Sea Horse Optimization-Deep Neural Network: A Medication Adherence Monitoring System Based on Hand Gesture...

Area of Science:

• Biomedical Engineering
• Health Informatics
• Machine Learning Applications

Background:

• Medication adherence is crucial for treatment success, yet accurate monitoring remains a global challenge.
• Sensor technology and machine learning (ML) offer promising solutions for continuous patient adherence observation.
• Existing methods lack standardized techniques, hindering effective patient regimen management.

Purpose of the Study:

• To develop a sensor-based hand gesture recognition model for predicting medication activities.
• To create a smart sensor device integrated with ML for accurate medication intake detection.
• To enhance patient monitoring through innovative wearable technology.

Main Methods:

• A smart sensor device with tri-axial gyroscope, geometric, and accelerometer sensors was used to collect hand gesture data.
• A smartphone application transmitted sensor data to a cloud database (.csv format).
• A novel Sea Horse Optimization-Deep Neural Network (SHO-DNN) model classified hand gestures to identify medication intake.

Main Results:

• The SHO-DNN model achieved high performance metrics: 98.59% accuracy, 97.82% sensitivity, 98.69% precision, and 98.48% F1 score.
• The proposed model demonstrated superior performance compared to existing available models.
• The Sea Horse Optimization technique effectively tuned the Deep Neural Network parameters for improved classification.

Conclusions:

• The developed sensor-based hand gesture recognition model is a highly effective tool for medication adherence monitoring.
• This innovative approach provides a reliable and accurate method for tracking patient medication behavior.
• The findings support the potential of this technology in advancing remote healthcare applications.