🌸 PCOS Predictor

A simple machine learning app to estimate the risk of Polycystic Ovarian Syndrome (PCOS)

---

About the Project

Polycystic Ovarian Syndrome (PCOS) is a common hormonal disorder affecting many women, and early diagnosis can significantly improve health outcomes. This project is my attempt to build a basic ML-powered web app that predicts PCOS risk based on patient health metrics.

You enter a few clinical/lifestyle parameters, and the model gives you a prediction along with a risk score—Low, Medium, or High.

Built with Python, Flask, and scikit-learn. Trained on real-world health data from Kaggle.

---

Dataset

• 📍 Source: Polycystic Ovary Syndrome (PCOS) | Kaggle
• 📄 File used: PCOS_data_without_infertility.xlsx → Full_new sheet
• 🧠 Features: Age, weight, height, BMI, hormone levels, cycle info, ultrasound results, and more
• 🎯 Target variable: PCOS (Y/N)

Some unnecessary columns (IDs, unnamed ones) are dropped. Missing values are handled, and BMI is calculated if not available.

---

How It Works

• The app loads the dataset and trains a Random Forest model on launch.

• User input is collected from a form and preprocessed.

• The model predicts:

  • 0 = No PCOS likely
  • 1 = Likely PCOS

• Based on prediction probability, it also returns a risk level:

  if prob > 0.7: risk = 'High'
  elif prob > 0.4: risk = 'Medium'
  else: risk = 'Low'

---

How to Run

1. Clone the repo

   git clone https://github.com/meoyushi/pcos_predictor
   cd pcos_predictor

2. Install dependencies

   pip install flask pandas scikit-learn numpy openpyxl

3. Start the app

   python app.py

4. Open your browser

   http://127.0.0.1:5000
   

---

App Endpoints

Route	Method	Purpose
/	GET	Main page with input form
/predict	POST	Makes prediction and returns results
/analysis	GET	Placeholder for future visualizations

---

Limitations

• Model retrains every time the server starts
• UI is basic and not yet styled for ease of use

---

About Me

Helloo, I’m Aayushi Thakre, an undergrad at IIITDM Jabalpur with a passion for applying tech to solve meaningful problems—especially in healthcare. This project was a mix of curiosity, learning, and the desire to know more about THIS PARTICULAR condition.

---