An Autonomous Robotic System for Multi-Modal, Non-Invasive Sensing and Early Illness Detection in Seven-Month-Old Infants

The early detection of illness in non-verbal infants, particularly at seven months of age, presents a significant challenge in pediatric care. Pre-verbal infants cannot articulate discomfort, leading to potential delays in diagnosis and treatment. This paper proposes the design and methodology for a novel, low-cost, and non-invasive monitoring system that leverages the Arduino microcontroller platform integrated with a thermal camera and passive infrared (PIR) sensors to create an early warning system for infant sickness. The system operates by continuously and unobtrusively monitoring two key physiological and behavioral correlates of illness: elevated core body temperature (fever) and alterations in sleep/wake activity patterns. The thermal camera (MLX90640) is employed to map facial temperature, identifying febrile states without physical contact. Concurrently, PIR sensors track gross motor activity and restlessness, which are often suppressed or increased during illness. Data from these sensors are processed by an Arduino Mega, which uses a rule-based algorithm to flag anomalies. If a potential sickness state is detected (e.g., sustained elevated temperature coupled with abnormal inactivity), the system triggers an alert to caregivers via a connected mobile application. This multi-modal approach aims to reduce false positives compared to single-parameter systems and provides a crucial tool for proactive parental intervention, potentially improving health outcomes for vulnerable infants.