MediPulse
Documentation Sign Up | Login

Hospital Recommendation System

Builds an algorithm using machine learning to recommend the best hospital based on user selections—medical issue, priority (e.g., quality, safety, or experience), and acceptable travel distance. Combines a Flask backend (serving hospital data from a CSV) with a user-facing frontend (HTML, JavaScript, and Leaflet for maps).

🎯 System Overview

Hospital Recommendation Flow

1. User Location & Medical Issue

(Frontend)
  • The user can choose a location (has anonymous options like San Diego Zoo or Petco Park in case the user isn't comfortable sharing).
  • Select a specific medical issue from a dropdown.

2. Priority Selection

(Frontend)
  • The frontend offers a choice among three key priorities:
    • Quality
    • Experience
    • Safety
  • A distance slider lets the user specify a maximum travel range.

3. Request Processing

(Frontend)
  • On form submission, the frontend sends a POST request to the Flask endpoint with the user's location, distance, medical issue, and priority.
  • Shows loading/error messages during processing.

4. Data Loading & Distance Calculation

(Backend)
  • Reads from a pre-cleaned CSV (`hospitaldatamodified.csv`) containing:
    • `hospital`, `performance measure` (medical issue)
    • `latitude`, `longitude`
    • `# of cases`, `# of adverse events`, `risk-adjusted rate`
    • (Optional) `quality` or similar metrics
  • Uses Haversine formula to compute distance from user's chosen lat/lon.
  • Filters out hospitals beyond the user's max travel distance.

5. ML Algorithm & Prioritization Logic

(Backend)
  • Enhanced Priority Weighting (RandomForest approach)
    • Encodes disease/priority, scales numeric features, and trains a random forest.
    • Increases the multiplier for the priority feature so that "What's most important" strongly affects which hospital is chosen.
  • Custom Scoring (Direct Formula)
    • For safety: fewer adverse events is better
    • For experience: higher # of cases
    • For quality: "better than expected" > "as expected" > "worse"
    • Incorporates distance as a penalty so closer facilities rank higher when priorities tie

6. Results Display

(Frontend)
  • Receives a recommended hospital—or a list of top hospitals—and displays them:
    • Basic hospital info (name, distance)
    • A Leaflet map with markers for each recommended facility and the user's location
  • Utilizes TailwindCSS for a responsive layout and user-friendly design.
  • Shows clear results section once the recommendation is computed.

🔧 Backend Dependencies

Here are the core Python libraries used in the backend to process the hospital CSV data and serve recommendations:

Core Libraries:

  • Pandas - For reading CSV files, cleaning and transforming the data, and handling imputation of missing values.
  • Scikit-learn - Provides the RandomForestClassifier (if using the machine-learning approach) and tools like `StandardScaler` for feature scaling.
  • Joblib - Used to serialize (save/load) the trained model and its associated encoders or scalers.
  • Math - For the Haversine formula to calculate distances between user and hospital coordinates.
  • Flask (and optionally Flask-RESTful) - To create the API endpoint that receives user input (disease, priority, distance, lat/lon) and returns a recommended hospital.

🔧 Key

  • 🟥 Frontend: User interface and interaction
  • 🟦 Backend: Data processing and ML algorithms

📡 API Endpoint

A POST endpoint `/api/predict` that accepts JSON:

````json { "disease": "Acute Stroke", "priority": "safety", "distance": 10, "user_lat": 32.71, "user_lon": -117.16 } `````