Voice Biomarkers: Cognitive Impairment, including Alzheimer's Disease,Dementia, or Mild Cognitive Impairment: Introduction to PeripheralNeuropathy

INTRODUCTION/OBJECTIVE: Alzheimer's disease (AD) can cause certain nervous disorders, which in turn can lead to voice disorders and abnormal values for certain acoustic parameters. Mild cognitive impairment (MCI) is probably the early stage of the disease. Dementia is one of the causes of Alzheimer's disease. Whether or not there is a link between these three cognitive impairments, lesions affecting the vocal cords or articulators can be caused by neurological structures that influence phonation. The potential of biomarkers in the detection of cognitive impairment is remarkable. In our study, we will examine vocal biomarkers obtained from the extraction of acoustic features. The aim of this study is to combine vocal biomarkers with cognitive diseases. METHODS: The standardised dataset used has recently been made publicly available. Cognitive impairment, including Alzheimer's disease, dementia, or MCI, is diagnosed from /a/ and diadochokinesis-pataka vocalisations using Mel-Frequency Cepstral Coefficients (MFCCs) transformed into 2D scalogram images. For processing, we will use the pre-trained OpenL3 network, and our less resource-intensive network called Op1Net to classify diseased and healthy groups. RESULTS: A significant difference was observed compared to the control group. For the /a/ vocalisation, classification accuracy across all ages and genders was 82.1%, and the AUC value was 88.3%, while for diadochokinesis-pataka, accuracy was 69.8% and the AUC value was 75.4%. In the group of women over 55 years of age, the accuracy was 80.93%, and the AUC value was 87.12%. DISCUSSION: Performance results clearly show that there is a correlation between the voice and the neurodegenerative disease AD, dementia, or MCI. We can see that the results of the data classification, including all ages and genders, for the sound /a/ are higher than those for the 'pataka' vocalisations. The prolonged vowel provides more information about the disease. CONCLUSION: This preliminary multidisciplinary study clearly demonstrates the existence of a link between neurological disease and the voice, and raises several questions concerning the nervous system, particularly the vagus nerve and associated neuropathy.