Skip to content

prd.md

Latest commit

 

History

History
111 lines (80 loc) · 2.92 KB

prd.md

File metadata and controls

111 lines (80 loc) · 2.92 KB

Product Requirements Document

Medical Case Learning App (AI-Powered Clinical Reasoning Platform)

1. Overview

A mobile/web application that helps medical students and junior doctors develop clinical reasoning skills through AI-generated spaced repetition case studies, real-time coaching, and performance feedback.

2. Goals

  • Help users practice clinical reasoning through realistic, AI-generated patient cases
  • Provide adaptive difficulty using spaced repetition
  • Give actionable performance feedback after each case
  • Support both self-directed learning and structured review

3. User Flow

3.1 Authentication

  • Phone number input
  • OTP verification (dummy/fake OTP for MVP)
  • On success → redirect to Home Dashboard

3.2 Home Dashboard

Users can:

  • Start a new case (Spaced Repetition Case Generator)
  • Review due cases (cases scheduled for review)
  • View performance & insights

3.3 Model Selection

Before starting a session, user selects a mode:

Mode Description
Clinical Reasoning Coach Step-by-step guidance through a case with feedback
Spaced Repetition Generator Creates cases/questions for repeated practice over time

3.4 Case Setup

User inputs:

  • Topic (optional) — e.g., "heart failure"
  • Difficulty — Easy / Medium / Hard

Case is generated via Gemini API (generate case function).

3.5 Case Interaction Screen (Core UX)

Three-panel layout:

Panel Content
Left Case progression: Chief Complaint → History → Exam → Labs
Center Chat interface with AI
Right Notes / Differential diagnosis list

3.6 Performance & Insight (Post-Case)

  • AI gives Socratic feedback on user's reasoning
  • LLM reviews user input vs. ideal response
  • Highlights what the user did well and what they missed

4. Features

MVP Features

  • Phone auth (dummy OTP)
  • Case generation via Gemini (topic + difficulty)
  • Chat-based case interaction
  • Three-panel case UI
  • Post-case performance feedback
  • Basic spaced repetition scheduling

Post-MVP

  • Real OTP (Twilio / Firebase)
  • User performance history & analytics
  • Specialty-specific case libraries
  • Peer comparison / leaderboard

5. Tech Stack

Layer Technology
Frontend React / Next.js
Backend FastAPI (Python)
Auth uvicorn + FastAPI, dummy OTP for MVP
AI / LLM Gemini API (case generation + feedback)
Database PostgreSQL or Supabase
Spaced Repetition SM-2 algorithm or similar

6. Non-Functional Requirements

  • Mobile-responsive design
  • Response latency < 3s for case generation
  • Support 100+ concurrent users (MVP target)
  • All medical content clearly labeled as educational, not clinical guidance

7. Out of Scope (MVP)

  • Real clinical data / EHR integration
  • Actual OTP SMS
  • Payment / subscription
  • Offline mode