Introduction
The demographic landscape is undergoing a significant transformation. By 2050, the segment of the population aged over 60 is set to double, as per the World Health Organization. This demographic shift presents a growing challenge to healthcare systems worldwide, particularly in the management of polypharmacy, defined as the concurrent use of five or more medications. Over 40% of older adults are now navigating the complexities of polypharmacy, often under the guidance of their primary care physicians (PCPs).
The Challenge of Polypharmacy
PCPs are witnessing an increased burden, with a 34% rise in specialist consultations for Medicare beneficiaries since 2000. The intricate web of medications prescribed by various specialists often culminates in the hands of PCPs, who must manage potential drug interactions and adverse effects. Deprescribing, the process of reducing or stopping medications that may no longer be beneficial or might be harmful, is a crucial but intricate strategy in mitigating these risks.
AI-Assisted Deprescribing: A New Frontier
Recent advancements in artificial intelligence (AI), specifically natural language processing (NLP), offer promising solutions for managing polypharmacy. ChatGPT, an AI model developed by OpenAI, has demonstrated high accuracy in medical applications, including the deprescribing process. Recent research presents the first use case of ChatGPT in this context, revealing its potential to align with physician decision-making when managing medication in geriatric patients.
The Impact of AI in Polypharmacy Management
With an ageing population and a decline in regular PCP visits, the need for efficient and accurate medication management is more critical than ever. ChatGPT has shown a propensity to deprescribe medications, particularly in patients without cardiovascular disease history. However, the decision-making process does not solely hinge on the presence of cardiovascular conditions; the severity of impairment in activities of daily living also plays a role.
Limitations and Considerations
Despite its potential, this study acknowledges limitations, including the standardised patient profile used in the vignettes, which may not represent the diversity of the geriatric population. Furthermore, the susceptibility of AI models like ChatGPT to generate incorrect or fabricated information, known as hallucinations, necessitates caution.
Conclusion
The integration of AI, such as ChatGPT, into polypharmacy management could signify a substantial stride forward in improving medication use and reducing adverse drug events among the elderly. Nonetheless, it is imperative to approach this technology with prudence, considering its current limitations and the evolving landscape of medical practice.