A hybrid data envelopment analysis—artificial neural network prediction model for COVID-19 severity in transplant recipients

• DOI: 10.1007/s10462-021-10008-0
• Publication Date: 2021-04-23T06:02:31Z
• Authors: Ignacio Revuelta; Francisco J. Sant...; Enrique Montagud-...; Pedro Ventura-Aguiar; Debora Di Caprio; Frederic Cofan; David Cucchiari; Vicens Torregrosa; Gaston Julio Piñeiro; Nuria Esforzado; Marta Bodro; Jessica Ugalde-Al...; Asuncion Moreno; Josep M. Campistol; Antonio Alcaraz; Beatriu Bayès; Esteban Poch; Federico Oppenheimer; Fritz Diekmann
• Keywords: Identification

In an overwhelming demand scenario, such as the SARS-CoV-2 pandemic, pressure over health systems may outburst their predicted capacity to deal with extreme situations. Therefore, in order successfully face a emergency, scientific evidence and validated models are needed provide real-time information that could be applied by any center, especially for high-risk populations, transplant recipients. We have developed hybrid prediction model whose accuracy relative several alternative configurations has been through battery of clustering techniques. Using hospital admission data from cohort hospitalized patients, our Data Envelopment Analysis (DEA)-Artificial Neural Network (ANN) extrapolates progression towards severe COVID-19 disease 96.3%, outperforming competing model, logistic regression (65.5%) random forest (44.8%). this regard, DEA-ANN allows us categorize evolution patients values analyses performed at admission. Our help guiding management identification key predictors permit sustainable resources patient-centered model.The online version contains supplementary material available 10.1007/s10462-021-10008-0.