1 - Serviço de Imunoalergologia, Centro Hospitalar Universitário de Lisboa Central
2 - NOVA Medical School, CHRC - Comprehensive Health Research Centre, Lisbon, Portugal
- CHRC - 2nd annual summit, may 2022
The COVID -19 pandemic has triggered the development of a wide range of diagnostic tests. As current alternatives for diagnosing COVID -19 are far from ideal, it has been proposed to process human speech and audio signals of coughing and breathing using artificial intelligence techniques. The project OSCAR (vOice Screening of CoronA viRus) aimed to develop an algorithm to screen for COVID -19 using a dataset of Portuguese participants with voice and cough.
It was a cross-sectional observational study in Portugal and divided into two phases. Phase I aimed to characterise the pattern of sounds produced by the vocal apparatus in patients with SARS-CoV-2 infection documented by a positive RT-PCR test to develop and validate an algorithm. In Phase II, the algorithm developed in Phase I was tested in a real-world setting.
In Phase I, after filtering, the training group consisted of 166 subjects who were effectively available to train the classification model (34.3% SARS-CoV-2 positive / 65.7% SARS-CoV-2 negative). Phase II enrolled 58 participants (69.0% SARS-CoV-2 positive / 31.0% SARS-CoV-2 negative). In terms of performance measures, we achieved a sensitivity of 85%, a specificity of 88.9% and an F1 score of 84.7%.
Voice screening algorithms are an attractive strategy for COVID -19 diagnosis. A quick, convenient, and easily accessible test can help improve the screening strategy for positive cases to reduce disease transmission. Our results suggest that the voice-based COVID -19 detection strategy is promising.
Palavras Chave: COVID-19, diagnostic tests machine learning, SARS-CoV-2, speech, voice