Towards the Early Detection of Alzheimer Disease Using Machine Learning Approach

Abstract

This paper proposes a machine learning technique to detect Alzheimer's disease in Bangladesh. The proposed system comprises of three components. First, a set of demographics, clinical and laboratory data are collected from web. This data is then pre-processed using feature selection and normalization techniques. Next, a supervised learning algorithm is used to train a model on the pre-processed data. The trained model is then used to detect Alzheimer's disease. Finally, an unsupervised learning algorithm is used to identify any other underlying diseases in the data. The proposed system is evaluated using a dataset consisting of data from patients in the world. The results indicate that the system has a good detection accuracy and can be used to detect Alzheimer's disease in Bangladesh. Alzheimer's is a disease that decreases thinking ability, it pieces of dementia. Those who are suffering from this disease they can’t think in a wide range, they have thinking limitations. The main reason for this disease is old age, the most cases are found in old age people. Patient with Alzheimer’s loses their memory partially, sometimes fully and can’t consider some things. In our country, Bangladesh's perspective though the number of patients is not much but is increasing in a plethora of ways. According to the WHO, a re-port turned in posted in BD. Over there showed up 14993 people or 2.09% of total died for the Alzheimer & Dementia in 2020. The death charge from the one’s illnesses are 13.89 people consistent with 1,00,000. Our major awareness within the paper is to apply machine learning set of rules for detecting Alzheimer sickness in its primitive degree. After the research we got our most accurate model with our processed dataset is K-Nearest Neighbor compared with four different techniques we played with. The model can provide output 96% accurately.