Gastrointestinal toxicities (GITs) in clinical trials often lead to dose-limitations that reduce drug efficacy and delay treatment optimization. Preclinical animal models do not accurately replicate human physiology, leaving few options for early detection of GITs, such as diarrhea, before human studies. Chemotherapeutic agents, known to cause clinical diarrhea, frequently target mitotic cells. Therefore, we hypothesized a model utilizing proliferative cell populations derived from human intestinal crypts would predict clinical diarrhea occurrence with high accuracy. Here, we describe the development of a diarrhea prediction assay utilizing RepliGut® Planar, a primary intestinal stem cell-derived platform. To evaluate the ability of this model to predict clinical diarrhea risk, we assessed toxicity of 30 marketed drugs by measuring cell proliferation (EdU incorporation), cell abundance (nuclei quantification), and barrier function (TEER) in cells derived from three human donors. Dose response curves were generated for each drug, and the IC15 to Cmax ratio was used to identify a threshold for assay positivity. This model accurately predicted diarrhea potential, achieving an accuracy of 91 % for proliferation, 90 % for abundance, and 88 % for barrier formation. In vitro toxicity screening using primary proliferative cells may reduce clinical diarrhea and ultimately lead to safer and more effective treatments for patients.