Dr. Younghoon Kim gave a presentation at the Eastern North American Region International Biometric Society Spring 2024 Meeting: ENAR – A Home for Every Biostatistician on March 11th in Baltimore, Maryland. His presentation was titled 'Detecting Emotionally Stressful Periods from Passive Sensing Data via Mobile Devices.'
Here is the abstract from his presentation: It is crucial for psychotherapy to prevent the negative effects of emotional stress in middle-aged and older adults with chronic pain and depression by detecting the early onset of stressful periods. This study proposes a data-driven emotional stress detection algorithm. The algorithm identifies participants' onset of emotionally stressful periods using passive sensing data related to physical activities, such as step counts and activity duration. The algorithm consists of three separate parts. Firstly, it estimates the time points of changes in the distributions of passive sensing data. Then, the algorithm validates whether the detected changes are statistically associated with changes in the response variable, constructed from self-reported stress levels and stress-related scores. Finally, the algorithm trains a classifier to predict whether the patient is in a stress period at future time points using statistical and physical features computed from each segmented passive sensing data. The algorithm is applied to the ALACRITY Phase 1 data, and its prediction performance and the effectiveness of each step are evaluated using various metrics.
