Find care now
If you are experiencing a medical emergency, please call 911 or seek care at an emergency room.
By Michelle Fischmann Magee, MD and Carine Nassar, RD, CDCES
Preliminary research indicates that patients with type 2 diabetes-built confidence and lowered hemoglobin A1C levels while participating in a novel chat-based diabetes education program.
Nearly 1 in 10 people in the U.S. have type 2 diabetes, in which the body does not make enough or is resistant to the hormone insulin. To successfully manage their diabetes, patients must learn and gain confidence in specific self-care routines.
Persons with Diabetes (PWD) can best learn and acquire these skills through consultation with a Certified Diabetes Care and Education Specialist (CDCES). But according to the U.S. Department of Health and Human Services, only half of PWD receive diabetes self-management training. Barriers to receiving education include lack of awareness of the benefits of these services – both from the patient and the provider perspective, poor referral systems, a scarcity of diabetes care and education specialists, and financial concerns including insurance reimbursement.
To expand the reach of diabetes education and support to more PWD and, in partnership with Conversa Health and the MedStar Diabetes Institute, our researchers have developed a digital chat algorithm, or “chatbot,” to help participants manage type 2 diabetes. The system delivers education and support to participants on their schedules, without needing office visits or insurance forms.
Our diabetes education chatbot, can automate the distribution of education messages and materials to patients from any internet-capable device while simultaneously identifying and elevating participants who need to see a doctor.
The preliminary data from pilot studies and anecdotal responses are encouraging and exciting. Early findings suggest that patients in our pilot chat program lowered their hemoglobin A1C and built self-care confidence through access to educational materials and provider support.
How the chatbot works: Human-supported Artificial Intelligence (AI).
Chatbots are a standard customer support feature many people encounter in daily internet use. These computer programs leverage artificial intelligence to interpret a user’s answers to survey questions and automate relevant responses, simulating human conversation.
In our pilot program from November 2020 to April 2022, persons with type 2 diabetes and hemoglobin A1C levels between 8 and 8.9 were eligible for the program through four clinical sites in the mid-Atlantic region. These were persons whose diabetes was not well managed, but not so severely as to make them candidates for our Diabetes Boot Camp, which provides patients with hands-on diabetes management training with providers in the medical center.
The average age of participants was 58 and most of the patients (65%) identified as African American. They selected text messaging or email delivery and their preferred day of the week to receive chats.
During the sessions, participants answered questions to determine how they managed their diabetes. Questions covered blood sugar measurements, medication use, and more. In response to their answers, the MedStar Diabetes Institute (MDI) prepared and provided patients with links to videos and other educational materials prepared by MDI.
When participants submitted answers demonstrating issues with diabetes management, such as high or low blood glucose in need of intervention, the chat system produced a “red flag” on a dashboard. Within 24 hours, a MedStar Diabetes Institute CDCES called the participant to offer support and education. The CDCES also contacted their provider to initiate follow-up care.
This system allowed MedStar care providers to intercede before negative consequences, such as a severe drop in blood sugar. About 40% of the red flags in the pilot program were generated for low blood sugar, which can lead to seizures and death. Without the chatbot engagement, the patient might not have sought or received prompt interventional care.
Results: happy patients, lower A1C.
When we deployed the chatbot, our primary goal was to get more educational materials into more people’s hands.
By February of 2022, we could establish baseline data and see a trend in health improvement in hemoglobin A1C, an essential quality indicator of diabetes management. High A1C levels indicate a greater risk of complications including diabetic retinopathy, heart disease, kidney failure, and more. Improvements in A1C are associated with reduced risk for these diabetes complications over time.
During this period, persons with type 2 diabetes who were not actively working with the chatbot showed slight increases in their A1C. A rigorous study will be required to confirm this data; however, this level of improvement from an automated educational tool is exciting and similar to what is expected when a single new diabetes medication is added to treat elevated blood sugars.
Patient satisfaction results were also encouraging. Through the chatbot dashboard and our technology partner, we deployed satisfaction surveys and learned that the platform had been easy to use and effective. Participant responses included:
- “I received a lot of good information about diabetes that I didn’t know.”
- “I thank them for helping me manage my diabetes and feel great again.”
- “I like that it connects me to my doctor when things are not right. It makes me feel like somebody cares.”
Overall, we were very pleased to see preliminary evidence that engaging with the chatbot may help people improve their blood glucose management. With future research, we hope to learn more about using chatbot technology to support persons living with diabetes and their providers.
Next steps: further study.
In June 2022, co-author Carine M. Nassar, RD, MS, CDCES, was invited to present on this project at the American Diabetes Association’s 82nd Scientific Sessions, confirming broad interest in these findings.
Our next steps will be to publish our preliminary data and conduct a rigorous scientific study to refine the user experience and better understand the value of chat-based diabetes management to patients, providers, and health care systems.
This chatbot program exemplifies how digital communications systems are beginning to transform healthcare. It fits into MedStar Health’s system-wide Connected Care Access, Research, Equity (CARE) initiative, which is an effort to increase and improve virtual care delivery.
With further research, we anticipate uncovering actionable data that may lend itself to applications using Artificial Intelligence delivery methods. These innovative digital health care delivery strategies will be an essential part of the technology ecosystem which supports persons with diabetes and their providers in improving patient outcomes.
