Volume 15, Issue 1 (Winter 2026)
aumj 2026, 15(1): 0-0 |
Back to browse issues page
Ethics code: IR.FUMS.REC.1404.174
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Rafatmagham S, Jafari N, Dehghan A, Roshanzamir M. Evaluation and comparison of machine learning models in predicting spirometric indices based on individual characteristics in Fasa county. aumj 2026; 15 (1) : 1
URL: http://aums.abzums.ac.ir/article-1-1983-en.html
URL: http://aums.abzums.ac.ir/article-1-1983-en.html
1- Internal Medicine Resident, Department of Internal Medicine, School of Medicine, Fasa University of Medical Sciences, Fasa, Iran
2- Assistant Professor of Pulmonology, Department of Internal Medicine, School of Medicine, Fasa University of Medical Sciences, Fasa, Iran ,nedajafari.med@gmail.com
3- Associate Professor of Epidemiology and Biostatistics, School of Health, Fasa University of Medical Sciences, Fasa, Iran
4- Assistant Professor of Computer Engineering, Department of Computer Engineering, School of Engineering, Fasa University, Fasa, Iran
2- Assistant Professor of Pulmonology, Department of Internal Medicine, School of Medicine, Fasa University of Medical Sciences, Fasa, Iran ,
3- Associate Professor of Epidemiology and Biostatistics, School of Health, Fasa University of Medical Sciences, Fasa, Iran
4- Assistant Professor of Computer Engineering, Department of Computer Engineering, School of Engineering, Fasa University, Fasa, Iran
Abstract: (30 Views)
Introduction: Chronic respiratory diseases impose a significant burden on public health, and accurate prediction of spirometric indices based on individual characteristics can be effective in the diagnosis and management of these conditions. The aim of this study was to evaluate and compare the performance of machine learning models in predicting spirometric indices based on individual features in the population of Fasa County.
Methods: This cross-sectional observational study was conducted on 5,450 individuals who visited pulmonary clinics between the spring of 2024 and the autumn of 2025. After excluding incomplete data, 4,615 eligible participants were included in the final analysis. Individual characteristics included age, sex, height, weight, and smoking status. Spirometric indices included FVC, FEV1, FEV1/FVC, FEF25–75, and PEF. Five machine learning algorithms—Linear Regression, K-Nearest Neighbors, Support Vector Machine, Random Forest, and Gradient Boosting—were evaluated using mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination (R²). The relative importance of individual features was analyzed using the feature_importances output of the Gradient Boosting model. In this study, all ethical principles were observed, and the ethics code IR.FUMS.REC.1404.174 was obtained.
Results: The Gradient Boosting model demonstrated the lowest error rates and the highest R² values across all spirometric indices, indicating superior performance compared to the other models. Age, height, and weight had the greatest influence on the predictions, whereas sex and smoking status had a relatively lower impact.
Conclusion: The Gradient Boosting model is capable of identifying nonlinear relationships and complex interactions between individual characteristics and spirometric indices. This study indicates that the use of machine learning can be beneficial for developing localized predictive models and improving the interpretation of spirometry results in native populations.
Methods: This cross-sectional observational study was conducted on 5,450 individuals who visited pulmonary clinics between the spring of 2024 and the autumn of 2025. After excluding incomplete data, 4,615 eligible participants were included in the final analysis. Individual characteristics included age, sex, height, weight, and smoking status. Spirometric indices included FVC, FEV1, FEV1/FVC, FEF25–75, and PEF. Five machine learning algorithms—Linear Regression, K-Nearest Neighbors, Support Vector Machine, Random Forest, and Gradient Boosting—were evaluated using mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination (R²). The relative importance of individual features was analyzed using the feature_importances output of the Gradient Boosting model. In this study, all ethical principles were observed, and the ethics code IR.FUMS.REC.1404.174 was obtained.
Results: The Gradient Boosting model demonstrated the lowest error rates and the highest R² values across all spirometric indices, indicating superior performance compared to the other models. Age, height, and weight had the greatest influence on the predictions, whereas sex and smoking status had a relatively lower impact.
Conclusion: The Gradient Boosting model is capable of identifying nonlinear relationships and complex interactions between individual characteristics and spirometric indices. This study indicates that the use of machine learning can be beneficial for developing localized predictive models and improving the interpretation of spirometry results in native populations.
Article number: 1
Type of Study: Research |
Subject:
Special
Received: 2025/12/06 | Accepted: 2026/01/04 | Published: 2026/02/21
Received: 2025/12/06 | Accepted: 2026/01/04 | Published: 2026/02/21
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |