Using wearable technology to detect prescription opioid self-administration.

Area of Science:

• Biomedical Engineering
• Machine Learning Applications
• Pain Management Technology

Background:

• Monitoring opioid use in pain patients is crucial but challenging.
• Post-dental surgery opioid self-administration requires effective oversight.

Purpose of the Study:

• To evaluate wearable sensors and machine learning for detecting opioid self-administration.
• To assess the accuracy of a machine learning model in identifying opioid use post-dental surgery.

Main Methods:

• 46 adult patients undergoing dental surgery wore Empatica E4 sensors.
• Sensors collected physiological data: accelerometer, heart rate, electrodermal activity.
• Machine learning models were trained to detect opioid self-administration.

Main Results:

• The bagged-tree machine learning model achieved 83.7% validation accuracy and 0.92 AUC.
• Model demonstrated high sensitivity (81-82%) and specificity (85-88%) in detecting opioid use.
• Opioid self-administration was identified with reasonable accuracy.

Conclusions:

• Wearable technology combined with machine learning can accurately detect opioid self-administration.
• This approach has significant potential for advancing opioid use prevention and treatment.
• Future applications in monitoring and patient safety are promising.