Speech-based Dementia Prediction using Artificial Intelligence

The global healthcare, social, and economic impacts that affect the patient, family, and society as a whole are significantly affected by dementia. This study's objective is to look at how the analysis of speech patterns might support the early diagnosis of dementia. We will create and assess machine learning models to categorize people with and without dementia using acoustic characteristics taken from voice data. In order to determine the best practices for dementia detection, the study will mix a number of conventional machine learning techniques. The goal of using speech data is to help identify the condition early, which can save a lot of money and allow for prompt interventions to stop its progression. Early detection is vital because it makes room for therapeutic methods that work and protect. It looks at important elements such as methods for collecting data, pre-processing strategies, features classifiers, and extraction techniques. The extract of acoustic features in Python modules such as Librosa was utilized for the analysis of speech data. The Praat software's Python interface, Parselmouth, utilized for OpenSMILE, a toolkit for phonetic analysis, and obtaining a variety of acoustic characteristics that are utilized deep acoustic embedding to efficiently record and reflecting essential aspects of speech that are intimately associated with cognitive deficits. The findings of this investigation could help the creation of accessible and noninvasive dementia-related screening tools.