Polypharmacy is a major challenge in home care, increasing risks of adverse drug events, readmission, and mortality. Existing medication review tools (e.g. the Medication Regimen Complexity Index (MRCI), Beers, STOPP/START, PIM-Taiwan) are often fragmented and time-consuming.This study aimed to develop an integrated medication optimization flowchart that incorporates these criteria and to evaluate the feasibility and effectiveness of using a Large Language Model (LLM) to implement this flowchart for home care patients with polypharmacy.Home care patients with MRCI scores above 40 were included. Comprehensive medication lists were collected for each patient. A LLM (ChatGPT, GPT-5.1 model)  was adapted using structured prompts and rule-based constraints to apply the Beers Criteria (2023), STOPP/START (v3), and PIM-Taiwan (2023). The AI generated prescribing suggestions, identified potentially inappropriate medications, and recommended potential omissions based on START criteria. AI-generated recommendations were reviewed by two family physicians and implemented when clinically appropriate. The MRCI scores were calculated pre- and post-intervention. Paired t-tests were conducted to assess the significance of changes in medication regimen complexity.A total of 26 patients (mean age: 86.3) were included. Participants had an average of 5.7 chronic diseases and used an average of 13 medications. The mean MRCI score decreased significantly from 48.29 ± 10.41 pre-intervention to 40.69 ± 11.05 post-intervention, representing a mean reduction of 7.6 points (t(25) = -4.4, p < 0.001). No adverse events related to deprescribing were reported during follow-up.The integrated flowchart demonstrated feasibility and efficiency in reducing medication regimen complexity. The LLM significantly improved the speed and consistency of applying multiple prescribing criteria, indicating its potential as a decision-support tool. However, AI outputs required mandatory clinician verification to mitigate risks of hallucination or misinterpretation. The focus was on simplifying regimens; the direct impact on clinical outcomes like readmission warrants future longitudinal studies.This study introduces an AI-assisted integrated medication optimization flowchart that enhances rational prescribing and reduces polypharmacy in home care settings. The use of a LLM shows strong potential as a valuable decision-support tool in primary care, particularly in resource-limited environments. Further studies are warranted to assess clinical outcomes and scalability across different healthcare systems.