An AI-powered chatbot for supporting the medication information needs of older adults

Principal Investigator:            Kirk Roberts, UTHealth-SBMI 

Co-Investigators:                    Muhammad Amith, UTHealth-SBMI

Mary E Ross, UTHealth-Cizik SON

Mandi Lyons, UTHealth-Cizik SON 

Older adults have complicated medication regimens, and suboptimal management of these medications can result in significant harm. A primary issue for older adults in these circumstances is access to the right information on their medications, which a Smart Apartment can aid in by the use of voice-activated artificial intelligence agents (chatbots) that can answer an older adult’s questions on a particular medication. Since they are AI agents, they are available 24/7 and can draw from a wide variety of publicly available drug resources. In this project, we have focused on assessing the feasibility of such a chatbot. First, we interviewed a range of geriatrics experts on the benefits, barriers, and information needs that such a chatbot would have for older adults. We found that at least for adults 65-75 years old, the technical barriers of using such a system would not be daunting to this age group. Second, we developed natural language processing (NLP) methods to extract information from FDA-mandated structured product labels (better known as drug labels), the canonical source for manufacturer-supplied drug information. We focused on 20 drugs of particular relevance to older adults. Finally, we piloted an ontology-driven voice chatbot based on a drug information ontology. The goal of this voice chatbot was to help older adults better adhere to their medication regimens. Overall, this pilot project demonstrated that the key pieces of a Smart Apartment-based voice chatbot for medication information for older adults is feasible. Future work will focus on scaling up the data and resources, as well as piloting the chatbot in realistic scenarios.

Publications:

Gudala M, Ross MET, Mogalla S, Lyons M, Ramaswamy P, Roberts K
Benefits of, Barriers to, and Needs for an Artificial Intelligence–Powered Medication Information Voice Chatbot for Older Adults: Interview Study With Geriatrics Experts
JMIR Aging 2022;5(2):e32169
doi: 10.2196/32169