White Paper: “From Correlation to Causation to Artificial Intelligence”
Correlation does not equal causation. And unfortunately, cause and effect cannot be proven from observational data – but we can gain important clues about causal relationships. Read how an intensive, algorithmic search for alternative explanations brings to light a small set of direct and potential causal factors. We show an example where we predict depressive episodes, revealing the effects and side effects of specific groups of drugs and how they affect different groups of patients. Thus, causality becomes an important pillar for future AI systems – not only in healthcare.