Ongoing Research Projects
- Secondary Data Analysis Modelling BMI: Using the National Health and Nutrition Examination Survey (NHANES), we are investigating factors affecting obesity. Demographic, behavioral, and social determinants of health variables are used to predict body mass index (BMI). The NHANES is a publicly available dataset without identifiers. The data is random, representative, and covers the full age range of the US population. It includes both interview and physical examination data.
- Self-diagnostic Computer Vision based Weight Monitoring Tools: A number of studies have been proposed that use machine/ deep learning models for (a) obesity prediction, and (b) understanding the key determinants of obesity to develop intervention strategies. A recent trend is in the development of computer vision based self-diagnostic tools for weight monitoring using facial images. The aim of this project is to design computer vision-based weight monitoring tools that are tailored for younger and older cohorts. To this aim, deep-learning and explainable AI based solutions are being explored to deduce BMI from facial images. Further, methods are being developed to understand and mitigate the bias of facial analysis technology in predicting weight across gender and age-groups. Lastly, the existing models are being tailored for older adults using fusing of facial analysis with other sources of information such as waist circumference, age, ethnicity, and self-reported overall health.