An Assorted Ensemble Method for Prediction of Terminal Care Preference by Caregivers of Alzheimer’s Victims

Healthcare is envisioning new and optimistic opportunities woven around the applications of machine learning that make it more effective and efficient. Machine learning is used to predict the chronic disease trends and its rate of progression which would help the healthcare stakeholders to strategize their treatment and healthcare provision. A 2-layered assorted ensemble model has been proposed to improve the prediction of terminal care preference by caregivers which is affected by the cognitive ability deterioration mentioned with respect to activities mentioned in Katz Index along with caregiver parameters. The main objective of the experiment is to reduce the variance in the learning models and stabilize the model performance across various predictors that perform in their own best ways based on the type and size of the sample. The assorted ensemble is performing at par with other ensemble models with respect to various performance metrics but outperformed them during the ROC curve analysis. This research study would help neuropsychologists to understand the effect of cognitive degeneration, and behavioural and caregiver parameters on the manageability of Alzheimer’s disease (AD) in its advanced stages thus enabling them to strategize their treatment and counselling. The comprehensibility of the model would drive its wide usage in neuroscience informatics and add value to the existing repository of chronic disease-related prognosis.