MultiMBNN: Matched and Balanced Causal Inference with Neural Networks
This work addresses confounding in observational studies for applications like healthcare and policy evaluation, representing an incremental improvement over existing methods.
The paper tackles confounding in causal inference for multiple treatment scenarios by proposing MultiMBNN, a neural network method that uses generalized propensity score matching and balanced representations, and it outperforms state-of-the-art algorithms like TARNet and Perfect Match on synthetic and real-world datasets using metrics such as PEHE and mean absolute percentage error over ATE.
Causal inference (CI) in observational studies has received a lot of attention in healthcare, education, ad attribution, policy evaluation, etc. Confounding is a typical hazard, where the context affects both, the treatment assignment and response. In a multiple treatment scenario, we propose the neural network based MultiMBNN, where we overcome confounding by employing generalized propensity score based matching, and learning balanced representations. We benchmark the performance on synthetic and real-world datasets using PEHE, and mean absolute percentage error over ATE as metrics. MultiMBNN outperforms the state-of-the-art algorithms for CI such as TARNet and Perfect Match (PM).