Real-Time Monitoring and Decision Modeling of Apis cerana cerana Swarming via Sensor-Embedded Intelligent Frames

• Intelligent beehive with multi-point temperature, humidity, and weight sensing, 5-min high-frequency monitoring. • Pre-swarming detection 7–15 min in advance via coupled temperature rise (0.8–3.15°C) and weight loss (0.02–0.22 kg, r > 0.93). • Dual-parameter early warning model with three-level decision rules, achieving 92.7% accuracy. • Low-power (< 100 mAh/day) system enabling real-time (< 3 s) cloud-based decision support. Non-invasive, continuous monitoring of internal beehive conditions can provide an effective alternative to labor-intensive, disruptive manual inspections. In this study, we developed and field-tested an intelligent frame (IF) integrating miniaturized, multi-point sensors for real-time monitoring of internal temperature, relative humidity, and weight in Apis cerana cerana colonies on Hainan Island, China. Over a 164-day deployment (September 25, 2023-March 5, 2024), the IFs transmitted readings to the cloud every 5 minutes. Six natural swarming events were recorded, with swarms occurring predominantly between 09:38 and 15:22. In the 8-15 minutes preceding swarming, characteristic pre-swarm signatures included internal temperature increases of 0.8-2.89 °C and weight losses of 0.02-0.15 kg. A rule-based decision model identified the combination of short-term rapid weight drops and temperature surges as a reliable indicator of imminent-swarm. By fusing multi-point sensing with frequent cloud reporting, the IF enables real-time behavioral detection and early intervention, offering a practical tool for precision apiculture.