The voice as a bio-marker of Covid-19: Preliminary resultsof the CO-VOICE-19 study

Background: Covid-19 has an impact on lung function and, consequently, on voice emission By registering an adequate numberof patients with Covid-19, we can 'train' artificial intelligence algorithms in order to highlight the disease status of any personwhose voice is registered Impact on voice increases with diseaseprogression, allowing staging Materials and Methods: Prospective pilot study to evaluate thecondition of Covid-19 affection of critically ill patients hospitalizedand monitored by evaluating their speech capacity through measurement and recording of the voice Primary End Point: remotelylocate people infected with Covid-19 Secondary end-points: establish the presence of any geographic areas with 'outbreaks', by'crossing' the geo-location data, staging the disease Results: 85 patients evaluated and 18 (10F and 8M) recruited,average age 62, subjected to intubation 3/18 WHO stage 2: 50%;comorbidity>3: 61%;only 2 with P/F<200 Conclusions: Recruited patients have features of lower functionalimpairment than other patients, however a high incidence of previous intubation Preliminary audio signal analysis of the patient's voice recordings are underway and will be treated with ArtificialIntelligence algorithms in order to select voice parameters thatcan identify the presence of the disease By training appropriatemachine learning and data classification systems it will be possible to determine whether the recorded voice belongs to a healthysubject or affected by Covid-19 and use these skills to screen suspect patients by telephone triage