Publication

Abstract : Medication non-adherence and inadequate real-time health monitoring remain critical challenges in elderly and chronic disease care. This paper presents MediTrack, a smart pill box solution that integrates internet-of-things (IoT) based automated pill detection, vital sign monitoring, and predictive health analytics. The system employs an ESP32 microcontroller, MAX30102 pulse oximeter, and DS18B20 temperature sensor to capture real-time health vitals including SpO2, heart rate, and body temperature. A Flutter mobile application communicates with the back end via the message queuing telemetry transport (MQTT) protocol and the MongoDB Atlas cloud database, enabling medication scheduling, automated reminders, and visualization of health data. Machine learning algorithms including logistic regression, random forest, and gated recurrent unit (GRU) neural networks analyze sensor data and user behavior to detect health anomalies and predict medication adherence patterns. Infrared (IR) sensors trigger light-emitting diode (LED) indicators in respective pill slots during medication times. Care-givers receive real-time alerts and weekly reports for remote monitoring. MediTrack achieves 98.5% pill detection accuracy, 88.90% adherence prediction accuracy, and demonstrates 25% reduction in missed doses, providing an integrated platform for medication adherence and proactive health management.