Scientists from the University of Granada design a method based on artificial intelligence that allows detecting COVID-19 from the voice or coughs of patients
The method proposed by the researchers is capable of detecting the COVID-19 disease with an accuracy of around 77%, analyzing only the patient’s voice.
It is a non-invasive and instant technology, so it could be used through an app or phone call to quickly diagnose the disease without the need for the patient to travel to the health center. An international team made up of scientists from the universities of Granada (UGR) and Eurecom (France), together with engineers from the company Biometric Vox (Murcia), have designed a method based on artificial intelligence techniques that allows the detection of the COVID-19 disease. 19 from voice and cough recordings made to patients.
The researchers have developed a technique that automatically analyzes these recordings in search of acoustic patterns that may be indicative of the person suffering from this disease.
The coronavirus cough is different
“As we know, among other associated symptoms, patients with coronavirus present a dry cough, a feeling of shortness of breath and increased respiratory rate, so the developed method looks for alterations in the patient’s voice that may be indicative that this suffers from the disease”, explains José Andrés González López, professor in the Department of Signal Theory, Telematics and Communications at the UGR, and one of the authors of this work.
As a result, the method proposed by the researchers is capable of detecting the COVID-19 disease with an accuracy of around 77% by analyzing only the patient’s voice.
The advantage of this method over others currently available to detect COVID-19, such as antigen tests or PCR, “is that our technology is completely non-invasive and instantaneous, so it could be used through an app or telephone calls as a method of screening or rapid diagnosis of the disease, without the need for the patient to have to travel to the health center and thus put other citizens at risk”, points out the UGR researcher.
Likewise, the researchers point out that this technology based on voice and cough recognition could also be applied in the future to detect or monitor other similar diseases that affect the respiratory tract, such as flu, colds, etc.
Bibliographic reference:
Kamble, MR; Gonzalez-Lopez, JA; Grau, T.; Espin, JM; Cascioli, L.; Huang, Y.; Gomez-Alanis, A.; Patino, J.; Font, R.; Hairstyle, AM; Gomez, AM; Evans, N.; Zuluaga, MA; Todisco, M., “PANACEA cough sound-based diagnosis of COVID-19 for the DiCOVA 2021 challenge,” Proc. Interspeech 2021, Brno, Czech Republic.