In a new study, researchers from the Mass General Brigham MESH Incubator found that AI-generated chatbots ChatGPT The study, published in the Journal of Medical Systems, shows the potential of AI models in managing polypharmacy and drug deprescribing.MedicalIn managementThe firstApplication cases.
To evaluate the usefulness of ChatGPT, the researchers presented it with different clinical scenarios and asked it a series of decision-making questions. Each scenario included the same elderly patient who was taking multiple medications but had varying histories of cardiovascular disease and levels of impairment in activities of daily living.
Source: The image is generated by AI, and the image is authorized by Midjourney
When asked whether to reduce prescribed medications, ChatGPT consistently recommended reducing medications in patients without a history of cardiovascular disease. However, when known cardiovascular disease was introduced, ChatGPT was more cautious and more likely to keep the patient's medication prescription unchanged. The researchers observed that regardless of the severity of impairment in activities of daily living, there did not seem to be an impact on decision outcomes.
The research team also noticed that ChatGPT tended to discount pain and favored reducing pain medications over other drug types, such as statins or blood pressure medications. In addition, ChatGPT’s responses varied when presented with the same scenario in a new chat session, which may reflect inconsistencies in the clinical deprescribing trends used to train the model.
Older adults with more than 40% met the criteria for polypharmacy. In recent years, older adults enrolled in Medicare have received care from more specialist teams, resulting in the need for primary care providers to monitorMedication managementAccording to the researchers, effective AI tools could help alleviate this situation.
ResearchSenior"Our study provides new insights into ChatGPT as a clinical support tool for medication management," said corresponding author Marc Succi, PhD, a professor of medicine at Harvard Medical School.The firstAlthough the accuracy of such models should be improved with caution, AI-Assisted“The management of polypharmacy can help reduce the burden on primary care physicians. Further research into specifically trained AI tools could significantly improve the care of older patients.”