Minimizing STOPP and Beers Criteria Risks in PIM Treatments Using PM-TOM and ChatGPT: A Case Study

PM-TOM (Personalized Medicine-Therapy Optimization Method) is a clinical decision-support tool designed to optimize polypharmacy treatments by minimizing their adverse drug reactions (ADRs) caused by individual drugs or drug interactions (DDIs, DCIs, DFIs, DGIs), along with the risks identified by the STOPP and Beers criteria. On the other hand, AI tools like ChatGPT 4.0, trained on medical literature texts, can provide broader clinical reasoning and insights tailored to individual patient contexts. By referring to a documented deprescribing case, this study demonstrates the synergistic power of PM-TOM and ChatGPT in optimizing potentially inappropriate medication (PIM) treatments. A malnourished older woman was admitted to a deprescribing facility with recurrent falls, hypertension, ischemic heart disease, depression, osteoarthritis, osteoporosis, and GERD. She was initially prescribed acetaminophen, alendronate, omeprazole, lisinopril, metoprolol, aspirin, citalopram, and vitamin D, which were assessed as inadequate. While the discharge regimen improved some conditions by replacing alendronate with zoledronic acid and reducing some drug dosages, PM-TOM revealed that key risks, stemming primarily from omeprazole, aspirin, and citalopram, remained unaddressed. The discharge treatment was optimized with PM-TOM after considering alternative drug classes suggested by ChatGPT and elaborated in the available medical literature. In the optimized treatment, omeprazole (PPI) was replaced with famotidine (H2-blocker), citalopram (SSRI) with agomelatine (atypical antidepressant), zoledronic acid (bisphosphonate) with denosumab (RANK ligand inhibitor), aspirin (NSAID) with ticagrelor (antiplatelet), and lisinopril with benazepril (ACE inhibitor). These changes significantly reduced possible ADRs and the geriatric care criteria risks. Finally, ChatGPT validated the proposed adjustments, confirming their alignment with the guidelines and highlighting the potential for longer-term benefits. This case study illustrates how a combined use of PM-TOM and AI tools can effectively support the clinical decision-making process by optimizing polypharmacy treatments and minimizing their PIMs, major contributors to morbidity in older adults and high healthcare costs.