AI Epidemiology: A linear regression modeling and structured machine learning protocol for the analysis of Alzheimer's Disease genomic data
By Reem Hamdan
My grandmother was my best friend. She laughed when I laughed; she cried when I cried. I wondered why she was suddenly forgetting things she had no problem remembering before. How could my grandma suddenly forget what country she was in? How could my grandma not even remember my name? … Alzheimer’s Disease is one of the most common neurodegenerative diseases. An emerging field of science is genomic neurology, which explores the genetic basis for neurodegenerative diseases. The purpose of this project was to identify genes associated with Alzheimer’s Disease. Genomic data of patients with dementia and healthy controls were collected from the Allen Institute Genome Browser. The following genes were analyzed: Apolipoprotein (APP), Complement Receptor 1 (CR1), Apolipoprotein E4 (APOE4), Clusterin (CLU). Linear regressions were run and showed the degree to which these four genes predicted AD diagnosis. Structured machine learning in the form of decision tree analyses identified which risk factors, including genes and at what levels, best predicted AD diagnosis. Significant results indicated that CR1, APOE4, and CLU were associated with the diagnosis of AD, but APP showed no significant association with the disease. The results of the study identify other potential pathways for new treatments and can help establish a more informed research process …