Mobile intelligence for continuous care.

My mobile-health AI research examines how phones, caregivers, practitioners, and data models can work together when care cannot depend on frequent in-person visits. The central example is mCARE, a mobile-based platform developed for longitudinal monitoring of children with autism spectrum disorder in Bangladesh. Related projects include MPredA, which applies machine learning to developmental progress, and VoiS, a voice-activated self-monitoring approach for adults managing diabetes and hypertension.

This work treats the mobile application as only one part of a larger care system. Families need questions that are understandable and practical to answer. Practitioners need longitudinal information that supports—not overwhelms—judgment. Data pipelines need consistent measures, clear missingness, privacy safeguards, and models whose limitations can be explained. A deployment also has to fit local connectivity, language, staffing, and cultural conditions. These human and organizational requirements shape the technical design from the beginning.

The mCARE research record includes end-to-end platform development, remote experience sampling, data-driven validation, machine-learning analysis of developmental milestones, and longitudinal study during the COVID-19 disruption. One milestone study analyzed data from 300 children, while a related risk-and-resilience study examined 56,290 caregiver-reported observations from 150 children in the monitored group. Those studies show the value of sustained digital contact, but they also show why a model cannot be separated from the context in which families report data and practitioners interpret it.