Irregular Sampling Handling

This paper presents a new approach to inferring unbiased treatment effects, using human-readable ordinary differential equations (ODEs) instead of traditional neural networks. This method enhances interpretability and accommodates irregular sampling, while introducing fresh identification assumptions. The innovation lies in transforming any ODE discovery into a treatment effects methodology, potentially revolutionizing the field.