Clinical Feasibility of Speech Phenotyping for Remote Assessment of Neurodegenerative and Psychiatric Disorders

Alzheimer Disease, Amyotrophic Lateral Sclerosis, Bipolar Disorder, Dementia, Vascular, Dementia With Lewy Bodies, Frontotemporal Dementia, Major Depressive Disorder, Mild Cognitive Impairment, Motor Neuron Disease, Parkinson Disease, Primary Progressive Aphasia

Speech, Acoustic, Language, Linguistic, Machine Learning, Artificial Intelligence

The primary objective of the study is to evaluate the feasibility of eliciting continuous narrative speech in different neurodegenerative and psychiatric indications, using remote, self-administered speech tasks, as measured by the average length of speech elicitation for each speech task during the first week of self-assessment. Secondary objectives include (1) evaluating the reliability of speech tasks in the remote self-administered setting, as measured by the intra- and inter-subject variance; (2) accessing the adherence of speech tasks in this setting, as measured by the subject average fraction of days during the first week, where at least one task response is submitted; (3) evaluating the feasibility of using speech tasks in the setting of a telemedicine videoconference, as measured by the average length of speech elicited in each group; (4) evaluate whether a set of acoustic and linguistic patterns can detect each indication, compare to either a control group or all other indications, as measured by the area under the receiver operating characteristic curve (AUC), sensitivity, specificity and Cohen's kappa of the relevant binary classifier; (5) evaluating how the performance of such algorithms can be impacted by speaker and environment covariates, as measured by the Kendall rank correlation coefficient of the AUC of each classifier and each of age group, gender and speech-to-reverberation modulation energy ratio.

None listed

United Kingdom