Coronavirus can be diagnosed in a patient based on the coordination of speech production subsystems, i.e. voice, as the disease affects the lungs.
This was stated by experts from MIT following studies, referred to in article of the journal Open Journal of IEEE Engineering in Medicine and Biology.
Scientists develop system modeling, speech and signal processing for detection and tracking COVID-19 in the asymptomatic and symptomatic stages.
Method the researchers is based on the complexity of the neuromotor coordination of the speech subsystems involved in respiration, phonation and articulation affected by the virus.
Specialists conducted preliminary research while only an audio interview of five subjects before and after infection with coronavirus. Studied coordination of breathing (measured by the amplitude of the acoustic wave form) and the movement of the larynx (the fundamental frequency and the cepstral peak performance), as well as coordination of the movement of the larynx and of the joint (forming the center frequency).
Scientists came to the conclusion that although there is a strong dependence of the subject, the morphology of size-effects at the group level “experimental” indicates a decrease in the complexity of the coordination of subsystems. They proved that when infected COVID-19 significantly reduced the diversity of audio signals.
Experts plan more research with more controlled data sets, and to eliminate contradictory influences, such as different recording conditions, the unbalanced amount of data and changes in the basic state of the voice compared to recordings before and after infection.