Detection of smoking events from confounding activities of daily living

Although smoking prevalence is declining in many countries, smoking related health problems still leads the preventable causes of death in the world. Several smoking intervention mechanisms have been introduced to help smoking cessation such as counselling program, motivational interview and pharmacotherapy. However, these methods lack providing real time personalized intervention messages to the smoking addicted users. The challenge is to develop an automated smoking behavior detection. We address this challenge by proposing a non-invasive sensor based automated framework for smoking behavior detection. We used a wristband based accelerometer and gyroscope sensors to detect smoking activities, differentiating with the closely confounding activities. We extract several features using learning algorithms and the empirical results with our participants show good accuracy in detecting the smoking activity in terms of precision, recall, and F1-score.