Volume 28, Issue 148, June 2024

Enhancing ordinal regression analysis with bootstrap techniques for a comprehensive study in dental health sciences

Muhammad Amirul Mat Lazin¹, Wan Mohd Nazlee Wan Zainon², Arsalan Humayun³, Ashwini M Madawana⁴, Akram Hassan⁵, Mohamad Arif Awang Nawi^6♦

¹School of Dental Sciences, Health Campus, Universiti Sains Malaysia, Health Campus,16150 Kubang Kerian, Kelantan, Malaysia.
²Family Medicine Department, Hospital Universiti Sains Malaysia, 16150, Kubang Kerian, Kelantan, Malaysia
³School of Dental Sciences, Health Campus, Universiti Sains Malaysia, Health Campus,16150 Kubang Kerian, Kelantan, Malaysia.
⁴Paediatric Dentistry Unit, School of Dental Sciences, Universiti Sains Malaysia, Health Campus, 16150 Kubang Kerian, Kelantan, Malaysia.
⁵Periodontology Unit, School of Dental Sciences, Universiti Sains Malaysia, Health Campus, 16150 Kubang Kerian, Kelantan, Malaysia.
⁶Biostatistics Unit, School of Dental Sciences, Health Campus, Universiti Sains Malaysia, Health Campus,16150 Kubang Kerian, Kelantan, Malaysia.

^♦Corresponding author
Mohamad Arif Awang Nawi, Biostatistics Unit, School of Dental Sciences, Health Campus, Universiti Sains Malaysia, Health Campus,16150 Kubang Kerian, Kelantan, Malaysia

ABSTRACT

The study proposes a novel combination of bootstrap and ordinal regression techniques in the R programming language to address the problem of analyzing small datasets in the health research domain, especially in medical imaging. Ordinal regression models are essential for assessing health-related quality-of-life measures since the responses are naturally ordered. However, their limitations include the need to analyze small data quality-of-life sets and the assumption of proportional odds, which can result in errors. Our methodology incorporates bootstrap techniques to improve prediction accuracy and dependability, particularly in limited datasets, improving upon the conventional ordinal regression method. Using sophisticated statistical methods, the study finds significant predictors of tooth wear, such as age and alcohol consumption, based on data regarding the severity of tooth wear among patients at Hospital Universiti Sains Malaysia. The results show that, although at the expense of higher model complexity, this novel method produces more accurate and dependable estimates than those produced by conventional ordinal regression models, as shown by improved log-likelihood values and lower standard errors. The improved method works exceptionally well when dealing with small sample sizes, a common challenge in medical research. It provides a vital instrument for accurate and thorough statistical analysis. This work promotes ordinal regression in the health sciences and emphasizes the importance of including trustworthy predictors in modeling endeavors to effectively evaluate tooth wear's severity. Ultimately, it emphasizes how advanced statistical techniques can enhance research findings and guide focused dental health measures.

Keywords: Ordinal Regression, Bootstrap Methods, Small Datasets, Health Research Domain, Tooth Wear Severity

Medical Science, 2024, 28, e53ms3358

PDF

DOI: https://doi.org/10.54905/disssi.v28i148.e53ms3358

Published: 07 June 2024