Publication

Abstract : Dementia is the loss of cognitive functioning that affects a person's ability to reason, think, and remember. Due to this, detecting the disease earlier is vital. This study focuses on detecting dementia from speech using the proposed model CogniWave. Models are evaluated to identify signs of dementia from speech patterns accurately. The pre-existing models exhibit a large error rate with less accuracy to address the disadvantages of these models, the proposed model named Cogni-Wave was devised. The study explores a set of speech features such as MFCCs 0 to MFCCs 19, Spectral Centroid, Spectral Rolloff, Mel Spectrogram 0 to Mel Spectrogram 4, Speech Rate, Energy, and ZCR, which are crucial for detecting dementia. These features are extracted from voice samples leveraging various data mining techniques. Additionally, it is being cross validated using a number of metrics, including accuracy, precision, and recall, with 92%, 96%, and 90% respectively.