Great article! Recent Machine Learning models do lack interpretability, for which Causal ML could be a better choice.
However, the assumptions that LiNGAM makes might not always be feasible in medical data. The relationships between variables could be non-linear too. For example, the effect of a drug might not scale linearly with dosage. I hope there are ways to overcome these limitations as well.
Great article! Recent Machine Learning models do lack interpretability, for which Causal ML could be a better choice.
However, the assumptions that LiNGAM makes might not always be feasible in medical data. The relationships between variables could be non-linear too. For example, the effect of a drug might not scale linearly with dosage. I hope there are ways to overcome these limitations as well.