Authors :
Hong Zheng
Volume/Issue :
Volume 8 - 2023, Issue 3 - March
Google Scholar :
https://bit.ly/3TmGbDi
Scribd :
https://bit.ly/3zCarB6
DOI :
https://doi.org/10.5281/zenodo.7807289
Abstract :
Cancer has been a prevalent medical concern among many scientists. Despite modern treatments,
many patients still have a low recovery rate due to late diagnosis. MicroRNAs (miRNAs) are
endogenous non-coding functional RNAs that regulate gene expression by inhibiting/promoting
certain signaling pathways.5 They could be a potential indicator of cancer and can be detected from
miRNA screening of patients’ blood samples. This indicator could allow scientists to detect potential
cancer at an early stage and begin targeted therapy or early treatment. In this project, the aim to was
to improve the understanding of gene expression in relation to cancer, using machine learning to
identify key miRNAs with a high relatedness to cancer and find pathways connected to the relatively
novel field of miRNAs. We carried out data collection (data from Toray Industries, Japan, for three
cancers: lung, esophageal, and gastric cancer), model development, pathway analysis, and app
development along with generating figures such as variable importance and partial dependence plots.
The models averaged a very high AUC of 0.99. All of this allowed for a further understanding of
miRNAs in relation to gene expression. Crucial statistics such as specific threshold levels of miRNA
expression most indicative of cancer were generated. Overall, this project serves as a prototypical
model which has a high diagnostic accuracy for predicting cancer outcomes from patient miRNA
data, and an app displaying the gene targets and basic descriptions of said targets was produced.
Keywords :
miRNA, Machine Learning, Liquid Biopsy, Lung Cancer, Esophageal Cancer, Gastric Cancer.
Cancer has been a prevalent medical concern among many scientists. Despite modern treatments,
many patients still have a low recovery rate due to late diagnosis. MicroRNAs (miRNAs) are
endogenous non-coding functional RNAs that regulate gene expression by inhibiting/promoting
certain signaling pathways.5 They could be a potential indicator of cancer and can be detected from
miRNA screening of patients’ blood samples. This indicator could allow scientists to detect potential
cancer at an early stage and begin targeted therapy or early treatment. In this project, the aim to was
to improve the understanding of gene expression in relation to cancer, using machine learning to
identify key miRNAs with a high relatedness to cancer and find pathways connected to the relatively
novel field of miRNAs. We carried out data collection (data from Toray Industries, Japan, for three
cancers: lung, esophageal, and gastric cancer), model development, pathway analysis, and app
development along with generating figures such as variable importance and partial dependence plots.
The models averaged a very high AUC of 0.99. All of this allowed for a further understanding of
miRNAs in relation to gene expression. Crucial statistics such as specific threshold levels of miRNA
expression most indicative of cancer were generated. Overall, this project serves as a prototypical
model which has a high diagnostic accuracy for predicting cancer outcomes from patient miRNA
data, and an app displaying the gene targets and basic descriptions of said targets was produced.
Keywords :
miRNA, Machine Learning, Liquid Biopsy, Lung Cancer, Esophageal Cancer, Gastric Cancer.
