“…In order to circumvent the parameter issue of classic SIR-derived methods while still allowing the mathematical model to cope with time-varying coefficients, the use of Machine Learning strategies has been a popular choice and a trend. Indeed, recent developments involving variable-parameter SIR variants to assess the course of Covid-19 can be found in [ 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 ], which include the use of effective Artificial Intelligence (AI) strategies, for example in [ 18 , 19 , 29 , 30 , 31 , 32 , 33 ]. Following these recent efforts in modeling Covid-19 dynamics from epidemic models tuned with learning mechanisms, in this paper we propose an effective, data-driven SIR model whose parameters are fully calibrated by temporal functions, learned from individual regressors and trained on different data sources.…”