EXPERIMENT 001 / HEALTHCARE AI

Clinical Intelligence Assistant

Can an AI assistant answer questions from synthetic patient records while showing its evidence—and admit when something isn't documented?

Status: BuildingPython · FHIR · RAG · EvaluationSynthetic records only
CLINICAL AI / MOCK INTERFACE
Patient: Synthetic Patient 001
FHIR Patient/001 · Sample only
Conditions: Hypertension · Type 2 diabetes
Recent HbA1c: 7.2%
Ask the assistant
“Summarize this patient's chronic conditions.”
Grounded example: Two chronic conditions are documented.
Sources: Condition/12 · Condition/18
PROJECT DEEP DIVE

Inside the build

An evolving case study

The problem

Patient data is spread across structured clinical resources. A useful assistant must retrieve relevant evidence, produce accurate summaries, and make its sources easy to inspect.

The goal

Build a synthetic-data prototype that answers chart questions using explicit evidence, distinguishes unknown facts from known facts, and supports reproducible evaluation.

Architecture, v0.1

Streamlit UI → FastAPI → validated retrieval tools → synthetic FHIR resources and indexed clinical context → hosted LLM → source verification → human review.

clinical-intelligence/
├── app/ # FastAPI and tools
├── ui/ # Streamlit interface
├── data/ # Synthetic FHIR fixtures
├── evals/ # Grounding and regression cases
└── tests/ # Python unit tests

What gets tested?

Factual grounding

Does every clinical claim have source evidence?

Missing information

Does the system say when a record doesn't support an answer?

Tool correctness

Does it retrieve the right patient and resource?

Target: 30 reproducible synthetic cases. Metrics and pass rates will be reported after implementation—not invented in advance.

The build journal

MILESTONE 01
FHIR patient explorer

Generate and render synthetic longitudinal records.

MILESTONE 02
Evidence-grounded RAG

Connect questions to explicit retrieved records.

MILESTONE 03
Evaluation and deployment

Test quality and publish a demonstration.

INTERACTIVE SAMPLE

Try the concept.

This is a deterministic frontend mock demonstration. It doesn't call a language model, access real patient records, or represent a clinically validated system.

Synthetic patientStatic test fixtures

Patient 001

Choose a sample question: