An AIoT-Based Home Monitoring System for Arteriovenous Fistula Surveillance in Hemodialysis Patients: Development, Evaluation, and Clinical Potential.

Background: Arteriovenous fistulas are critical for maintaining effective blood circulation during hemodialysis. Undetected fistula dysfunction can lead to severe complications or death. Existing monitoring approaches rely heavily on hospital-based assessment, creating challenges for early intervention in home care settings.

Methods: This study developed an AIoT-based home care device that enables patients to monitor their fistula function at home. The device captures vascular sound signals through a microphone and analyses them using a convolutional neural network model trained on 245 labelled audio samples. The device provides real-time alerts using LED and audio indicators and transmits data to the hospital information system via LoRa wireless communication. Additionally, user feedback was gathered through qualitative interviews based on the Technology Acceptance Model (TAM).

Results: The neural network achieved an F1-score of 1.00 for detecting blockages (n=33), 0.93 for slight blockages (n=54), and 1.00 for normal conditions (n=158). Wireless signal transmission was reliable over distances ranging from 6.17 to 8.68 km with RSSI values between -107.2 dBm and -97.2 dBm. TAM-based interviews showed that patients found the device easy to operate and were willing to recommend its use to others.

Conclusion: The proposed system offers a reliable, non-invasive, and user-friendly solution for early detection of fistula dysfunction. It enhances patient safety and facilitates real-time communication with medical institutions, making it a promising tool for remote hemodialysis management.