Objective Detection of Cigarette Smoking from Physiological Sensor Signals

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Wearable Technology

Background:

• Cigarette smoking poses significant health risks. Objective monitoring of smoking behavior is crucial for public health research and interventions.
• Existing methods for detecting smoking episodes using wearable sensors include analyzing breathing patterns, hand-to-mouth movements, and lighting events.

Purpose of the Study:

• To propose and validate a novel method for identifying cigarette smoking events by analyzing heart rate parameters.
• To differentiate smoking-induced heart rate changes from those caused by physical activity using breathing rate and motion data.

Main Methods:

• A human study involving 20 daily smokers was conducted, collecting ECG, bioimpedance, and motion data using a chest-worn sensor.
• Participants underwent laboratory sessions and ~24-hour free-living monitoring, smoking cigarettes in both settings.
• A support vector machine classifier was trained using 15 selected features to detect smoking episodes.

Main Results:

• The proposed method achieved a sensitivity of 0.87 and an F-score of 0.79 in the laboratory setting.
• Under free-living conditions, the method demonstrated a sensitivity of 0.77 and an F-score of 0.61 for detecting smoking events.
• The study successfully detected 187 out of 232 smoking events across both conditions.

Conclusions:

• Characteristic changes in heart rate parameters, when combined with breathing and motion data, can serve as a useful indicator for detecting cigarette smoking.
• The findings validate the proof-of-concept for using heart rate variability in wearable sensors for objective smoking detection.
• Further research is recommended to enhance the performance and robustness of the method for real-world applications.