Handling Missing Data in Longitudinal Anthropometric Data Using Multiple Imputation Method

EasyChair Preprint 12591

Abstract

Diabetes mellitus, a prevalent and an ever-growing metabolic syndrome, has grown to be a widespread global health challenge. Given its tremendous occurrence, complexity, and the continuously rising healthcare fees related to it, there is an urgent need for research to advance our knowledge and remedy of this condition. This paper specializes in addressing missing data in a longitudinal clinical study targeted around diabetes. The study, published in October 2003, aimed to examine 14 different strategies for imputing missing data within a long-term study including older adults. Missing data encompassed an extensive range of variables, along with factors like despair, weight, cognitive functioning, and self-rated fitness, especially applicable to older adults. To address the problem of missing data, we carried out a radical exam of properly-established imputation strategies: K-Nearest Neighbors (KNN) and Multiple Imputation by using Chained Equations (MICE). Additionally, we tested the MICE technique, which iteratively imputes missing information while respecting temporal dependencies, resulting in the formation of multiple imputed datasets. Our study found out that the MICE imputation method outperformed KNN approach in terms of maintaining the mean and standard deviation. Also, the rigorous statistical evaluation confirmed the MICE approach's remarkable potential to preserve the nuanced temporal characteristics of the data. In conclusion, this study underscores the paramount significance of preserving temporal consistency in longitudinal research, specifically when coping with diabetes-related statistics.

Keyphrases: Anthropometric data, Data Imputation, K-Nearest Neighbors (KNN), Multiple Imputation by Chained Equations (MICE) Diabetes, missing data

@booklet{EasyChair:12591,
  author    = {Dhruv Varma and Chittaranjan S Yajnik and Aniket Thorave and Neha Sharma},
  title     = {Handling Missing Data in Longitudinal Anthropometric Data Using Multiple Imputation Method},
  howpublished = {EasyChair Preprint 12591},
  year      = {EasyChair, 2024}}