An IoT-Enabled Multi-Sensor Smart IV Infusion Monitoring System with Edge–Cloud Integration

Conventional gravity-fed intravenous (IV) infusion relies largely on visual inspection, and abnormal events such as incorrect drip rate, depletion, and blood backflow may occur between nursing rounds. This study develops and evaluates an IoT-enabled multi-sensor bedside monitoring system with edge–cloud integration. A dual-core ESP32 edge node integrates SD01 optical drip-rate sensing, XKC-Y25 non-contact depletion detection, TCS34725 optical blood-backflow detection, MAX30105 heart-rate and SpO2 monitoring, and PN532 NFC/RFID staff authentication. FreeRTOS task partitioning separates high-rate physiological processing from communication tasks. Deterministic alarms are generated locally on the edge node, independent of cloud availability. Processed states are synchronized to a Supabase cloud backend every 1 s and displayed on a multi-bed dashboard with privacy-aware public and NFC-authenticated clinical views. Laboratory tests showed an aggregate IV flow-rate mean absolute error below 3.5% and a backflow alarm response time below 1 s; the sensitivity, specificity, and false-positive rate of the backflow detector were not quantified. In a preliminary comparison with a commercial pulse oximeter (five paired readings), the mean differences were 4.2 bpm for heart rate and 0.6 percentage points for SpO2, with mean absolute differences of 5.0 bpm and 1.8 percentage points, respectively. These results demonstrate the feasibility of the integrated architecture at laboratory level but do not establish general measurement accuracy, reliable backflow detection, or clinical accuracy. A knowledge-guided decision-support layer is outlined as future work.