Distinguish the Severity of Illness Associated with Novel Coronavirus (COVID-19) Infection via Sustained Vowel Speech Features.

The authors are currently conducting research on methods to estimate psychiatric and neurological disorders from a voice by focusing on the features of speech. It is empirically known that numerous psychosomatic symptoms appear in voice biomarkers; in this study, we examined the effectiveness of distinguishing changes in the symptoms associated with novel coronavirus infection using speech features. Multiple speech features were extracted from the voice recordings, and, as a countermeasure against overfitting, we selected features using statistical analysis and feature selection methods utilizing pseudo data and built and verified machine learning algorithm models using LightGBM. Applying 5-fold cross-validation, and using three types of sustained vowel sounds of /Ah/, /Eh/, and /Uh/, we achieved a high performance (accuracy and AUC) of over 88% in distinguishing "asymptomatic or mild illness (symptoms)" and "moderate illness 1 (symptoms)". Accordingly, the results suggest that the proposed index using voice (speech features) can likely be used in distinguishing the symptoms associated with novel coronavirus infection.

A web visualization tool using T cell subsets as the predictor to evaluate COVID-19 patient's severity.

Wuhan, China was the epicenter of the 2019 coronavirus outbreak. As a designated hospital for COVID-19, Wuhan Pulmonary Hospital has received over 700 COVID-19 patients. With the COVID-19 becoming a pandemic all over the world, we aim to share our epidemiological and clinical findings with the global community. We studied 340 confirmed COVID-19 patients with clear clinical outcomes from Wuhan Pulmonary Hospital, including 310 discharged cases and 30 death cases. We analyzed their demographic, epidemiological, clinical and laboratory data and implemented our findings into an interactive, free access web application to evaluate COVID-19 patient's severity level. Our results show that baseline T cell subsets results differed significantly between the discharged cases and the death cases in Mann Whitney U test: Total T cells (p < 0.001), Helper T cells (p <0.001), Suppressor T cells (p <0.001), and TH/TSC (Helper/Suppressor ratio, p<0.001). Multivariate logistic regression model with death or discharge as the outcome resulted in the following significant predictors: age (OR 1.05, 95% CI, 1.00 to 1.10), underlying disease status (OR 3.42, 95% CI, 1.30 to 9.95), Helper T cells on the log scale (OR 0.22, 95% CI, 0.12 to 0.40), and TH/TSC on the log scale (OR 4.80, 95% CI, 2.12 to 11.86). The AUC for the logistic regression model is 0.90 (95% CI, 0.84 to 0.95), suggesting the model has a very good predictive power. Our findings suggest that while age and underlying diseases are known risk factors for poor prognosis, patients with a less damaged immune system at the time of hospitalization had higher chance of recovery. Close monitoring of the T cell subsets might provide valuable information of the patient's condition change during the treatment process. Our web visualization application can be used as a supplementary tool for the evaluation.