Related Experiment Video
Updated: Jun 28, 2025

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
Proactive Polypharmacy Management Using Large Language Models: Opportunities to Enhance Geriatric Care
Arya Rao1,2,3, John Kim1,2,3, Winston Lie1,2,3
1Harvard Medical School, Boston, MA, USA.
Large language models (LLMs) like ChatGPT show potential in aiding polypharmacy management by making deprescribing recommendations. Their decisions align with clinical factors, suggesting future support for primary care physicians.
Area of Science:
- Artificial Intelligence in Medicine
- Clinical Decision Support Systems
- Geriatric Pharmacotherapy
Background:
- Polypharmacy presents a significant challenge in managing complex patients, especially with an aging population and primary care shortages.
- Effective management strategies for polypharmacy are critical to reduce healthcare burdens.
- The role of artificial intelligence (AI), specifically large language models (LLMs), in polypharmacy management remains largely unevaluated.
Purpose of the Study:
- To evaluate the performance of ChatGPT, a large language model, in making deprescribing decisions for patients with polypharmacy.
- To assess ChatGPT's capacity to aid in clinical decision-making for medication management in complex patients.
Main Methods:
- Standardized clinical vignettes, originally used to study general practitioners' deprescribing decisions, were inputted into ChatGPT 3.5.
- ChatGPT's performance was evaluated based on binary yes/no deprescribing recommendations and list-based prompts for medication selection.
- Responses regarding deprescribing decisions, including the number and types of medications, were recorded and analyzed.
Main Results:
- ChatGPT universally recommended deprescribing without considering activities of daily living (ADL) status in patients without cardiovascular disease (CVD) history.
- In patients with CVD history, ChatGPT's recommendations varied across technical replicates.
- The number of medications recommended for deprescribing (2.67-3.67 out of 7) increased with ADL impairment severity but was not influenced by CVD status.
- ChatGPT preferentially recommended deprescribing pain medications.
Conclusions:
- ChatGPT's deprescribing decisions demonstrated variability based on ADL status, CVD history, and medication type, indicating some alignment with clinical reasoning.
- These findings suggest that specifically trained LLMs could offer valuable clinical support to primary care physicians in managing polypharmacy.
More Related Videos
Related Concept Videos
Drug Therapy
Antianxiety Medications
Methods of Documentation VI: Case Management Model
For example, a patient with a chronic...
Analysis of Population Pharmacokinetic Data
Model Approaches for Pharmacokinetic Data: Distributed Parameter Models
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
Factors Affecting Drug Response: Overview
Pharmacokinetic Models: Overview
There are three primary types of models: empirical, compartment, and physiological. Empirical models, with minimal...

