About
An AI‑assisted clinical communication hub that improves patient-clinician interactions through safe triage, clear plain‑English explanations, clinical note summarisation, and a transparent rule‑based chest‑pain risk scoring system.
This project is an AI-assisted clinical communication hub designed to address well‑documented communication failures across the NHS. Research shows poor communication is a major driver of patient dissatisfaction, misunderstood instructions, and delays in care, while long‑term patients frequently struggle with unclear symptom exchange that leads to missed escalation and avoidable harm. Clinicians simultaneously face administrative and documentation overload, which continues to be highlighted as a critical pressure point in healthcare environments. Our system improves communication across all directions: patient‑to‑clinician, clinician‑to‑patient, and clinician‑to‑clinician.
The hub consists of four modules. A Triage Assistant analyses vague patient messages, surfaces hidden risk, and generates safe clarifying responses. A Clinical Notes Summariser reduces cognitive load by extracting diagnoses, medications, and timelines from long histories. A Letter and Results Explainer rewrites complex NHS letters into accessible plain English with clear next steps. Finally, a specialised Chest Pain Assessment module performs structured symptom capture and transparent scoring using a deterministic rule‑based engine. This scoring system weights red‑flag features such as radiation, breathlessness, sudden onset, duration, exertion, age, sex, and cardiac risk factors, producing Very Low to Very High risk categories with safety‑netting, without performing diagnosis.
Built as a prototype for NHS Hack Day, this system demonstrates how hybrid AI combining deterministic rule‑based logic with natural‑language generation can support safer, clearer and more efficient communication. It showcases a patient‑centred and clinician‑supportive approach that aligns with NHS safety culture by keeping clinical logic transparent, explainable, and reviewable.
Presentation Video
Website
https://preview--ai-patient-whisperer.lovable.app/
Code and Licence
Source code: https://github.com/sadiqueamin-1999/ai-healthcare-assistant