[Abstract]: Objective: To optimise the current strategy of prostate puncture biopsy by analysing the clinical data of high-risk prostate cancer patients, establishing a model to predict the pathological nature of their prostate lesions, and identifying patients who can be exempted from systematic biopsy and undergo targeted puncture alone. Methods: Clinical data of patients who underwent prostate puncture at the First Affiliated Hospital of Nanjing Medical University from January 2022 to June 2024 were retrospectively analysed, and eligible patients were screened and divided into a training set and a validation set. Single-factor and multifactor logistic regression analyses were used to screen the significant correlates of the pathology of prostate targeted biopsy. A predictive model was constructed based on the data from the training set, and the model efficacy was verified in the validation set. Receiver Operating Characteristic (ROC) curves were used to assess the diagnostic performance of the model in both data sets.Results: Age (X1), number of lesions (X2), histological region in which the lesions are located (X3), Prostate Imaging Reporting and Data System (PI-RADS) score (X4), and prostate-specific antigen density (X5) were the variables associated with the patient's prostate targeted biopsy pathology. The mathematical expression of the predictive model is: p=1/[1+e^(-15.770+0.067×X1-0.658×X2+0.381×X3+2.271×X4+5.742×X5)]. The Area under the curve (AUC) of the prediction model was 0.856 (95% confidence interval: 0.812-0.900) in the training set and 0.886 (95% confidence interval: 0.776-0.995) in the validation set.Conclusion: The constructed prediction model for targeted biopsy pathology in high-risk prostate cancer patients can guide the clinical strategy of prostate puncture to maintain good diagnostic performance while reducing the number of puncture needles, thus reducing the occurrence of complications and the waste of medical resources. Keywords: prostate cancer; prostate puncture biopsy; MRI; predictive model;