Modeling and variable selection of healthcare trip behaviours using statistical learning techniques

IF 6.3 2区工程技术 Q1 ECONOMICS

Transport Policy Pub Date : 2025-05-09 DOI:10.1016/j.tranpol.2025.05.008

Cagdas Kara , Abdulkadir Ozden , Arzu Altin Yavuz , Safak Bilgic

{"title":"Modeling and variable selection of healthcare trip behaviours using statistical learning techniques","authors":"Cagdas Kara , Abdulkadir Ozden , Arzu Altin Yavuz , Safak Bilgic","doi":"10.1016/j.tranpol.2025.05.008","DOIUrl":null,"url":null,"abstract":"<div><div>In an increasingly complex and fast-paced world, understanding healthcare-related travel behavior has become a critical challenge in transportation planning. Traditional models, including the Four-Step Transportation Model (FSTM), require further refinements to better capture evolving travel patterns, particularly the growing share of home-based other trips, which include shopping, leisure, and healthcare-related journeys. Among these, healthcare trips require special attention due to the increasing proportion of the aging population and the essential nature of medical accessibility.</div><div>This study aims to identify key variables influencing healthcare-related travel behaviors using variable selection techniques, specifically Least Absolute Shrinkage and Selection Operator (Lasso) and Elastic Net (ENet). The analysis is based on zonal-level aggregated data derived from household-based hospital travel surveys conducted in Eskişehir City. The model incorporates regional socioeconomic variables, including household characteristics, vehicle ownership rates, employment ratios, and age group distributions. Model performance is evaluated using Mean Squared Error (MSE) and Mean Absolute Error (MAE) as key success criteria.</div><div>The results indicate that the ENet model achieved the lowest Mean Squared Error (MSE), reducing the error by approximately 37 % compared to the OLS model, while the Lasso model yielded the lowest Mean Absolute Error (MAE), reflecting a 38 % improvement. Both methods effectively performed variable selection, retaining 10 out of 17 predictors in the final model. Significant variables positively associated with healthcare travel frequency include the proportions of individuals aged 30–49, 50–64, and over 65, as well as family density. These results suggest that household accompaniment patterns and age-related healthcare needs increase the frequency of such trips. In contrast, negative associations were observed between healthcare travel frequency and the share of the 6–17 age group, employment ratio, average number of cars per family, and the number of healthcare centers.</div></div>","PeriodicalId":48378,"journal":{"name":"Transport Policy","volume":"170 ","pages":"Pages 12-23"},"PeriodicalIF":6.3000,"publicationDate":"2025-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Transport Policy","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0967070X25001878","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ECONOMICS","Score":null,"Total":0}

引用次数: 0

Abstract

In an increasingly complex and fast-paced world, understanding healthcare-related travel behavior has become a critical challenge in transportation planning. Traditional models, including the Four-Step Transportation Model (FSTM), require further refinements to better capture evolving travel patterns, particularly the growing share of home-based other trips, which include shopping, leisure, and healthcare-related journeys. Among these, healthcare trips require special attention due to the increasing proportion of the aging population and the essential nature of medical accessibility.

This study aims to identify key variables influencing healthcare-related travel behaviors using variable selection techniques, specifically Least Absolute Shrinkage and Selection Operator (Lasso) and Elastic Net (ENet). The analysis is based on zonal-level aggregated data derived from household-based hospital travel surveys conducted in Eskişehir City. The model incorporates regional socioeconomic variables, including household characteristics, vehicle ownership rates, employment ratios, and age group distributions. Model performance is evaluated using Mean Squared Error (MSE) and Mean Absolute Error (MAE) as key success criteria.

The results indicate that the ENet model achieved the lowest Mean Squared Error (MSE), reducing the error by approximately 37 % compared to the OLS model, while the Lasso model yielded the lowest Mean Absolute Error (MAE), reflecting a 38 % improvement. Both methods effectively performed variable selection, retaining 10 out of 17 predictors in the final model. Significant variables positively associated with healthcare travel frequency include the proportions of individuals aged 30–49, 50–64, and over 65, as well as family density. These results suggest that household accompaniment patterns and age-related healthcare needs increase the frequency of such trips. In contrast, negative associations were observed between healthcare travel frequency and the share of the 6–17 age group, employment ratio, average number of cars per family, and the number of healthcare centers.

查看原文本刊更多论文

使用统计学习技术的医疗旅行行为建模和变量选择

在一个日益复杂和快节奏的世界中，了解与医疗保健相关的旅行行为已成为交通规划中的一个关键挑战。包括四步交通模型（FSTM）在内的传统模型需要进一步改进，以更好地捕捉不断变化的旅行模式，特别是以家庭为基础的其他旅行，包括购物、休闲和医疗相关的旅行。其中，由于老龄化人口比例的增加和医疗可及性的本质，医疗旅行需要特别关注。本研究旨在利用变量选择技术，特别是最小绝对收缩和选择算子（Lasso）和弹性网（ENet），确定影响医疗保健相关旅行行为的关键变量。该分析基于在eski希尔市进行的以家庭为基础的医院旅行调查得出的区域一级汇总数据。该模型结合了区域社会经济变量，包括家庭特征、车辆拥有率、就业率和年龄组分布。使用均方误差（MSE）和平均绝对误差（MAE）作为关键成功标准来评估模型性能。结果表明，ENet模型实现了最低的均方误差（MSE），与OLS模型相比，误差减少了约37%，而Lasso模型产生了最低的平均绝对误差（MAE），反映了38%的改进。两种方法都有效地进行了变量选择，在最终模型中保留了17个预测因子中的10个。与医疗差旅频率呈正相关的显著变量包括30-49岁、50-64岁和65岁以上的个体比例，以及家庭密度。这些结果表明，家庭陪伴模式和与年龄相关的医疗保健需求增加了此类旅行的频率。相反，医疗旅行频率与6-17岁年龄组的比例、就业率、家庭平均汽车数量和医疗中心数量呈负相关。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Transport Policy Multiple-

CiteScore

12.10

自引率

10.30%

发文量

282

期刊介绍： Transport Policy is an international journal aimed at bridging the gap between theory and practice in transport. Its subject areas reflect the concerns of policymakers in government, industry, voluntary organisations and the public at large, providing independent, original and rigorous analysis to understand how policy decisions have been taken, monitor their effects, and suggest how they may be improved. The journal treats the transport sector comprehensively, and in the context of other sectors including energy, housing, industry and planning. All modes are covered: land, sea and air; road and rail; public and private; motorised and non-motorised; passenger and freight.