Artificial Intelligence-based Clinical Decision Support for COVID-19 -- Where Art Thou?

Prior to January 2020, the artificial intelligence and machine learning (AI/ML) for healthcare community had many reasons to be pleased with the recent progress of their field. Learning-based algorithms had been shown to accurately forecast the onset of septic shock [1], MLbased pattern recognition methods classified skin lesions with dermatologist level accuracy [2], diagnostic AI systems successfully identified diabetic retinopathy during routine primary care visits [3], AIbased breast cancer screening outperformed radiologists by a fairly large margin [4], MLdriven triaging tools improved outcome differentiation beyond the emergency severity index [5], AIenabled assistance systems simplified interventional workflows [6], and algorithm-driven organizational studies enabled redesign of infusion centers [7]. Many would have argued that, after nearly 60 years on the test bench [8], AI in healthcare had finally reached a level of maturity, performance, and reliability that was compatible with the unforgiving requirements imposed by clinical practice. Today, only a few months later, this rather sunny outlook has become overcast. The worlds healthcare systems are facing the outbreak of a novel respiratory disease, COVID-19.