Early Postoperative Prediction of Complications and Readmission After Colorectal Cancer Surgery Using an Artificial Neural Network.

Autor:	Agnes A; Department of General Surgery, Fondazione Policlinico Universitario Agostino Gemelli IRCCS, Rome, Italy., Nguyen ST; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., Konishi T; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., Peacock O; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., Bednarski BK; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., You YN; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., Messick CA; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., Tillman MM; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., Skibber JM; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., Chang GJ; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas., Uppal A; Department of Colon and Rectal Surgery, University of Texas MD Anderson Cancer Center, Houston, Texas.
Jazyk:	angličtina
Zdroj:	Diseases of the colon and rectum [Dis Colon Rectum] 2024 Oct 01; Vol. 67 (10), pp. 1341-1352. Date of Electronic Publication: 2024 Jul 03.
DOI:	10.1097/DCR.0000000000003253
Abstrakt:	Background: Early predictors of postoperative complications can risk-stratify patients undergoing colorectal cancer surgery. However, conventional regression models have limited power to identify complex nonlinear relationships among a large set of variables. We developed artificial neural network models to optimize the prediction of major postoperative complications and risk of readmission in patients undergoing colorectal cancer surgery. Objective: This study aimed to develop an artificial neural network model to predict postoperative complications using postoperative laboratory values and compare the accuracy of models to standard regression methods. Design: This retrospective study included patients who underwent elective colorectal cancer resection between January 1, 2016, and July 31, 2021. Clinical data, cancer stage, and laboratory data from postoperative days 1 to 3 were collected. Complications and readmission risk models were created using multivariable logistic regression and single-layer neural networks. Setting: National Cancer Institute-Designated Comprehensive Cancer Center. Patients: Adult patients with colorectal cancer. Main Outcome Measures: The accuracy of predicting postoperative major complications, readmissions, and anastomotic leaks using the area under the receiver operating characteristic curve. Results: Neural networks had larger areas under the curve for predicting major complications compared to regression models (neural network 0.811; regression model 0.724, p < 0.001). Neural networks also showed an advantage in predicting anastomotic leak ( p = 0.036) and readmission using postoperative day 1 to 2 values ( p = 0.014). Limitations: Single-center, retrospective design limited to cancer operations. Conclusions: In this study, we generated a set of models for the early prediction of complications after colorectal surgery. The neural network models provided greater discrimination than the models based on traditional logistic regression. These models may allow for early detection of postoperative complications as early as postoperative day 2. See the Video Abstract . Prediccin Post Operatoria Temprana De Complicaciones Y Reingreso Despus De La Ciruga De Cncer Colorrectal Mediante Una Red Neuronal Artificial: ANTECEDENTES:Los predictores tempranos de complicaciones postoperatorias pueden estratificar el riesgo de los pacientes sometidos a cirugía de cáncer colorrectal. Sin embargo, los modelos de regresión convencionales tienen un poder limitado para identificar relaciones no lineales complejas entre un gran conjunto de variables. Desarrollamos modelos de redes neuronales artificiales para optimizar la predicción de complicaciones postoperatorias importantes y riesgo de reingreso en pacientes sometidos a cirugía de cáncer colorrectal.OBJETIVO:El objetivo de este estudio fue desarrollar un modelo de red neuronal artificial para predecir complicaciones postoperatorias utilizando valores de laboratorio postoperatorios y comparar la precisión de estos modelos con los métodos de regresión estándar.DISEÑO:Este estudio retrospectivo incluyó a pacientes que se sometieron a resección electiva de cáncer colorrectal entre el 1 de enero de 2016 y el 31 de julio de 2021. Se recopilaron datos clínicos, estadio del cáncer y datos de laboratorio del día 1 al 3 posoperatorio. Se crearon modelos de complicaciones y riesgo de reingreso mediante regresión logística multivariable y redes neuronales de una sola capa.AJUSTE:Instituto Nacional del Cáncer designado Centro Oncológico Integral.PACIENTES:Pacientes adultos con cáncer colorrectal.PRINCIPALES MEDIDAS DE RESULTADO:Precisión de la predicción de complicaciones mayores postoperatorias, reingreso y fuga anastomótica utilizando el área bajo la curva característica operativa del receptor.RESULTADOS:Las redes neuronales tuvieron áreas bajo la curva más grandes para predecir complicaciones importantes en comparación con los modelos de regresión (red neuronal 0,811; modelo de regresión 0,724, p < 0,001). Las redes neuronales también mostraron una ventaja en la predicción de la fuga anastomótica ( p = 0,036) y el reingreso utilizando los valores del día 1-2 postoperatorio ( p = 0,014).LIMITACIONES:Diseño retrospectivo de un solo centro limitado a operaciones de cáncer.CONCLUSIONES:En este estudio, generamos un conjunto de modelos para la predicción temprana de complicaciones después de la cirugía colorrectal. Los modelos de redes neuronales proporcionaron una mayor discriminación que los modelos basados en regresión logística tradicional. Estos modelos pueden permitir la detección temprana de complicaciones posoperatorias tan pronto como el segundo día posoperatorio. (Traducción-Dr. Mauricio Santamaria ). (Copyright © The ASCRS 2024.)
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu