Abstract
Alzheimer’s disease (AD) is a neurodegenerative disease with typical clinical symptoms in the form of progressive cognitive impairment and memory loss. To facilitate early diagnosis of AD and a greater understanding of the mechanisms underlying its clinical expression, the use of biomarkers is necessary. Furthermore, it is believed that biomarkers provide a more objective and accessible means of diagnosis. Currently, established biomarkers include neuroimaging markers, such as those based on positron emission tomography (PET), and biochemical markers such as cerebrospinal fluid (CSF) markers. However, neuroimaging is expensive and may not be widely available and CSF testing is invasive. Blood-based biomarkers offer the potential for the development of minimally invasive, low-cost and time-efficient methods for AD detection to complement CSF and neuroimaging. In this work, a data-driven approach, machine learning in particular, was exploited to identify blood-based biomarker panels consisting of a few markers (as no single marker provides sufficient performance) that may serve as screening tools in a multi-stage diagnostic procedure. Novel contributions were made in biomarker discovery, including identification of novel panels as well as panel selection procedures that emphasize performance and robustness. Identified biomarker panels have remarkable classification performance at discriminating between Alzheimer’s dementia as well as mild cognitive impairment subjects and normal controls. Another set of identified blood-based biomarkers could classify individuals with abnormal/normal levels of CSF amyloid β42, which is one of the key early markers of AD. Furthermore, a novel software prototype was developed to demonstrate the possible clinical use of identified biomarker panels. A significance of this work is its potential contribution to the development of rapid testing and cost-effective point of care devices to facilitate AD diagnosis.
Document Type
Thesis
Publication Date
2022-01-01
DOI
10.24382/1259
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Eke, C. S. (2022) MACHINE LEARNING-BASED EXPLORATION OF BLOOD-BASED BIOMARKERS FOR ALZHEIMER’S DISEASE DIAGNOSIS. Thesis. University of Plymouth. Available at: https://doi.org/10.24382/1259
