Analysis of the characteristics and influencing factors of the duration of day surgery for children in a tertiary hospital
某三甲医院儿童日间手术时长特点及其影响因素分析

摘要

目的 回顾性分析某三甲医院儿童日间手术时长分布特征，采用多层感知器（MLP）模型预测手术时长并探讨其关键影响因素，为手术室智能化排程提供循证依据。方法 采用回顾性研究设计，收集2020年1月至2024年12月某三甲儿童医院73 533例日间手术病例资料。手术时长为患者进入至离开手术室的时间间隔。应用描述性统计分析患者年龄、疾病谱构成及手术时长分布特征，并通过MLP对手术时长进行预测，评估各影响因素的重要性。结果 73 533例手术以3～6岁学龄前儿童为主（37.40%），疾病谱前三位为鼾症（31.80%）、腹股沟斜疝（13.78%）及鞘膜积液（11.68%）。手术时长呈右偏态分布，92.84%的手术可在1.5 h内完成。MLP模型预测的相对误差为0.354，预测准确率为64.6%。影响因素重要性分析显示，术前诊断是预测能力最强的变量（正态化重要性19.2%），其后依次为手术医生（11.6%）、院区（11.2%）和巡回护士（8.4%）。结论 儿童日间手术时长分布具有明显偏态特征，术前诊断、手术医生、院区资源配置及巡回护士协作效率是影响手术时长的主要因素，基于多层感知器神经网络为手术排程由经验驱动向数据驱动转型提供了可行路径。

Abstract

Objective To analyze the distribution characteristics of the duration of pediatric day surgeries in a tertiary hospital, predict the duration of surgeries using a multi-layer perceptron (MLP) neural network, and explore the key influencing factors, providing evidence-based support for the intelligent scheduling of operating rooms.
Methods A retrospective study design was adopted, collecting data from 73,533 pediatric day surgeries conducted in a tertiary children’s hospital from January 2020 to December 2024. The duration of surgery was defined as the time interval from the patient’s entry to the operating room to their departure. Descriptive statistical analysis was applied to examine the age of patients, the composition of disease spectra, and the distribution characteristics of surgery duration. A multi-layer perceptron neural network prediction model was used to predict the duration of surgeries, and the importance of various influencing factors was evaluated.
Results Among the 73,533 surgeries, the majority were performed on preschool children aged 3 to 6 years (37.40%), with the top three disease spectra being snoring (31.80%), inguinal hernia (13.78%), and hydrocele (11.68%). The duration of surgeries showed a right-skewed distribution, with 92.84% of surgeries completed within 1.5 hours. The multi-layer perceptron neural network prediction model had an accuracy rate of 64.6% in predicting surgery duration, and the ranking of the importance of influencing factors was: preoperative diagnosis (importance19.2%), surgeon (11.6%), hospital area (11.2%), and circulating nurse (8.4%).
Conclusion   The duration of pediatric day surgeries has a distinct right-skewed distribution. Preoperative diagnosis, surgeon, hospital area resource allocation, and the efficiency of circulating nurse collaboration are the main factors influencing the duration of surgeries. The multi-layer perceptron neural network provides a feasible path for transforming the scheduling of surgeries from experience-driven to data-driven.