We present CACTI, a masked autoencoding approach for imputing tabular data that leverages the structure in missingness patterns and contextual information. Our approach employs a novel median truncated copy masking training strategy that encourages the model to learn from empirical patterns of missingness while incorporating semantic relationships between features — captured by column names and text descriptions — to better represent feature dependence. These dual sources of inductive bias enable CACTIto outperform state-of-the-art methods — an average $R^2$ gain of 7.8\% over the next best method (13.4%, 6.1%, and 5.3% under missing not at random, at random and completely at random, respectively) — across a diverse range of datasets and missingness conditions. Our results highlight the value of leveraging dataset-specific contextual information and missingness patterns to enhance imputation performance.

Patients with end-stage liver disease comprise a patient population whose only definitive treatment is liver transplantation. Anesthetic technique for orthotopic liver transplantation (OLT) requires specialized knowledge and skills because of the complications associated with the procedure and institution-specific equipment and guidelines. Normal "on the job" training may be inadequate, as transplants do not happen on a fixed schedule and the number of transplants that occur in a given year are very small relative to other types of surgeries. Therefore, there is a need for additional training in proper anesthetic technique for OLT. The objective of this project is to develop a serious game designed to teach best practices for the anesthetic management of an OLT that can be used by practitioners to better their skills.