Complexity and a certain level of uncertainty are inherent to the clinical scenarios faced by the treating teams, often inundated with exhaustive details. Disease arises from a multitude of factors, each with varying impact on patients’ outcomes. Inevitably, patients have intrinsic and contextual aspects that can modify the pathological magnitude from banal to life-threatening. Additionally, the effectiveness of treatments and supportive therapies varies, again, depending on numerous variables. While clinicians endeavor to comprehensively characterize distinct pathological conditions using multidimensional clinical constructs, practical algorithms for treatment and supportive therapies should incorporate probabilities derived from prediction models. Among the most pertinent input variables in the majority of acute and critical care decision support systems is a chronological age. This is unsurprising, considering that even in the realm of health, this deeply human trait, explored by Cicerone over two millennia ago, bears a substantial correlation with the risk of mortality. However, this correlation has changed since the ancient times gaining a significant degree of complexity, which nowadays hinders the setting of treatment proportionality. In the era of artificial intelligence, deep neural networks have identified age as one of the few variables monotonically associated with lower survival in patients receiving venoarterial extracorporeal membrane oxygenation (V-A ECMO), but it has been shown that old patients with certain characteristics may benefit from invasive strategies