Random forest machine learning analysis of 390 ambulatory patients aged 65+ in Nepal identified polypharmacy and need for medication assistance as the primary drivers of both medication burden and poor adherence. The study found median medication burden scores of 110 (moderate level) and adherence scores indicating moderate non-adherence, with complex medication regimens creating the greatest challenges for older adults. This research addresses a critical gap in geriatric care as Nepal's population rapidly ages. The machine learning approach offers unprecedented insight into medication management complexity, revealing that while polypharmacy typically worsens adherence, the negative effects are partially offset when patients receive physical or cognitive assistance. Financial factors and employment status also emerged as significant predictors. These algorithmic insights could revolutionize medication optimization strategies for aging populations globally, suggesting that clinical interventions should prioritize regimen simplification and targeted support systems. However, as a preprint awaiting peer review, these findings require validation before clinical implementation. The work represents an important step toward data-driven geriatric care, though broader geographic validation would strengthen generalizability beyond Nepal's healthcare context.