Lifestyle and Dietary Factors Causing Microsatellite Instability Gastric Cancer Detected using Ensemble Modeling

Authors

  • S. K. Brindha Department of Computer Engineering, Mizoram University, Aizawl - 796004, Mizoram, India
  • P. Charkarborthy Deparment of Biotechnology, Mizoram University, Aizawl – 796004, Mizoram, India
  • S. Chenkual Department of Surgery, Civil Hospital, Aizawl – 796001, Mizoram, India
  • J. Zohmingthanga Department of Pathology, Civil Hospital, Aizawl – 796001, Mizoram, India
  • J. L. Pautu Department of Oncology, Mizoram State Cancer Institute, Zemabawk, Aizawl, Mizoram, India
  • P. Nath Department of Computer Engineering, Mizoram University, Aizawl - 796004, Mizoram, India
  • A. Maitra National Institute of Biomedical Genomics, Kalyani - 741251, West Bengal, India
  • N. S. Kumar Deparment of Biotechnology, Mizoram University, Aizawl – 796004, Mizoram, India
  • L. Hmingliana Department of Computer Engineering, Mizoram University, Aizawl - 796004, Mizoram, India

DOI:

https://doi.org/10.3329/jsr.v14i3.58331

Abstract

Aim of this study is to identify diet and lifestyle patterns that cause microsatellite instability gastric cancer (MSI-GC) using supervised machine learning algorithms. There were 142 genetic variants acquired via targeted resequencing of 60 biomarker genes from gastric tumor samples and tabulated with respect to MSI status, diet and lifestyle characteristics. Four classifiers (logistic regression, random forest, logistic regression, multilayer perceptron) were used to train the data and evaluated based on their classification efficiency. Data analysis revealed features extracted using ridge regression: extra salt, smoked food, smokeless tobacco products (Khaini /sadha), alcohol and betel nut leaf with lime (khuva) were the core factors for causing MSI-GC. The extracted features were exploited using random forest and multilayer perceptron classifiers, which has produced accuracy, precision, recall, F1 score, and Receiver operating characteristics (ROC) curve of 96 %. The brier score was 0.04 and Matthews correlation coefficient (MCC) was +0.91. Linear regression results revealed khuva was main driving factor and extra salt, smoked food, khaini/sadha and alcohol were confounding factors to cause MSI-GC. This is a first-time report that integrates mutation and diet-lifestyle data using machine learning, to precisely identify the driving and confounding factors for causing MSI-GC.

Downloads

Download data is not yet available.
Abstract
60
pdf
70

Downloads

Published

2022-09-01

How to Cite

K. Brindha, S., Charkarborthy, P., Chenkual, S., Zohmingthanga, J., Pautu, J. L., Nath, P., Maitra, A., Kumar, N. S. ., & Hmingliana, L. (2022). Lifestyle and Dietary Factors Causing Microsatellite Instability Gastric Cancer Detected using Ensemble Modeling. Journal of Scientific Research, 14(3), 943–958. https://doi.org/10.3329/jsr.v14i3.58331

Issue

Section

Section B: Chemical and Biological Sciences