Causal AI in Personalised Healthcare
Tobias Hatt · 2023
Personalised healthcare aims to tailor healthcare to the individual patient by estimating the effect of a specific treatment on that patient. To estimate the effect of a treatment, conventional medicine uses clinical trials. However, clinical trials are too small to capture all possible variations among patients and can only provide estimates for the average patient rather than the individual. Real-world data (RWD), which are generated outside of clinical trials and can capture more patient variations, have the potential to provide estimates for individualised treatment effects. However, standard statistical methods may not accurately estimate treatment effects from RWD, because treatment effects are causal and require the ability to distinguish causation from correlation. To address this challenge, we turn to the field, which focuses on estimating causal effects using artificial intelligence (AI) techniques, called causal AI. By using causal AI methods to learn from RWD, we can accurately estimate individualised treatment effects and approach personalised healthcare.