Data mining and machine learning for the development of risk-predicting models in patients with Covid-19 pneumonia, based on the chest x-ray Brixia-severity score and lab tests

The project is part of the actions aimed at contrasting Covid-19 infection through the development of prognostic risk models based on quantitative data related to lung damage and biochemical tests, detected in 1300 patients affected by Covid-19 in course of hospitalization at ASST Spedali Civili Brescia. Lung damage assessment is performed by Brixia-severity radiological score and the state of tissue inflammation through the biochemical data PCR, ferritin, LDH, troponin, D-Dimer, fibrinogen, WBC. The methods are based on multivariate statistical analysis, data mining and artificial intelligence algorithms.

Scientific coordinators: Roberto Maroldi, Alfonso Gerevini, Paola Zuccolotto

Predictive model for survival probability (up to 12 days) of patients to be hospitalized

Cox model based on sex, age, Brixia Severity Score, C-reactive protein level. Developed by M. Sandri and P. Zuccolotto.

Go to the app