(Georgia Institute of Technology, 2020-05)
Nakirikanti, Anudeep Sai
The impact of Alzheimer’s Disease (AD) on today’s society and healthcare is unprecedented. As a larger portion of today’s population enters an age for which AD becomes a health concern, there is growing support among health practitioners to prevent the disease’s progression and development. Early identification of the disease may serve as a critical step towards combating the disease, allowing earlier interventions in the disease process to foster healthy aging. The focus of such interventions includes alleviating risk factors of AD, two of which include cortisol and degeneration in the basal forebrain. Importantly, increased levels of cortisol and reduced volume in the basal forebrain are attributed to higher risks of AD. In the present study, we make use of machine learning and the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database to characterize individuals with AD by using data from cortisol levels and basal forebrain degeneration. This allowed us to test whether cortisol and basal forebrain degeneration were predictively valuable for AD diagnosis. Our data partially supported our prediction—the machine learning classifier yielded significantly above chance classification accuracy for basal forebrain degeneration, but the classification accuracy for cortisol was not significantly above chance. Consequently, our results indicate that basal forebrain degeneration might serve as a diagnostically useful biomarker for AD, while cortisol’s role in AD characterization necessitates further investigation.