Developing a novel sodium-based calorie model with non-parametric regression: A case study with non-normally distributed data
Keywords:
Non-parametric regression model; Kendall-Theil Sen Siegel; Calories; SodiumAbstract
Background This paper introduces a comprehensive approach to modelling calories and sodium intake using non-parametric regression, bootstrap resampling, and data splitting into training and testing sets, aiming to enhance precision in understanding their intricate relationship and inform evidence-based strategies for optimal health promotion. Objective This study seeks to build a non-parametric regression model that links Sodium levels with Calorie levels, to enhance prediction accuracy for Calorie levels in the analyzed patient population by employing diagnostic capabilities. Materials and Methods In cases where the linear regression assumption is not met, the resulting model may exhibit biased estimates. To address this limitation, a nonparametric regression model is introduced in this study, incorporating an enhanced bootstrap method to optimize the model estimation. Specifically, the study utilizes the robust estimator known as the Kendall-Theil Sen Siegel slope to assess the regression line’s slope, mitigating the impact of outliers or extreme values. This method facilitates a flexible exploration of the relationship between variables, eliminating the need for specific parametric assumptions. The dataset was divided into training and testing groups using this method. The training dataset will be utilized for model development, whereas the testing dataset will be employed to validate the model. Result The statistical analysis conducted with R revealed that the nonparametric regression approach demonstrated enhanced predictive capabilities, particularly in cases where the data did not satisfy the normality assumption. The resulting model is as follows: Calories = −182.2940b+1.8361s(Sodium). The R-squared values for the training and testing sets are 74.2% and 70.0%, respectively. This suggests that the developed methodology has exhibited an exceptional level of performance. Conclusion The conclusion of the study underscores the higher performance of the hybrid model approach utilized.
Bangladesh Journal of Medical Science Vol. 25 No. 04 October’26 Page: 1213-1219
16
16
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Wan Muhamad Amir W Ahmad, Hazik Bin Shahzad, Farah Muna Mohamad Ghazali, Mohamad Nasarudin Adnan, Nor Azlida Aleng, Nor Farid Mohd Noor
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish in the Bangladesh Journal of Medical Science agree to the following terms that:
- Authors retain copyright and grant Bangladesh Journal of Medical Science the right of first publication of the work.
Articles in Bangladesh Journal of Medical Science are licensed under a Creative Commons Attribution 4.0 International License CC BY-4.0.This license permits use, distribution and reproduction in any medium, provided the original work is properly cited.- Authors are able to enter into separate, additional contractual arrangements for the distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted to post their work online (e.g., in institutional repositories or on their website) as it can lead to productive exchanges, as well as greater citation of published work.
