Development Of An AI-Driven Indoor Air Quality Index Prediction System for Healthcare Environments

Poor indoor air quality (IAQ) in healthcare environments increases hospital acquired infections, exacerbates respiratory diseases, and impairs staff performance. This study develops an AI driven system that predicts the Indoor Air Quality Index (IAQI) in real time using low cost sensor data and machine learning. A dataset of 18,540 hourly records from five hospitals was collected, measuring PM2.5, PM10, CO₂, CO, TVOCs, temperature, and relative humidity. Three AI models (Random Forest, XGBoost, LSTM) were trained and compared against a traditional deterministic model. The ensemble LSTM XGBoost predictor achieved an R² of 0.94 (RMSE = 4.21) for IAQI, outperforming the deterministic model (R² = 0.68). The system provides 15 minute ahead predictions with an accuracy of 92.3%, enabling proactive ventilation control. Deployment in a 200 bed hospital reduced high IAQ events by 58% over 6 months. Ethical considerations (data privacy, model explainability) are addressed. This AI system offers a cost effective, scalable solution for continuous IAQ management in low resource healthcare settings.