Abstract
Alzheimer's Disease is a gradual, irreversible brain disease that deteriorates a patient's memory, cognitive functions and shrinks the brain's size, eventually leading to death. Based on recent research, it is found that AD is the third leading cause of death. Presently there is no available medication for the treatment of AD. Though, diagnosis of AD at early onset may delay the progression of the disease and thus aid in improving the subject's well-being. Early detection and classification of divergent phases of AD using EHR (Electronic Health Record) and ML (Machine Learning) algorithms can prove to be a productive approach as AD evolves with time and thus patients at distinct stages need to be treated differently. Hence, classification of different stages is crucial for the realization of purpose that it can improve patient's quality of life as treatment of symptoms can be performed accordingly. The use of contemporary computing technology and resources is becoming a boon to new trends in healthcare and diagnosis. EHR is setting a gauge to record patient's data electronically through the replacement of conventional methods that comprise the collection of data in paper-based form. ML with AI techniques can be applied to EHR to provide an accurate and comprehensive diagnosis to improve the quality and productivity of healthcare. In this article, four diverse machine learning algorithms are applied on ADNI-Longitudinal data for the classification of five different stages of AD and thus identifying the most relevant biomarkers and features that can lead to reliable and effective detection and diagnosis of AD. Wherein, RF (Random Forest) exhibits the highest accuracy of 99.8 % followed by ANN (Artificial Neural Network). In this study, we utilized the TADPOLE (The Alzheimer's Disease Prediction of Longitudinal Evolution) grand challenge data generated from ADNI (Alzheimer's Disease Neuroimaging Initiative). The proposed study provides a promising solution for the management of AD.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have