🏥 Post-Partum Health Detection

Data-Driven Early Identification of Post-Partum Health Concerns — Machine Learning for Maternal Mental Health Screening

📝 Notes

Postpartum depression (PPD) is a serious mental health condition affecting many women after childbirth, often going unnoticed due to lack of timely screening. This project leverages machine learning to provide an accessible, data-driven screening tool that can assist healthcare professionals in early detection and intervention.

📄 Project Description

Post-Partum Health Detection is a machine learning-based binary classification system that detects postpartum depression risk based on structured questionnaire responses. The system processes patient responses through a comprehensive data pipeline to deliver rapid screening results.

📋 Input: Standardized questionnaire responses covering emotional, behavioral, and physical health indicators.
🧹 Preprocessing: Text standardization, missing value imputation, punctuation removal, and class balancing using SMOTE (Synthetic Minority Over-sampling Technique).
🔤 Feature Engineering: One-Hot Encoding for nominal variables, Label Encoding for ordinal variables, and TF-IDF vectorization for free-text fields.
🤖 Models Evaluated: Logistic Regression, SVM, Decision Tree, Random Forest, and Naive Bayes — comprehensively benchmarked on accuracy, precision, recall, and F1 score.
🏆 Best Model: Logistic Regression achieving 99.33% accuracy and 99.5% F1 score.
✅ Output: Binary classification — Yes/No indicating likelihood of postpartum depression, enabling early intervention.

🛠️ Languages & Tools Used

Python
Scikit-learn
Pandas, NumPy
SMOTE (Imbalanced-learn)
TF-IDF Vectorization
Jupyter Notebook

📊 Dataset

Initially sourced from Kaggle and refined by adding additional relevant health-indicator columns. Preprocessing includes class balancing via SMOTE and multi-encoding for mixed data types.

👨‍💻 Project Members

Rhithika M Pradeep
Sreelakshmi P
Harinandana K Biju
Gouri S

👨‍🏫 Mentors

Pankaj Kumar G — Asst. Professor, Dept. of CSE
Hansa J Thatil

📈 Results

Model	Accuracy	F1 Score
Logistic Regression	99.33%	99.5%
Random Forest	98.7%	98.9%
SVM	97.5%	97.8%
Decision Tree	95.2%	95.5%
Naive Bayes	91.8%	92.1%

🌍 Impact

This project demonstrates the potential of machine learning as a supportive tool in mental health care, aiding healthcare professionals in the early detection and intervention of postpartum depression — potentially saving lives through timely screening.