The classification of medical diagnoses and treatments, a process known as ‘medical coding’, is typically undertaken solely by human medical coders to turn disparate data elements into discrete actionable information. This coding process can be costly, time consuming, and vulnerable to human error. In order to supplement the human coding process, we introduce Hank.ai, which leverages state-of-the-art Artificial Intelligence (AI) technologies to reduce the cost, time and inaccuracies inherent in human processing.

Primarily, Hank utilizes innovative deep learning and machine learning methodologies to predict codes for humans during their typical medical coding workflow. However, Hank’s infrastructure also leverages the indispensable knowledge of the human coder, allowing them to approve or correct any prediction and provide relevant context for that decision in the form of natural language.

In essence, Hank is a self-programming expert system that leverages human knowledge to enhance the predictive capabilities of its underlying deep learning models.