How Does the System Know It’s Right? Automated Confidence Assessment for Compliant Coding

Because there are numerous gray areas in code assignment, it can be difficult to define a single notion of correctness in medical coding. Beyond coding guidelines that may not be fully determinate, the desired outputs for a computer-assisted coding (CAC) system can vary depending on the consumer of the technology, the payer, the date of service, and other factors. Nonetheless, CAC systems need a welldefined basis for evaluating their own correctness in order to avoid presenting subpar coding results as trustworthy. In this paper, we present a solution to that problem.