Biomarkers Identification for Parkinson’s Disease using Machine Learning

Alzheimer’s disease or Parkinson’s disease is a neurodegenerative disease that starts developing at an early age however symptoms are demonstrated quite late.Parkinson’s disease is a neuro degenerative disease with a brain neuron loss causing shaking, stiffness and difficulty in motor movements. These symptoms worsen over time as the disease progresses. Computational biology and bioinformatics domains have witnessed advancement with the development of powerful methods to collect, process and analyze health informatics data such as molecular-genomic, proteomic, transcriptomic data revealing hidden patterns. Several machine learning techniques are widely used to mine the voluminous data with large feature space. Biomarkers identification process using machine learning helps to detect the minute changes that might have occurred at the molecular level. This paper presents preliminary work of identifying biomarkers using machine learning for Parkinson’s disease through differentially expressed genes. The dataset GSE54536 - Gene Expression Omnibus is obtained from Gene Expression Omnibus repository and pre-processed. This pre-processed data is used to construct a linear model indicating disease states. Then least square regression along with statistical tests such as t-test and fold change are used to find differentially expressed genes. Total 8 differentially expressed Parkinson’s disease genes-TLR10, OSBPL10, FCRLA, MS4A1,FOS, FOSB,EGR1,SLC11A2 are recognized.