Causal Machine Learning and Decision Making

The field of causal machine learning and decision making is moving towards the development of more advanced and dynamic methods for estimating treatment effects and making decisions under uncertainty. Researchers are focusing on overcoming the limitations of traditional methods, such as the inability to capture temporal dynamics and the presence of unmeasured confounders. Novel frameworks and techniques, such as continuous-time modeling and counterfactual reasoning, are being proposed to address these challenges. These innovations have the potential to significantly improve the accuracy and effectiveness of decision-making systems in a variety of domains, including healthcare and reinforcement learning. Notable papers in this area include:

  • CAST, which introduces a novel framework for modeling treatment effects as continuous functions of time, allowing for more accurate estimation of treatment effects in medical survival data.
  • Counterfactual Reasoning Decision Transformer, which enhances decision-making abilities by generating and utilizing counterfactual experiences, enabling improved decision-making in unseen scenarios.

Sources

CAST: Time-Varying Treatment Effects with Application to Chemotherapy and Radiotherapy on Head and Neck Squamous Cell Carcinoma

An Identifiable Cost-Aware Causal Decision-Making Framework Using Counterfactual Reasoning

Sequential Treatment Effect Estimation with Unmeasured Confounders

Beyond the Known: Decision Making with Counterfactual Reasoning Decision Transformer

Counterfactual Strategies for Markov Decision Processes

Built with on top of