Megha Thukral
I am a PhD student at the School of Interactive Computing, Georgia Tech. My research focuses on generalizable representation learning for health sensing, building machine learning models for sensor data and clinical structured text that generalize across users, devices, home layouts, and clinical settings. My research tackles distribution shift in health sensing under varying levels of data and label availability: adapting wearable HAR models with limited supervision, transferring across heterogeneous smart-home environments , and learning generalizable representations with self-supervised pretraining for physiological signals and for structured clinical records. I previously interned at the Samsung Research America Digital Health Lab (Biomarkers team), and at Bloomberg AI Group. I am thankful to be advised by Prof. Thomas Ploetz, and grateful to be funded by Optum AI and NSF AI-CARING.
