Use of machine learning to predict early biochemical recurrence after robot-assisted prostatectomy

被引:76
作者
Wong, Nathan C. [1 ]
Lam, Cameron [1 ]
Patterson, Lisa [1 ]
Shayegan, Bobby [1 ]
机构
[1] McMaster Univ, Div Urol, Dept Surg, Hamilton, ON, Canada
关键词
predictive model; prostate cancer; biochemical recurrence; machine learning; NATURAL-HISTORY; CANCER; MORTALITY; DISEASE; FUTURE;
D O I
10.1111/bju.14477
中图分类号
R5 [内科学]; R69 [泌尿科学(泌尿生殖系疾病)];
学科分类号
1002 ; 100201 ;
摘要
ObjectivesTo train and compare machine-learning algorithms with traditional regression analysis for the prediction of early biochemical recurrence after robot-assisted prostatectomy. Patients and MethodsA prospectively collected dataset of 338 patients who underwent robot-assisted prostatectomy for localized prostate cancer was examined. We used three supervised machine-learning algorithms and 19 different training variables (demographic, clinical, imaging and operative data) in a hypothesis-free manner to build models that could predict patients with biochemical recurrence at 1 year. We also performed traditional Cox regression analysis for comparison. ResultsK-nearest neighbour, logistic regression and random forest classifier were used as machine-learning models. Classic Cox regression analysis had an area under the curve (AUC) of 0.865 for the prediction of biochemical recurrence. All three of our machine-learning models (K-nearest neighbour (AUC 0.903), random forest tree (AUC 0.924) and logistic regression (AUC 0.940) outperformed the conventional statistical regression model. Accuracy prediction scores for K-nearest neighbour, random forest tree and logistic regression were 0.976, 0.953 and 0.976, respectively. ConclusionsMachine-learning techniques can produce accurate disease predictability better that traditional statistical regression. These tools may prove clinically useful for the automated prediction of patients who develop early biochemical recurrence after robot-assisted prostatectomy. For these patients, appropriate individualized treatment options can improve outcomes and quality of life.
引用
收藏
页码:51 / 57
页数:7
相关论文
共 12 条
[1]   Radical Prostatectomy versus Watchful Waiting in Early Prostate Cancer [J].
Bill-Axelson, Anna ;
Holmberg, Lars ;
Ruutu, Mirja ;
Garmo, Hans ;
Stark, Jennifer R. ;
Busch, Christer ;
Nordling, Stig ;
Haggman, Michael ;
Andersson, Swen-Olof ;
Bratell, Stefan ;
Spangberg, Anders ;
Palmgren, Juni ;
Steineck, Gunnar ;
Adami, Hans-Olov ;
Johansson, Jan-Erik .
NEW ENGLAND JOURNAL OF MEDICINE, 2011, 364 (18) :1708-1717
[2]   Variation in the definition of biochemical recurrence in patients treated for localized prostate cancer: The American Urological Association Prostate Guidelines for Localized Prostate Cancer update panel report and recommendations for a standard in the reporting of surgical outcomes [J].
Cookson, Michael S. ;
Aus, Gunnar ;
Burnett, Arthur L. ;
Canby-Hagino, Edith D. ;
D'Amico, Anthony V. ;
Dmochowski, Roger R. ;
Eton, David T. ;
Forman, Jeffrey D. ;
Goldenberg, S. Larry ;
Hernandez, Javier ;
Higano, Celestia S. ;
Kraus, Stephen R. ;
Moul, Judd W. ;
Tangen, Catherine ;
Thrasher, J. Brantley ;
Thompson, Ian .
JOURNAL OF UROLOGY, 2007, 177 (02) :540-545
[3]   Impact of positive surgical margins on prostate cancer recurrence and the use of secondary cancer treatment: Data from the CaPSURE database [J].
Grossfeld, GD ;
Chang, JJ ;
Broering, JM ;
Miller, DP ;
Yu, J ;
Flanders, SC ;
Henning, JM ;
Stier, DM ;
Carroll, PR .
JOURNAL OF UROLOGY, 2000, 163 (04) :1171-1177
[4]   Utilizing Machine Learning and Automated Performance Metrics to Evaluate Robot-Assisted Radical Prostatectomy Performance and Predict Outcomes [J].
Hung, Andrew J. ;
Chen, Jian ;
Che, Zhengping ;
Nilanon, Tanachat ;
Jarc, Anthony ;
Titus, Micha ;
Oh, Paul J. ;
Gill, Inderbir S. ;
Liu, Yan .
JOURNAL OF ENDOUROLOGY, 2018, 32 (05) :438-444
[5]   Predicting the Future - Big Data, Machine Learning, and Clinical Medicine [J].
Obermeyer, Ziad ;
Emanuel, Ezekiel J. .
NEW ENGLAND JOURNAL OF MEDICINE, 2016, 375 (13) :1216-1219
[6]  
Pedregosa F, 2011, J MACH LEARN RES, V12, P2825
[7]   Natural History of Early, Localized Prostate Cancer: A Final Report from Three Decades of Follow-up [J].
Popiolek, Marcin ;
Rider, Jennifer R. ;
Andren, Ove ;
Andersson, Sven-Olof ;
Holmberg, Lars ;
Adami, Hans-Olov ;
Johansson, Jan-Erik .
EUROPEAN UROLOGY, 2013, 63 (03) :428-435
[8]   Natural history of progression after PSA elevation following radical prostatectomy [J].
Pound, CR ;
Partin, AW ;
Eisenberger, MA ;
Chan, DW ;
Pearson, JD ;
Walsh, PC .
JAMA-JOURNAL OF THE AMERICAN MEDICAL ASSOCIATION, 1999, 281 (17) :1591-1597
[9]   The use of machine learning for the identification of peripheral artery disease and future mortality risk [J].
Ross, Elsie Gyang ;
Shah, Nigam H. ;
Dalman, Ronald L. ;
Nead, Kevin T. ;
Cooke, John P. ;
Leeper, Nicholas J. .
JOURNAL OF VASCULAR SURGERY, 2016, 64 (05) :1515-+
[10]   Lymphovascular invasion is a pathological feature of biologically aggressive disease in patients treated with radical prostatectomy [J].
Shariat, SF ;
Khoddami, SM ;
Saboorian, H ;
Koeneman, KS ;
Sagalowsky, AI ;
Cadeddu, JA ;
McConnell, JD ;
Holmes, MN ;
Roehrborn, CG .
JOURNAL OF UROLOGY, 2004, 171 (03) :1122-1127