AI-Powered Health Risk Prediction for Elderly Care

GeriRisk is a wearable-data health monitoring system designed for senior care. It ingests real-time metrics — heart rate, SpO₂, steps, sleep, and temperature — and uses machine-learning models to predict cardiac stress, fall likelihood, and respiratory irregularities. Results are presented through an intuitive, alert-driven dashboard so caregivers can act fast.

Features · Tech Stack · Architecture · Project Structure · Getting Started· API Reference · License

---

Features

Category	Highlights
ML Risk Prediction	Cardiac stress · Fall likelihood · Respiratory risk — each scored with High / Moderate / Low levels
CSV Upload Pipeline	Drag-and-drop wearable data upload → preprocessing → feature extraction → ML inference
Real-Time Dashboard	Sparkline charts · Activity rings · Sleep timeline & distribution panels · Risk score cards
Intelligent Alerts	Data-driven alert engine with contextual clinical messages and severity-based prioritization
Book Appointment	In-dashboard appointment scheduling with doctor availability based on selected dates
Data Persistence	Supabase-backed storage for all uploaded health records
Apple-Inspired UI	Clean, minimal design with Inter typography, smooth Framer Motion animations & glassmorphic cards

---

Tech Stack

Frontend

Technology	Purpose
Next.js 16	React framework with App Router & API routes
React 19	Component library with the React Compiler
TypeScript 5	Type-safe development
Tailwind CSS 4	Utility-first styling with CSS variables
Framer Motion	Smooth, Apple-like animations & transitions
Recharts	Data visualization (sparklines, charts)
Lucide React	Icon system
Inter Font	Typography (via next/font/google)

Backend & ML

Technology	Purpose
Next.js API Routes	/api/upload and /api/process endpoints
Python 3	ML inference runtime
scikit-learn	Pre-trained risk models (Random Forest / Logistic Regression)
NumPy & Joblib	Numerical computation & model serialization

Infrastructure

Technology	Purpose
Supabase	PostgreSQL database + authentication
PapaParse	Client-side CSV parsing

---

Architecture

---

Project Structure

GeriRisk/
├── geririsk-ai/                  # Main application
│   ├── ml/                       # Machine learning module
│   │   ├── models/               # Pre-trained model files (.pkl)
│   │   │   ├── cardiac_risk_model.pkl
│   │   │   ├── cardiac_scaler.pkl
│   │   │   ├── fall_risk_model.pkl
│   │   │   ├── fall_scaler.pkl
│   │   │   ├── respiratory_risk_model.pkl
│   │   │   └── respiratory_scaler.pkl
│   │   └── predict.py            # Inference entry point
│   ├── public/                   # Static assets & branding
│   ├── src/
│   │   ├── app/
│   │   │   ├── api/
│   │   │   │   ├── upload/       # CSV upload endpoint
│   │   │   │   └── process/      # ML processing endpoint
│   │   │   ├── dashboard/        # Patient dashboard page
│   │   │   ├── upload/           # Upload wizard page
│   │   │   ├── login/            # Authentication page
│   │   │   ├── page.tsx          # Landing page
│   │   │   ├── layout.tsx        # Root layout
│   │   │   └── globals.css       # Global styles
│   │   ├── components/
│   │   │   ├── ActivityRing.tsx   # Circular progress rings
│   │   │   ├── AlertPanel.tsx     # Risk alert notifications
│   │   │   ├── BookAppointment.tsx # Doctor appointment scheduler
│   │   │   ├── DataTable.tsx      # Tabular data display
│   │   │   ├── MetricCard.tsx     # KPI metric tiles
│   │   │   ├── RiskCard.tsx       # Risk score display cards
│   │   │   ├── SleepDistribution.tsx
│   │   │   ├── SleepTimeline.tsx
│   │   │   └── SparklineChart.tsx
│   │   └── lib/
│   │       ├── api.ts            # Client-side API helpers
│   │       ├── csvParser.ts      # CSV parsing utilities
│   │       ├── features.ts       # Feature engineering logic
│   │       ├── generateAlerts.ts # Intelligent alert generation engine
│   │       ├── preprocess.ts     # Data preprocessing pipeline
│   │       └── supabaseClient.ts # Supabase client singleton
│   ├── package.json
│   ├── tsconfig.json
│   └── next.config.ts
├── requirements.txt              # Python dependencies
├── LICENSE                       # MIT License
└── README.md

---

Getting Started

Prerequisites

Tool	Version
Node.js	≥ 18.x
npm	≥ 9.x
Python	≥ 3.10
Supabase Account	supabase.com

1. Clone the Repository

git clone https://github.com/atharvapawar9/GeriRisk.git
cd GeriRisk

2. Install Python Dependencies

pip install -r requirements.txt

3. Install Node.js Dependencies

cd geririsk-ai
npm install

4. Configure Environment Variables

Create a .env.local file inside geririsk-ai/:

NEXT_PUBLIC_SUPABASE_URL=your_supabase_project_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_anon_key

Note: You can find these values in your Supabase project dashboard under Settings → API.

5. Run the Development Server

npm run dev

The app will be available at http://localhost:3000.

---

API Reference

POST /api/upload

Upload a CSV file containing wearable health data.

Parameter	Type	Description
file	FormData	CSV file with health metrics

POST /api/process

Run ML inference on preprocessed feature data.

Request Body (JSON):

{
  "avgHeartRate": 78,
  "maxHeartRate": 120,
  "minHeartRate": 55,
  "minSpO2": 94,
  "totalSteps": 3200,
  "recordCount": 48
}

Response:

{
  "cardiacRisk": { "score": 0.312, "level": "Low" },
  "fallRisk": { "score": 0.651, "level": "Moderate" },
  "respiratoryRisk": { "score": 0.142, "level": "Low" }
}

---

License

Distributed under the MIT License. See LICENSE for more information.

---

Design Philosophy

GeriRisk's frontend follows an modern design language:

• Typography: Inter font family (closest to SF Pro available via Google Fonts)
• Color Palette: Deep brand blue #0000c9 primary with #a8bcff accents, Apple's #f5f5f7 warm gray for section backgrounds, #1d1d1f charcoal footer
• Corners & Shadows: Large 1rem border-radius with soft, diffused box-shadows
• Animations: Smooth Framer Motion entrances with custom cubic-bezier easing curves
• Glass Effects: Frosted-glass navbar with backdrop-blur and backdrop-saturate

---