Logistic Regression Analysis of Functional Constipation Factors in the Elderly

Zeinab  Shiravi; Zahra  Abdollahzade; Saeed  Talebian; Fatemeh Alsadat  Mirzadeh

doi:10.18502/jmr.v19i4.19778

Zeinab Shiravi Department of Elderly Health, School of Rehabilitation, Tehran University of Medical Sciences, Tehran, Iran.
Zahra Abdollahzade Department of Physiotherapy, Faculty of Paramedical and Rehabilitation Sciences, Mashhad University of Medical Sciences, Mashhad, Iran.
Saeed Talebian Department of Physiotherapy, School of Rehabilitation, Tehran University of Medical Sciences, Tehran, Iran.
Fatemeh Alsadat Mirzadeh Department of Geriatric Medicine, Ziaeian Hospital, Tehran University of Medical Sciences, Tehran, Iran.

DOI: https://doi.org/10.18502/jmr.v19i4.19778

Keywords: Chronic constipation; Elderly; Supervised logistic machine learning

Abstract

Introduction: Machine learning software programs are of great interest in medical sciences for their diagnostic and therapeutic applications. Elderly individuals can greatly benefit from these technologies due to their physical limitations. This study aimed to develop and evaluate a supervised machine learning model for predicting functional constipation (FC) in older adults.

Materials and Methods: Specific software was developed in Excel as a logistic regression supervised machine learning model (LR-SML 402). This software was developed based on a secondary analysis of existing data, including articles and doctoral dissertations on elderly individuals with FC who underwent colorectal evaluations using advanced laboratory equipment. The correlation between labeled data and colorectal parameter outputs was measured using 480 datasets from published sources and research laboratories. Strong correlations were observed between variables, such as age, body mass index, Wexner’s questionnaire scores, and FC parameters.

Results: To validate the performance of LR-SML 402, the results were compared with those of a neural network model using SPSS software. The Excel-based model demonstrated strong performance in terms of sensitivity, specificity, and area under the curve.

Conclusion: The LR-SML 402 model shows that supervised machine learning using logistic regression may yield meaningful clinical predictions of FC indicators in the elderly. This approach can reduce diagnostic time and cost.