Tacrolimus is a commonly used medication for the treatment of nephrotic syndrome. Due to its narrow therapeutic window and significant pharmacokinetic differences among individuals, therapeutic drug monitoring is required during its clinical use. In the process of therapeutic drug monitoring, machine learning-based personalized dosing prediction models for tacrolimus can excavate medication patterns from a large amount of clinical data, assist in clinical decision-making, and achieve individualized precise medication. Machine learning models, the application progress of machine learning in personalized administration of tacrolimus for patients with nephrotic syndrome, modeling points of machine learning prediction models, and the limitations of current prediction models were reviewed in this paper, which could provide references for future research in this field.