Machine Learning-Based Clustering and Classification of Obesity Associated with Elevated Screen Time

Authors

  • Muhammad Ammar Genetic and Molecular Epidemiology Research Group, Department of Bioinformatics and Biosciences, Capital University of Science and Technology, Islamabad, Pakistan
  • Abdur Rauf Genetic and Molecular Epidemiology Research Group, Department of Bioinformatics and Biosciences, Capital University of Science and Technology, Islamabad, Pakistan & National University of Science and Technology, Islamabad, Pakistan
  • Nasreen Noor Genetic and Molecular Epidemiology Research Group, Department of Bioinformatics and Biosciences, Capital University of Science and Technology, Islamabad, Pakistan
  • Mahnoor Azhar Genetic and Molecular Epidemiology Research Group, Department of Bioinformatics and Biosciences, Capital University of Science and Technology, Islamabad, Pakistan.
  • Syeda Marriam Bakhtiar Capital University of Science And Technology, Islamabad, Pakistan

DOI:

https://doi.org/10.52700/pjbb.v5i2.241

Keywords:

Machine Learning, Obesity, Screen Time, Public Health, Decision Tree,, K–Nearest Neighbor, Random Forest, K Mean Clustering

Abstract

Obesity is a complex chronic disease characterized by excessive fat deposits that can undermine health. Obesity can lead to increased risk of type 2 diabetes and cardiovascular diseases; it can affect bone health and reproduction; and it also increases the risk of certain cancers. WHO in 2022 reported 2.5 billion people are suffering from obesity. It is a multifactorial disease due to obesogenic environments, psychosocial factors, and genetic predispositions. Lacks of physical activity, sedentary lifestyle, mode of transportation, and screen time have also shown significant impact on it. Because of the strong link between elevated screen time and obesity, effective and advanced studies are required to address and understand this issue. Classification, prediction, and population-level analysis help us to build future prevention strategies to combat. Machine learning algorithms proved their remarkable results in the latest studies for prediction analysis. We have used data from the open source library UCI and applied three machine learning approaches for classification, prediction, and cluster analysis: K-mean clustering, the K-nearest neighbor algorithm (KNN), and a decision tree with a random forest. Among them all, the decision tree with random forest has shown the best results with an accuracy of 95.64% and a relative absolute error of 17.96%. All approaches proved strong associations between increased screen time and obesity level.

Published

2024-10-30

How to Cite

Muhammad Ammar, Abdur Rauf, Nasreen Noor, Mahnoor Azhar, & Syeda Marriam Bakhtiar. (2024). Machine Learning-Based Clustering and Classification of Obesity Associated with Elevated Screen Time. Pakistan Journal of Biochemistry and Biotechnology, 5(2), 21-29. https://doi.org/10.52700/pjbb.v5i2.241