Academic
Alzeimer Disease Prediction Using ML
Introduction
This project leverages machine learning techniques to aid in the diagnostic process, classifying patients into Dementia and Non-Dementia groups based on clinical data. By harnessing the predictive power of algorithms, this work aims to support healthcare providers in identifying Alzheimer’s with increased accuracy.
Dataset Collection
The data for this study was obtained from Kaggle, consisting of records of patients with and without dementia. The dataset includes the following primary attributes, each essential for understanding dementia-related factors:
- Group: Dementia or Non-Dementia
- MMSE (Mini-Mental State Examination): Cognitive assessment score
- Age: Patient age
- EDUC (Years of education): Number of years of formal education
- SES (Socioeconomic Status): Socioeconomic background of the patient
- eTIV (Estimated Total Intracranial Volume): Measure of brain volume
- nWBV (Normalized Whole Brain Volume): Normalized measure of brain matter
- CDR (Clinical Dementia Rating): Scale used to stage dementia
Models used and accuracy
In this analysis, several machine learning models were employed to predict Alzheimer's disease, each with varying degrees of accuracy. Here are the models and their corresponding accuracy rates:
- Support Vector Machine (SVM) - 87% accuracy
- Random Forest - 90% accuracy
- Logistic Regression - 83% accuracy
- AdaBoost - 88% accuracy