Current physiologically-based biopharmaceutics modeling (PBBM) neglects the effect of gastrointestinal stress events on the disintegration and dissolution of oral solid dosage forms. Biorelevant dissolution testing can simulate the behavior of drug products under physiological agitation but a workload limits variability examination. In this study, we overcame these deficiencies by inputting dissolution profiles generated by machine-learning (ML) into PBBM-based simulations. Our specific aim was to examine how the varied timing of intragastric stress and housekeeping wave (GET) and fasted stomach pH affect dabigatran exposure from the Pradaxa capsule. Twenty experimental dissolution profiles of dabigatran etexilate from the flow-through apparatus PhysioCell and 1,036 ML-derived profiles representing various gastric motility patterns were a basis for single-dose simulations. A novel timewise dissolution model, which estimates the first-order rate constants at consecutive two-point time intervals, provided an excellent fit to the highly irregular and variable dissolution curves (coefficient of determination ≥ 0.9835, median 0.9992). The time between the onset of dissolution (T