Polypharmacy, the simultaneous use of five or more medications, is common in older adults and increases the risk of adverse drug interactions. Although discontinuing unnecessary medications can mitigate this risk, the decision-making process can be complex and time-consuming. There is a growing need for effective polypharmacy management tools that can support primary care physicians facing staff shortages.
In a new study, researchers from Mass General Brigham MESH Incubator found that ChatGPT, a generative artificial intelligence (AI) chatbot, is a promising tool for managing polypharmacy and prescriptions. Published on April 18 in the Journal of Medical Systems, these findings represent the first use case of AI models in medical management.
To assess its utility, researchers provided ChatGPT with various clinical scenarios and posed a series of decision-making questions. In each scenario, the same older adult patient was taking a mix of medications but had differences in their history of cardiovascular disease (CVD) and the degree of impairment in activities of daily living (ADL).
When asked yes-or-no questions about reducing prescribed medications, ChatGPT consistently recommended against prescribing medications for patients with no history of cardiovascular disease. However, when a comorbid cardiovascular disease was introduced, it was more cautious, more likely to maintain the patient’s medication regimen. In both cases, researchers found that the severity of ADL impairment did not appear to influence the decision outcomes.
The team also noted that ChatGPT tended to neglect pain and favored prescribing pain relievers over other types of medications, such as statins or blood pressure medications. Furthermore, ChatGPT’s responses varied when the same scenario was presented in new chat sessions—which the authors suggest could be due to inconsistencies in commonly reported clinical prescribing trends that the model was trained on.
More than 40 percent of older adults meet the criteria for polypharmacy. The number of seniors receiving Medicare has increased their care teams with more specialists in recent years, requiring primary care physicians to oversee medication management. According to the researchers, an effective AI tool could support this practice.
https://link.springer.com/article/10.1007/s10916-024-02058-y
