Washington State University researchers will test an AI-powered chatbot and continuous glucose monitors with roughly 20 Type 2 diabetes patients in rural Washington, backed by a grant from the National Institutes of Health's AIM-AHEAD Program. The three-month study aims to show whether a modest investment in AI tools can help patients manage the daily habits that influence blood sugar, such as diet, exercise, and sleep.
How the system works
Patients will wear continuous glucose monitors that feed data to an AI-powered interface built by Clinic Chat LLC, a company specializing in machine learning and natural language processing for healthcare. The system sends messages in English and Spanish about healthy behaviors, and patients can ask questions to a chatbot trained exclusively on peer-reviewed medical literature-a focused approach to AI for healthcare that relies on curated, high-quality data rather than general-purpose models.
"Type 2 diabetes is a condition where you might see a healthcare provider every so often, and what truly matters is your daily habits, like what you eat at home or what your exercise routine is," said Anna Zamora-Kapoor, lead investigator and associate professor at WSU.
Rural health equity
The project is a joint effort between WSU, Three Rivers Family Medicine in Brewster, and Clinic Chat LLC. Zamora-Kapoor had previously worked with the rural clinic and said they were eager to continue the partnership. "All Washingtonians deserve access to the best tools we have, and our rural clinics have not been receiving the investments they need to fully leverage the promise of artificial intelligence," she said. "It's important that we're leveling the playing field, and these projects show that with a very small investment, we can improve rural health in Washington state."
Measuring what works
After the monitoring period, the research team will evaluate patient usage data, satisfaction scores, and changes in glucose levels and daily routines. The goal is to determine whether the combination of continuous monitoring and an AI assistant can produce measurable health improvements.
Zamora-Kapoor emphasized the community-driven approach: "We want to conduct research together with our communities, as true partners that collaborate and participate in every step of the research process."
Why this matters for IT and Development
For developers and IT teams working in health tech, the study offers a real-world test of how AI components-continuous data streams, natural language processing, and domain-specific training data-can be combined into a lightweight intervention. The project's constrained scope and focus on a rural population also highlight the importance of designing systems that function reliably with limited infrastructure and support multilingual users from the start. The choice to train the chatbot only on vetted medical literature, rather than a general-purpose model, is a design decision worth noting for anyone building AI tools where accuracy and safety are critical.
Your membership also unlocks:
