MP64-01 PREDICTING RISK OF SIDE-SPECIFIC EXTRAPROSTATIC EXTENSION IN MEN WITH PROSTATE CANCER USING EXPLAINABLE ARTIFICIAL INTELLIGENCE

You have accessJournal of UrologyProstate Cancer: Localized: Surgical Therapy VI (MP64)1 Sep 2021MP64-01 PREDICTING RISK OF SIDE-SPECIFIC EXTRAPROSTATIC EXTENSION IN MEN WITH PROSTATE CANCER USING EXPLAINABLE ARTIFICIAL INTELLIGENCE Jethro Kwong, Adree Khondker, Christopher Tran, Emily Evans, Amna Ali, Munir Jamal, Thomas Short, Frank Papanikolaou, John Srigley, and Andrew Feifer KwongJethro Kwong More articles by this author , KhondkerAdree Khondker TranChristopher Tran EvansEmily Evans AliAmna Ali JamalMunir Jamal ShortThomas Short PapanikolaouFrank Papanikolaou SrigleyJohn Srigley FeiferAndrew View All Author Informationhttps://doi.org/10.1097/JU.0000000000002104.01AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Current machine learning (ML) models are limited poor interpretability, precluding their routine use in planning nerve-sparing at radical prostatectomy (RP). We aimed leverage explainable artificial intelligence techniques provide accurate, interpretable, personalized predictions for side-specific extraprostatic extension (ssEPE). METHODS: A retrospective sample 900 prostatic lobes (450 patients) from RP specimens our institution between 2010 2020, was used as the training cohort. Features (ie: variables) included patient demographics, clinical, sonographic, site-specific data transrectal ultrasound-guided prostate biopsy. The label outcome) interest presence ssEPE specimen. ten-fold stratified cross-validation method performed train a gradient-boosted model, optimize hyperparameters, internal validation. Our model further externally validated using testing cohort 122 (61 separate 2016 2020. An existing literature which has highest performance predicting selected baseline comparison. Discriminative capability quantified area under receiver-operating-characteristic (AUROC) precision-recall curve (AUPRC). Clinical utility determined decision analysis. Shapley Additive exPlanations were interpret ML model’s predictions. RESULTS: incidence cohorts 30.7 41.8%, respectively. outperformed with mean AUROC 0.81 vs 0.75 (p<0.01) AUPRC 0.69 0.60, respectively, on Similarly, favourably an 0.76 (p=0.03) 0.78 0.72. On analysis, achieved higher net benefit threshold probabilities 0.15 0.65. web application incorporating developed de-identified can be inputted generate individualized map annotated explanations highlight features had greatest impact (www.ssepe.ml). CONCLUSIONS: user-friendly that enables physicians without prior experience assess risk understand factors driving these aid surgical counselling. Source Funding: None © 2021 American Urological Association Education Research, Inc.FiguresReferencesRelatedDetails Volume 206Issue Supplement 3September 2021Page: e1110-e1110 Advertisement Copyright & Permissions© Inc.MetricsAuthor Information Expand Loading ...