Abbykairu65/CONTRACEPTIVE-RECOMMENDATION-KIOSK-FOR-COMFORT-AWARE-FAMILY-PLANNING

A bilingual self-service contraceptive decision support kiosk that combines WHO MEC safety rules with a machine learning based comfort ranking model (Random Forest), deployed on a Raspberry Pi touchscreen system for low-resource clinical environments. # CONTRACEPTIVE-RECOMMENDATION-KIOSK-FOR-COMFORT-AWARE-FAMILY-PLANNING A bilingual self-service contraceptive decision support kiosk that combines WHO MEC safety rules with a machine learning based comfort ranking model (Random Forest), deployed on a Raspberry Pi touchscreen system for low-resource clinical environments. # Contraceptive Decision Support Kiosk A bilingual, offline-capable self-service kiosk that supports contraceptive method selection using a two-tier decision system: 1. A deterministic WHO MEC safety gate for clinical eligibility filtering 2. A machine learning based comfort ranking model for personalised recommendation The system is designed for low-resource clinical settings and runs on a Raspberry Pi with a touchscreen interface. --- ## 🧠 Problem Statement Many women discontinue contraceptive methods not due to access or safety issues, but due to poor comfort matching and side-effect mismatch. Existing clinical workflows prioritise safety but do not systematically optimise for individual comfort preferences in a structured way. --- ## 🎯 Project Objective To develop a structured, clinically grounded decision support system that: - Ensures WHO MEC safety compliance - Personalises contraceptive recommendations based on user comfort profiles - Operates as a self-service kiosk without requiring a clinician or internet access --- ## ⚙️ System Architecture The system operates in two stages: ### 1. Tier 1: WHO MEC Safety Gate - Rule-based filtering system - Eliminates unsafe contraceptive methods based on clinical contraindications - Fully aligned with WHO Medical Eligibility Criteria ### 2. Tier 2: Machine Learning Ranking Model - Random Forest classifier - Ranks only safe methods from Tier 1 - Uses 19 user features including: - Age, BMI, blood pressure - Medical history - Reproductive goals - Comfort concern profile ### Output - Primary recommendation - Secondary recommendation - Displayed via touchscreen interface (English and Kiswahi …