Is an algorithm based on Deep Learning techniques able to predict the prognosis of a Covid 19 patient in the same way as an expert medical doctor?
The study published in The Journal of the Acoustical Society of America shows that the percentage of agreement between the analyzes on Covid-19 data made at DISI on lung ultrasounds with Deep Learning techniques and those made by medical experts on COVID-19 data is 85.96%, a very high correlation.
The algorithm, which was developed by the research team led by prof. Libertario Demi, co-author of the article, acquires patient images according to a standardized protocol and is then able to distinguish patients at high risk of clinical worsening from those at low risk. The prognostic results obtained are in line with those of the leading clinical experts in lung ultrasound imaging.
The publication, entitled "Deep learning applied to lung ultrasound videos for scoring COVID-19 patients: A multicenter study", is the result of the collaboration between the Department of Information Engineering and Computer Science of the University of Trento, the Fondazione Policlinico Universitario Agostino Gemelli IRCCS in Rome and the Fondazione IRCCS Policlinico San Matteo in Pavia.
This is the first study in the world on the subject. The results demonstrate the potential of Deep Learning models when applied to high quality data acquired according to a standardized imaging protocol.