Health self-monitoring through natural conversation.

My conversational-health research studies how voice and language technologies can reduce the effort required to record and understand information about chronic conditions. The central system is VoiS, a voice-activated self-monitoring approach for adults managing diabetes and hypertension. Its workflow connects conversational input, mobile access, feedback, and the possibility of sharing records with healthcare providers. The design question is not simply whether a speech interface can capture a number; it is whether the interaction fits daily self-management and produces information that people and care teams can interpret responsibly.

The VoiS publication series develops this question from system architecture to user perspectives and medical-language understanding. An early design grounds the application in self-management and behavior-change theory. A usability study with 14 participants examines perceived usefulness and ease of use. A later approach incorporates the Unified Medical Language System so the conversational agent can connect everyday expressions with medical concepts rather than depending only on rigid commands. A related survey places VoiS within broader conversational-agent architectures for chronic-condition management.

Complementary natural-language-processing work analyzes how people discuss diabetes on social media and maps lay vocabulary to UMLS concepts and semantic types. Together, these studies connect conversational interaction, medical ontologies, health informatics, and patient–provider communication. They also surface practical risks: speech recognition errors, ambiguous expressions, privacy in shared environments, and automation that sounds more certain than the underlying data. The research therefore treats understandable confirmation, user correction, and clear boundaries around advice as core system requirements.