Decision trees for COVID-19 prognosis learned from patient data: Desaturating the ER with Artificial Intelligence

ObjectivesTo present a model that enhances the accuracy of clinicians when presented with a possibly critical Covid-19 patient. MethodsA retrospective study was performed with information of 5,745 SARS-CoV2 infected patients admitted to the Emergency room of 4 public Hospitals in Madrid belonging to Quiron Salud Health Group (QS) from March 2020 to February 2021. Demographics, clinical variables on admission, laboratory markers and therapeutic interventions were extracted from Electronic Clinical Records. Traits related to mortality were found through difference in means testing and through feature selection by learning multiple classification trees with random initialization and selecting the ones that were used the most. We validated the model through cross-validation and tested generalization with an external dataset from 4 hospitals belonging to Sanitas Hospitals Health Group. The usefulness of two different models in real cases was tested by measuring the effect of exposure to the model decision on the accuracy of medical professionals. ResultsOf the 5,745 admitted patients, 1,173 died. Of the 110 variables in the dataset, 34 were found to be related with our definition of criticality (death in <72 hours) or all-cause mortality. The models had an accuracy of 85% and a sensitivity of 50% averaged through 5-fold cross validation. Similar results were found when validating with data from the 4 hospitals from Sanitas. The models were found to have 11% better accuracy than doctors at classifying critical cases and improved accuracy of doctors by 12% for non-critical patients, reducing the cost of mistakes made by 17%.

Compassionate Use of Tocilizumab in Severe SARS-CoV2 Pneumonia. When late administration is too late.

IntroductionTocilizumab is an interleukin 6 receptor antagonist which has been used for the treatment of severe SARS-CoV-2 pneumonia (SSP), aiming to ameliorate the cytokine release syndrome (CRS) -induced acute respiratory distress syndrome (ARDS). However, there is no data about the best moment for its administration along the course of the disease. MethodsWe provided tocilizumab on a compassionate-use basis to patients with SSP hospitalized (excluding intensive care and intubated cases) who required oxygen support to have a saturation >93%. Primary endpoint was intubation or death after 24 hours of its administration. Patients received at least one dose of 400 mg intravenous tocilizumab during March 8-2020, through April 20-2020. FindingsA total of 207 patients were studied and 186 analysed. The mean age was 65 years and 68% were male. A co-existing condition was present in 68 % of cases. At baseline, 114 (61%) required oxygen support with FiO2 >0.5 % and 72 (39%) [≤]0.5%. Early administration of tocilizumab, when the need of oxygen support was still below FiO2 [≤]0.5%, was significantly more effective than given it in advanced stages (FiO2 >0.5 %), achieving lower rates of intubation or death (13% vs 37% repectively, p<0{middle dot}001). InterpretationThe benefit of tocilizumab in severe SARS-Cov-2 pneumonia is only expected when it is administrated before the need of high oxygen support. FundingNone.

COVID-19 Outcomes in 4712 consecutively confirmed SARS-CoV2 cases in the city of Madrid.

There is limited information describing features and outcomes of patients requiring hospitalization for COVID19 disease and still no treatments have clearly demonstrated efficacy. Demographics and clinical variables on admission, as well as laboratory markers and therapeutic interventions were extracted from electronic Clinical Records (eCR) in 4712 SARS-CoV2 infected patients attending 4 public Hospitals in Madrid. Patients were stratified according to age and stage of severity. Using multivariate logistic regression analysis, cut-off points that best discriminated mortality were obtained for each of the studied variables. Principal components analysis and a neural network (NN) algorithm were applied. A high mortality incidence associated to age >70, comorbidities (hypertension, neurological disorders and diabetes), altered vitals such as fever, heart rhythm disturbances or elevated systolic blood pressure, and alterations in several laboratory tests. Remarkably, analysis of therapeutic options either taken individually or in combination drew a universal relationship between the use of Cyclosporine A and better outcomes as also a benefit of tocilizumab and/or corticosteroids in critically ill patients. We present a large Spanish population-based study addressing factors influencing survival in current SARS CoV2 pandemic, with particular emphasis on the effectivity of treatments. In addition, we have generated an NN capable of identifying severity predictors of SARS CoV2. A rapid extraction and management of data protocol from eCR and artificial intelligence in-house implementations allowed us to perform almost real time monitoring of the outbreak evolution.