aaj92/Preeclampsia-Home-Shield-AI

An offline, low-resource machine learning triage engine for early preeclampsia pre-detection in crisis-stricken communities. # 🩺 Preeclampsia Early-Detection AI: Crisis-Context Home-Shield Triage Engine An open-source, production-grade clinical stratification application leveraging machine learning to predict preeclampsia risk profiles within low-resource environments, conflict-displaced populations, and home settings. ## 🌟 Core Innovations * **Self-Healing Clinical Pipeline:** Integrates an iterative missing-data imputation array (`IterativeImputer`) to calculate missing clinical parameters safely on the fly, eliminating system crashes caused by missing or broken checkup log entry variables. * **Calibrated Recall Optimization:** Tuned specifically for public health triage frameworks, optimizing decision margins to minimize fatal false-negative errors while maintaining robust statistical classification power. * **100% Offline Database Architecture:** Leverages a localized SQLite data-logging system, allowing field workers and families to run analytical assessments and track patient profiles securely entirely without internet connectivity. * **Dual-Tier Emergency Evacuation Protocol:** Features clear, non-technical instructions for family response routines alongside professional intervention protocols for field medics. ## 🏗️ Technical Architecture Blueprint The system utilizes a multi-layered, offline-first pipeline to process clinical input safely and generate highly critical triage protocols: ```text [ Mother's Input Form ] │ ▼ [ Step 1: Symptom Scan Override ] ─── (Active Checkboxes Detected) ───► [ AUTOMATIC EMERGENCY ALERT ] │ ▲ │ (All Symptoms Clear) │ ▼ │ [ Step 2: Self-Healing Imputer ] ───► (Fills Blank Inputs / NaN Data) │ (Risk >= 28% OR │ │ Vitals Critic …