A deep learning-based medication behavior monitoring system.

Area of Science:

• Medical Informatics
• Artificial Intelligence
• Internet of Things

Background:

• The COVID-19 pandemic highlighted the need for remote patient monitoring systems.
• Existing wearable devices for medication behavior monitoring present challenges such as user inconvenience and measurement inconsistencies.
• There is a growing demand for intelligent systems to ensure stable patient monitoring in medical domains.

Purpose of the Study:

• To develop an improved medication behavior monitoring system using the Internet of Things (IoT) and deep learning.
• To overcome the limitations of wearable devices by avoiding sensing errors and enhancing user experience.
• To enable accurate and timely detection of patient medication status for remote healthcare.

Main Methods:

• The proposed system utilizes an IoT device for real-time image capture of patients.
• Deep learning, specifically OpenPose, is employed for human activity recognition to identify medication situations.
• The MQTT protocol facilitates reliable, periodic data transmission between the system and healthcare providers.

Main Results:

• The system accurately detects all medication behaviors.
• Notifications regarding medication status are efficiently sent to patients' mobile devices.
• Medication status information is reliably transmitted to doctors for remote treatment, improving monitoring accuracy.

Conclusions:

• The developed IoT and deep learning system effectively monitors patient medication behavior.
• The system enhances user experience by eliminating the need for inconvenient wearable devices.
• This approach facilitates accurate remote patient monitoring and timely medical intervention.