The Role Of Predictive Models Based On Multiple Clinical Parameters And Various Artificial Intelligence Algorithms In Predicting Adverse Pathology Of Prostate Cancer

Background: The present study aimed to develop three nomograms by incorporating multiple clinical characteristics to identify those prostate cancer(PCa)patients with high probability of incorrect biopsy Gleason grade group(GG)before making treatment decisions.Methods: We retrospectively collected data from PCa patients who underwent systematic biopsy and radical prostatectomy from January 2012 to December 2019 at Tongji Hospital of Tongji Medical College,Huazhong University of Science and Technology.Univariable and multivariable logistic regression analyses were preformed to identify independent risk factors associated with upgrading,upstaging and downgrading.By incorporating selected clinical parameters with high predictive value,we constructed three nomograms to predict the probability of upgrading,upstaging and downgrading.Discrimination of nomograms was evaluated by receiver operating characteristic(ROC)analysis with corresponding area under the curve(AUC).Decision curve analysis(DCA)and calibration curves were performed to evaluate calibration and the clinical usefulness of nomograms.Performance of the three nomograms was validated in the testing dataset.Results: There were 585 PCa patients in total enrolled in this study who met the inclusion criteria.Of the 585 patients,the disease of 262(44.8%)was upgraded and 68(11.6%)was downgraded,and the disease of 67(11.5%)was upstaged.With regard to findings of multivariable analyses,patients’ age and biopsy GG(GG 2,GG 3,GG 4 versus GG 1)were significantly associated with upgrading.Moreover,maximum diameter of the index lesion(D-max),clinical T stage(c T3 a,c T3 b versus c T1-2),number of positive cores and total tumor length were significantly associated with upstaging.Furthermore,D-max,%f PSA(> 0.16 versus ≤ 0.16)and biopsy GG(GG 3,GG 4,GG 5 versus GG 2)were independent predictors of downgrading.The three nomograms displayed good calibration in respective calibration plots.ROC analyses showed good discrimination with satisfactory AUC values and DCA plots demonstrated that the upgrading-risk nomogram,upstaging-risk nomogram and downgrading-risk nomogram were all clinically useful.Conclusions: The upgrading-risk nomogram,upstaging-risk nomogram,and downgrading-risk nomogram were developed and correctly predicted the probability of incorrect Gleason grade group assigned to patients undergoing systematic biopsy.Background: This study aimed to develop a machine learning(ML)-assisted model capable of accurately predicting the probability of biopsy Gleason grade group upgrading before making treatment decisions.Methods: We retrospectively collected data from prostate cancer(PCa)patients.Four ML-assisted models were developed from 16 clinical features using logistic regression(LR),logistic regression optimized by least absolute shrinkage and selection operator(Lasso)regularization(Lasso-LR),random forest(RF)and support vector machine(SVM).The area under the curve(AUC)was applied to determine the model with the highest discrimination.Calibration plots and decision curve analysis(DCA)were performed to evaluate the calibration and clinical usefulness of each model.Results: A total of 530 PCa patients were included in this study.The Lasso-LR model showed good discrimination with an AUC,accuracy,sensitivity,specificity,positive predictive value(PPV)and negative predictive value(NPV)of 0.776,0.712,0.679,0.745,0.730 and 0.695,respectively,followed by SVM(AUC 0.740,95% confidence interval [CI]: 0.690–0.790),LR(AUC 0.725,95% CI: 0.674–0.776)and RF(AUC 0.666,95% CI: 0.618–0.714).Validation of the model showed that the Lasso-LR model had the best discriminative power(AUC 0.735,95% CI: 0.656–0.813),followed by SVM(AUC 0.723,95% CI: 0.644–0.802),LR(AUC 0.697,95% CI: 0.615–0.778)and RF(AUC 0.607,95% CI: 0.531–0.684)in the testing dataset.Both the Lasso-LR and SVM models were well-calibrated.DCA plots demonstrated that the predictive models except RF were clinically useful.Conclusions: The Lasso-LR model had good discrimination in the prediction of patients at high risk of harboring incorrect Gleason grade group assignment,and the use of this model may be greatly beneficial to urologists in treatment planning,patient selection,and the decision-making process for PCa patients.Background: To develop novel models for predicting extraprostatic extension(EPE),seminal vesicle invasion(SVI),or upgrading in prostate cancer(PCa)patients using clinical parameters,biparametric magnetic resonance imaging(bp-MRI),and transrectal ultrasonography(TRUS)-guided systematic biopsies.Methods: We retrospectively collected data from PCa patients who underwent standard(12-core)systematic biopsy and radical prostatectomy.To develop predictive models,the following variables were included in multivariable logistic regression analyses: total prostate-specific antigen(TPSA),central transition zone volume(CTZV),prostate-specific antigen(PSAD),maximum diameter of the index lesion at bp-MRI,EPE at bp-MRI,SVI at bp-MRI,biopsy Gleason grade group,and number of positive biopsy cores.Three risk calculators were built based on the coefficients of the logit function.The area under the curve(AUC)was applied to determine the models with the highest discrimination.Decision curve analyses(DCAs)were performed to evaluate the net benefit of each risk calculator.Results: A total of 222 patients were included in this study.Overall,83(37.4%),75(33.8%),and 107(48.2%)patients had EPE,SVI,and upgrading at final pathology,respectively.The addition of bp-MRI data improved the discrimination of models for predicting SVI(0.807 vs 0.816)and upgrading(0.548 vs 0.625)but not EPE(0.766 vs 0.763).Similarly,models including clinical parameters,bp-MRI data,and information on systematic biopsies achieved the highest AUC in the prediction of EPE(0.842),SVI(0.913),and upgrading(0.794),and the three corresponding risk calculators yielded the highest net benefit.Conclusions: We developed three easy-to-use risk calculators for the prediction of adverse pathological features based on patient clinical parameters,bp-MRI data,and information on systematic biopsies.This may be greatly beneficial to urologists in the decision-making process for PCa patients.