Duncan: Broadly, the coding solution is an automated tool which works by aggregating the multiple clinical data sources relating to a patient episode (such as pathology tests, operating notes, discharge summary), then ingesting and cleansing the raw data.  It then structures this data (if it is not already in a structured format) by identifying relevant medical terminology, using named entity recognition techniques, and subsequently applies algorithms to predict appropriate clinical codes, which can be validated by the hospital’s coding team.

The solution has been developed using publicly available dummy data.  Over time, feedback from coders, and machine learning techniques, will help it refine its accuracy automatically, and adapt to new coding conventions, diagnoses and treatments.

Duncan: Automated clinical coding helps support hospital strategy in multiple ways. Firstly, it has the potential to improve operational efficiency by converting patient notes from diverse sources (and in various formats) into clinical codes in a consistent way, reducing the work effort for clinical coders and improving accuracy and consistency. This does not mean that manual coders are no longer needed, of course.  The solution can take over low-complexity coding and high volume coding tasks, leaving coders to focus on the more challenging patient cases.  This can also free up coders to work closely with clinicians to support them in developing accurate clinical notes at source - for example helping with the identification and recording of secondary diagnoses, and enabling coders to improve their efficiency and maximise value. Coder shortages also mean many hospitals have significant coding backlogs - automating the clinical coding process could help to reduce these backlogs.

Furthermore, the insights from structured, consolidated, comprehensive patient data provided by the solution support better service planning to inform funding discussions for public hospitals, and more efficient (and accurate) billing for private patients. 

Duncan: The base technology of analysing clinical data has numerous use cases. Whilst more accurate billing and mitigating the shortage of clinical coders may be the initial opportunities, further down the line I can see the technology having a great impact when applied during patient episodes to help standardise care and improve patient outcomes. Patient notes could be analysed in real time to identify trends and correlations that could trigger proactive care interventions, and help support the deployment of established and emerging good practice as new research and data comes to light. 

While the automated coding solution is still in its development phase, it is an exciting step in providing impactful innovation to hospitals using the latest artificial intelligence technology. The insights gained from the AI-based solution can greatly improve operational efficiency, service planning and could potentially be used to uplift patient outcomes and consistency of care. This would mean a more positive experience for both patients and clinicians, and who wouldn’t want that? 

If you would like to hear more, please contact Pia Clinton-Tarestad.