The use of artificial neural networks to predict delayed discharge and readmission in enhanced recovery following laparoscopic colorectal cancer surgery.

摘要: Artificial neural networks (ANNs) can be used to develop predictive tools enable the clinical decision-making process. This study aimed investigate use of an ANN in predicting outcomes from enhanced recovery after colorectal cancer surgery. Data were obtained consecutive patients undergoing laparoscopic surgery within (ERAS) program between 2002 and 2009 a single center. The primary assessed delayed discharge readmission 30-day period. data analyzed using multilayered perceptron network (MLPNN), prediction created for each outcome. results compared with conventional statistical method logistic regression analysis. A total 275 included study. median length stay was 6 days (range 2–49 days) 67 (24.4 %) staying longer than 7 days. Thirty-four (12.5 %) readmitted 30 days. Important factors related failure compliance ERAS, particularly postoperative elements first 48 h. MLPNN had area under receiver operator characteristic curve (AUROC) 0.817, AUROC 0.807 tool developed Factors overall ERAS pathway receiving neoadjuvant treatment rectal cancer. 0.68. These may plausibly suggest that reliable outcome multifactorial intervention such as ERAS. Compliance reliably predict both following