Cargando…

Factibilidad de la utilización de la inteligencia artificial para el cribado de pacientes con COVID-19 en Paraguay

OBJECTIVE. Study the feasibility of using artificial intelligence as a sensitive and specific method for COVID-19 screening in patients with respiratory conditions, using chest CT scan images and a telemedicine platform. METHODS. From March 2020 to June 2021, the authors conducted an observational d...

Descripción completa

Detalles Bibliográficos
Autores principales: Galván, Pedro, Fusillo, José, González, Felipe, Vukujevic, Oraldo, Recalde, Luciano, Rivas, Ronald, Ortellado, José, Portillo, Juan, Borba, Julio, Hilario, Enrique
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Organización Panamericana de la Salud 2022
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8942285/
https://www.ncbi.nlm.nih.gov/pubmed/35350452
http://dx.doi.org/10.26633/RPSP.2022.20
Descripción
Sumario:OBJECTIVE. Study the feasibility of using artificial intelligence as a sensitive and specific method for COVID-19 screening in patients with respiratory conditions, using chest CT scan images and a telemedicine platform. METHODS. From March 2020 to June 2021, the authors conducted an observational descriptive multicenter feasibility study based on artificial intelligence (AI) for COVID-19 screening using chest images of patients with respiratory conditions who presented at public hospitals. The AI platform was used to diagnose chest CT scan images; this was then compared with molecular diagnosis (RT-PCR) to determine whether they matched and to analyze the feasibility of AI for screening patients with suspected COVID-19. A telemedicine platform was used to send images and diagnostic results. RESULTS. Screening of 3 514 patients with a suspected COVID-19 diagnosis was performed in 14 hospitals around the country. Most patients were aged 27 to 59 years, followed by those over 60. The average age was 48.6 years; 52.8% were male. The most frequent findings were severe pneumonia, bilateral pneumonia with pleural effusion, bilateral pulmonary emphysema, and diffuse ground glass opacity, among others. There was an average of 93% matching and 7% mismatching between images analyzed by AI and RT-PCR. Sensitivity and specificity of the AI system, obtained by comparing AI and RT-PCR screening results, were 93% and 80% respectively. CONCLUSIONS. The use of sensitive and specific AI for stratified rapid detection of COVID-19 in patients with respiratory conditions by using chest CT scan images and a telemedicine platform in public hospitals in Paraguay is feasible.