Alzheimer’s Disease Detection Using AI

Abstract: Alzheimer’s disease is a progressive neurological condition that impacts memory, thinking ability, and speech functions. Early detection of Alzheimer’s is very crucial because timely intervention can help slow the progression of symptoms and enhance the overall quality of life for patients. Conventional diagnostic methods mainly depend on neuroimaging, cognitive assessments, and clinical evaluations, which are often costly, time-consuming, and not easily accessible to all individuals.

Recent studies indicate that speech patterns and language usage can provide early indications of cognitive decline. Individuals affected by Alzheimer’s disease frequently show variations in speech fluency, vocabulary diversity, pause pat-terns, and sentence formation. These variations can be exam-ined using computational methods along with machine learning techniques.

In this work, we present a machine learning-based approach for identifying Alzheimer’s disease through speech recordings and linguistic analysis. Acoustic features such as Mel Frequency Cepstral Coefficients (MFCC), pitch, energy, speech rate, and pause duration are derived from audio signals. Linguistic features are obtained from transcribed speech using TF-IDF representation.

Multiple models, including text-based, audio-based, and combined feature models, are tested using the XGBoost classifier. The experimental findings indicate that integrating both acoustic and linguistic features leads to a noticeable improvement in prediction performance. The proposed hybrid model attains an accuracy of 76.44

1