Toward adaptive therapeutic timing: integration of mechanistic pharmacology and artificial intelligence in precision dosing

Artificial intelligence (AI) is reshaping the dosing of medication. It is shifting practice from fixed schedules to flexible timing that better reflects the needs of individual patients. This review synthesizes recent work in which AI identifies improved windows for drug delivery. It spans multiple time horizons, ranging from minute-by-minute infusion control to regimen planning over weeks or months. These approaches integrate population pharmacokinetic modeling with machine learning and reinforcement learning. Each component serves a distinct purpose. Together, they can recommend the dose timing, adjust dosing intervals, and indicate when a planned treatment pause may be appropriate. The models link these recommendations to the physiological rhythms and current disease status of patients, so timing becomes a defined part of the dosing strategy rather than a simple clock-based rule. Recent advances cluster into three directions. First, reinforcement learning supports sequential dosing decisions in long-term therapies, where early choices shape later outcomes. Second, AI enables chronotherapy by aligning drug delivery with daily circadian biology. Third, hybrid models improve exposure prediction, which supports more accurate interval personalization for individual patients. However, key limitations remain. Many models show limited generalizability across clinical settings, and many methods still require rigorous clinical validation. Even with these constraints, the trajectory is consistent. AI is positioned to move therapeutic timing from a static calendar task to an adaptive, patient-centered element of precision medicine.