Multi-treatment Effect Estimation from Biomedical Data

TL;DR

This paper introduces M3E2, a neural network model for estimating multiple treatment effects in biomedical data, capable of handling various treatment types and covariates, outperforming existing methods in synthetic benchmarks.

Contribution

The paper presents M3E2, a novel multi-task neural network that estimates multiple treatment effects simultaneously, handling continuous, binary treatments, and numerous covariates, with superior performance over baselines.

Findings

M3E2 outperforms baseline methods in synthetic benchmarks.

It effectively estimates multiple treatment effects simultaneously.

The model handles both continuous and binary treatments.

Abstract

This work proposes the M3E2, a multi-task learning neural network model to estimate the effect of multiple treatments. In contrast to existing methods, M3E2 can handle multiple treatment effects applied simultaneously to the same unit, continuous and binary treatments, and many covariates. We compared M3E2 with three baselines in three synthetic benchmark datasets: two with multiple treatments and one with one treatment. Our analysis showed that our method has superior performance, making more assertive estimations of the multiple treatment effects.