Mp07-13 Contextualized Surgical Gestures in Nerve-Spare Reveal Heterogeneity in Techniques and Predicts Erectile Function Recovery

You have accessJournal of UrologySurgical Technology & Simulation: Artificial Intelligence I (MP07)1 May 2024MP07-13 CONTEXTUALIZED SURGICAL GESTURES IN NERVE-SPARE REVEAL HETEROGENEITY IN TECHNIQUES AND PREDICTS ERECTILE FUNCTION RECOVERY Runzhuo Ma, Jiayun Wang, Joseph Rich, Sherry L. Huang, Maxwell X. Otiato, Istabraq Dalieh, Jonathan You, Alvin C. Goh, Geoffrey A. Sonn, Christian Wagner, Animashree Anandkumar, Andrew J. Vickers, Jim C. Hu, and Andrew J. Hung Runzhuo MaRunzhuo Ma , Jiayun WangJiayun Wang , Joseph RichJoseph Rich , Sherry L. HuangSherry L. Huang , Maxwell X. OtiatoMaxwell X. Otiato , Istabraq DaliehIstabraq Dalieh , Jonathan YouJonathan You , Alvin C. GohAlvin C. Goh , Geoffrey A. SonnGeoffrey A. Sonn , Christian WagnerChristian Wagner , Animashree AnandkumarAnimashree Anandkumar , Andrew J. VickersAndrew J. Vickers , Jim C. HuJim C. Hu , and Andrew J. HungAndrew J. Hung View All Author Informationhttps://doi.org/10.1097/01.JU.0001008728.41882.d7.13AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Heterogeneity in nerve-spare (NS) techniques during robotic-assisted radical prostatectomy (RARP) results in variation in postoperative recovery of erectile function (EF). Deconstructing and classifying surgical gestures during NS provide a novel method to capture variances in NS techniques. Herein, we aim to delineate the heterogeneity of NS by contextualized surgical gesture data and utilize machine learning (ML) to predict EF recovery after RARP. METHODS: RARPs from four international centers were prospectively captured from July 2016 to January 2022. Inclusion criteria comprised preop Sexual Health Inventory for Men (SHIM) ≥ 17 and use of NS technique (69 full NS, 28 partial NS). We defined EF recovery as return to baseline SHIM≥17 at 1 year. NS videos were annotated into sequences of surgical gestures using a validated classification system, with the context of the gestures in specific anatomical locations and bundling gestures together for a surgical task (Figure 1a-c). Multilayer perceptron model was trained to read gesture data and patient features (e.g., age, comorbidity) to predict EF recovery. RESULTS: Ninety-seven patients from 26 surgeons were included. 42/97 (43%) patients recovered EF at 12 months. The median number of gestures/case was 499 (IQR 290-784), with significant variances in their distribution across anatomical locations and surgical tasks (Figure 1b-c). Different dissection techniques were revealed: clip usage ranged from 0 (clipless) to 33 times (median 8, IQR 6-13); energy usage ranged from 0 (athermal) to 99 times (median 17, IQR 9-36); sharp/blunt dissection gesture ratio ranged from 0.09 to 4.17 (median 0.47, IQR 0.31-0.83). Energy usage was found to be adversely associated with EF recovery (p=0.018, Chi-square test, Figure 1d), while clip usage (p=0.7) and sharp/blunt dissection gesture ratio (p=0.4) were not associated with EF recovery. Using contextualized surgical gesture data, the ML model predicted EF recovery with an AUC of 0.69 (95% CI 0.59-0.79) over a 5-fold cross validation, and it increased to 0.73 (95% CI 0.63-0.83) after adding patient features. CONCLUSIONS: Surgical gesture data revealed significant heterogeneity in NS dissection techniques. Using contextualized surgical gesture data in NS, ML model has potential to predict EF outcome. Download PPT Source of Funding: Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under Award Number R01CA273031. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health © 2024 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 211Issue 5SMay 2024Page: e109 Advertisement Copyright & Permissions© 2024 by American Urological Association Education and Research, Inc.Metrics Author Information Runzhuo Ma More articles by this author Jiayun Wang More articles by this author Joseph Rich More articles by this author Sherry L. Huang More articles by this author Maxwell X. Otiato More articles by this author Istabraq Dalieh More articles by this author Jonathan You More articles by this author Alvin C. Goh More articles by this author Geoffrey A. Sonn More articles by this author Christian Wagner More articles by this author Animashree Anandkumar More articles by this author Andrew J. Vickers More articles by this author Jim C. Hu More articles by this author Andrew J. Hung More articles by this author Expand All Advertisement PDF downloadLoading ...