / Report Output

Analysis dashboard focused on the report output area and execution controls.

Sessions overview

Explore past sessions and improve DILI assessment iteratively.

Model Configuration (Settings)

Runtime source selection and current model configuration summary.

Data Inspection

Session inspection table with status, timing, and record actions.

5. Desktop Packaging (Tauri)

Build Windows desktop artifacts:

release\tauri\build_with_tauri.bat

Generated outputs:

• release/windows/installers
• release/windows/portable

6. Setup and Maintenance

Run:

setup_and_maintenance.bat

Use this script for offline maintenance operations (for example DB initialization and cleanup tasks).

6.1 Regression Validation Shortcuts

From repository root:

app\tests\run_tests.bat modelconfig
app\tests\run_tests.bat modelconfigfull

• modelconfig: validated regression slice (model-config unit + focused model-config/app-flow e2e checks, including conflict-feedback handling)
• modelconfigfull: model-config unit + full test_app_flow.py + test_model_config_api.py
  • If uv --with pytest-playwright cannot access package indexes on first use, run the PowerShell runner directly after cache warmup.

Equivalent PowerShell runners:

.\app\tests\run_model_config_regression.ps1
.\app\tests\run_model_config_full_regression.ps1

7. Database and Ollama Requirements

• Database schemas are not upgraded in place across this cleanup; recreate the schema (or local SQLite DB file) when upgrading.
• Runtime startup does not perform SQLite schema salvage/deletion.
• Ollama must support the chat-capable /api/chat API; /api/generate fallback behavior has been removed.

8. Documentation Map

• assets/docs/USER_MANUAL.md: end-user operation, journeys, key commands.
• assets/docs/ARCHITECTURE.md: system boundaries and data flow.
• assets/docs/BACKGROUND_JOBS.md: job lifecycle and semantics.
• assets/docs/RUNTIME_MODES.md: runtime profiles and packaging.
• assets/docs/ERROR_HANDLING.md: backend/frontend error strategy.
• assets/docs/UI_STANDARDS.md: frontend design standards.

9. License

Non-commercial use is covered by the Polyform Noncommercial License 1.0.0; commercial licensing is available separately. See LICENSE.