A real-time integrated framework to support clinical decision making for covid-19 patients

• DOI: 10.1016/j.cmpb.2022.106655
• Publication Date: 2022-01-29T23:31:23Z
• Authors: Rita Murri; Carlotta Masciocchi; Jacopo Lenkowicz; Massimo Fantoni; Andrea Damiani; Antonio Marchetti; Paolo Domenico An...; Giovanni Arcuri; Alfredo Cesario; Stefano Patarnello; Massimo Antonelli; Rocco Bellantone; Roberto Bernabei; Stefania Boccia; Paolo Calabresi; Andrea Cambieri; Roberto Cauda; Cesare Colosimo; Filippo Crea; Ruggero De Maria; Valerio De Stefano; Francesco Franceschi; Antonio Gasbarrini; Raffaele Landolfi; Ornella Parolini; Luca Richeldi; Maurizio Sanguinetti; Andrea Urbani; Maurizio Zega; Giovanni Scambia; Vincenzo Valentini
• Keywords: Clinical Decision Making; 2019-20 coronavirus outbreak

The COVID-19 pandemic affected healthcare systems worldwide. Predictive models developed by Artificial Intelligence (AI) and based on timely, centralized standardized real world patient data could improve management of to achieve better clinical outcomes. objectives this manuscript are describe the structure technologies used construct a Data Mart architecture present how large hospital has tackled challenge supporting daily emergency, creating strong retrospective knowledge base, time environment integrated information dashboard for practice early identification critical condition at level. This framework is also as an informative, continuously enriched lake, which base several on-going predictive studies. technology research was described. It using SAS Institute software analytics tool SAS® Vyia® Open-Source R ® Python fast prototyping modeling. included variables source extraction procedures were presented. covers cohort 5528 patients with SARS-CoV-2 infection. People who died older, had more comorbidities, reported frequently dyspnea onset, higher d-dimer, C-reactive protein urea nitrogen. support three levels: hospital, single ward individual care integration collection AI-based been developed, set automated mining retrieval, transformation integration, embedded in help managing care. Benefits from availability include opportunity build machine learning approach identify undescribed phenotypes foster networks. A real-time updated built may represent valid epidemiological features COVID-19, especially when multiple waves observed, well epidemic events same nature (e. g. conditions leading severe pulmonary inflammation). Therefore, we believe presented paper find applications comparable situations even region or state levels. Finally, predicting course future new pandemics largely benefit network DataMarts.