GENESIGHTS Machine Learning Genetic Predictor

### GENESIGHTS: Predicting Genetic Disorders Using Machine Learning Approaches

#### Overview
GENESIGHTS is a machine learning-based system designed to predict genetic disorders using stacked model techniques. It integrates an interactive web interface for seamless user experience and provides detailed analytical insights.

---

### Tech Stack
#### Machine Learning Backend
- **Python** (Jupyter Notebook for development)
- **Libraries:**
- `pandas`, `numpy`, `scikit-learn` → Data processing
- `seaborn`, `matplotlib` → Visualization
- `XGBoost`, `LightGBM`, `RandomForest`, `SVM`, `Logistic Regression` → Model stacking
- `TensorFlow/Keras` (optional) → Deep learning enhancements
- `Flask` or `FastAPI` → API to serve the ML model

#### Web Development
- **Frontend:** React.js / Next.js, Tailwind CSS
- **Backend:** Node.js with Express.js OR Django (if using Python for both backend & ML)
- **Database:** PostgreSQL / MongoDB
- **Authentication:** Firebase/Auth0

---

### Machine Learning Approach
#### Data Collection & Preprocessing
- Genetic disorder datasets sourced from public repositories
- Exploratory Data Analysis (EDA)
- Handling missing values, outliers, and feature encoding

#### Stacked Model Architecture
A stacked model approach combines multiple models for better predictions.
- **Base Models:** Decision Trees, Random Forest, XGBoost, SVM, and k-NN
- **Meta Learner:** Logistic Regression or Neural Networks
- Base models make predictions, and the meta learner combines their outputs for final classification.

---

### Web Application Features
#### User Authentication
- Secure login/signup for doctors & researchers
- Role-based access control

#### Genetic Disorder Prediction System
- Upload patient data (CSV or manual entry)
- Real-time prediction using trained ML model (API call to backend)
- Result visualization (charts, risk factors, probability score)

#### Reports & Analytics
- Generate PDF reports for users
- Track patient history and trends over time
- Interactive dashboard for doctors

---

### Deployment
- **ML Model Deployment:** Flask API on Google Cloud Run / AWS Lambda
- **Web Application Hosting:** Vercel / Netlify (Frontend) + Heroku / AWS EC2 (Backend)
- **Database:** Cloud Firestore / AWS RDS

---

### Project Impact
✔️ Accurate genetic disorder risk prediction using stacked models
✔️ User-friendly UI for doctors and researchers
✔️ Secure, cloud-based system for real-world use

---

### Required Inputs for Prediction
#### Blood Test Parameters:
- White Blood Cell Count (thousand per microliter)
- Blood Cell Count (mcL)

#### Demographic Information:
- Patient Age
- Father's Age
- Mother's Age

#### Pregnancy & Birth History:
- Number of Previous Abortions
- History of Anomalies in Previous Pregnancies
- Assisted Conception (IVF/ART)
- Birth Asphyxia
- Autopsy Shows Birth Defect

#### Symptoms & Clinical Features:
- Symptoms 4 & 5
- Heart Rate (rates/min)
- Respiratory Rate (breaths/min)

#### Other Medical Factors:
- Blood Test Result (categorical: 0, 1, 2, 3)
- Folic Acid Intake (Peri-conceptional)
- Follow-up Status

---

### Web Interface Workflow
1️⃣ User enters details manually or uploads a CSV file.
2️⃣ Backend sends the data to the trained ML model via API.
3️⃣ Model predicts the probability score of a genetic disorder.
4️⃣ Results displayed with risk factors & visualization.
5️⃣ Option to download report as PDF for reference.

---

### GENESIGHTS Web Interface Blueprint
#### UI Layout (React + Tailwind CSS)
- **Home Page:**
- Welcome message + brief project description
- "Get Started" button leading to prediction form
- **User Authentication Page (Optional):**
- Login/Signup for doctors/researchers
- Secure access using Firebase/Auth0
- **Genetic Disorder Prediction Form (Main Feature):**
- Patient Information Section (Name, Age, Gender)
- Medical History Section (Abortions, IVF, Birth Anomalies, etc.)
- Blood & Symptom Details Section (WBC Count, Blood Test, Symptoms)
- Upload CSV Option for bulk predictions
- Submit Button → Sends data to ML Model via API
- **Prediction Result Page:**
- Displays Genetic Disorder Prediction (Yes/No) with Probability
- Risk factors & highlighted values
- Data visualization (Charts/Graphs)
- Download Report (PDF) button
- **Dashboard for Doctors:**
- View previous patient predictions
- Analytics on disorder trends

---

### GENESIGHTS: End-to-End Project Roadmap ?
#### Phase 1: Data Analysis & Preprocessing (EDA & Feature Engineering)
✔ Exploratory Data Analysis (EDA)
✔ Feature Engineering & Data Transformation
✔ Splitting Data for Training & Testing
? Output: Clean dataset ready for model training ✅

#### Phase 2: Model Building (Stacked Model Approach)
✔ Base Models: RandomForest, XGBoost, LightGBM, SVM, k-NN
✔ Meta Learner: Logistic Regression / Neural Networks
✔ Hyperparameter Tuning & Model Optimization
✔ Evaluate & Improve Model Performance
? Output: Trained ML Model with 90-95% accuracy ✅

#### Phase 3: Backend API Development (Flask / FastAPI)
✔ Build API to Serve the Model
✔ Secure & Optimize API
? Output: Working API to handle predictions ✅

#### Phase 4: Frontend Development (React.js / Next.js + Tailwind CSS)
✔ User Authentication
✔ UI Design
✔ Interactive Dashboard
? Output: Functional UI for genetic disorder predictions ✅

#### Phase 5: Deployment (Full Stack Integration)
✔ ML Model Deployment
✔ Frontend Deployment
✔ Database Setup
? Output: Live & accessible GENESIGHTS Web App ✅

#### Phase 6: Final Testing & Improvements
✔ Test entire pipeline (API + UI)
✔ Fix UI/UX bugs
✔ Enhance model performance if needed
? Final Output: Fully functional & deployed project ✅

---

### Free Tech Stack Options
✔ Frontend → React.js + Tailwind CSS (Vercel)
✔ Backend → FastAPI / Node.js (Render/Railway)
✔ Database → PostgreSQL (Neon.tech) / MongoDB (MongoDB Atlas)
✔ Authentication → Firebase / Auth0
✔ ML Model Hosting → Google Cloud Run