AI for clinical coding PoV garners mixed results

NEWS - eHealthNews.nz editor Rebecca McBeth

Health New Zealand Te Whatu Ora has finished a proof of value (PoV) around artificial intelligence (AI) for clinical coding  and will move forward with one company to develop a solution.

Jon Herries, group manager of emerging health technology and innovation, told the MTANZ Conference in June that the adoption of AI could help to address the acute shortage of clinical coders, particularly in smaller centres.

“We are really struggling at the moment, particularly in smaller centres, to have clinical coders available to code discharges,” he said. 

On average, a hospital admission generates around 60 pages of content, which needs to be condensed into around 12 codes. 

The initiative aims to reduce the workload of clinical coders by leveraging AI to summarise these vast amounts of patient data into manageable codes. 

Herries said Health New Zealand conducted a proof-of-value study involving five teams to explore the potential of AI in clinical coding, but acknowledged the results “weren’t great”.

However, the organisation plans to collaborate with one of these companies, which it cannot yet name, on developing its plan and vision for how it could work in the future.

Herries said if successful, this could free up clinical coders to focus on more critical tasks, such as identifying risk factors in hospitals and examining variations in patient recovery times. 

Digitising the process would also allow data sharing and reduce the reliance on physical document transfers.

A Registration of Interest (ROI) for the AI coding project says clinical codes are currently identified and entered manually by individual teams nationwide, who work independently with no common worklist.

Clinically coded summaries of discharges are forwarded to National Collections where the information is loaded into the National Minimum Data Set (NMDS), which supports service planning, research and policy development.

You’ve read this article for free, but good journalism takes time and resource to produce. Please consider supporting eHealthNews by becoming a member of HiNZ, for just $17 a month.

To comment on or discuss this news story, go to the eHealthNews category on the HiNZ eHealth Forum

Read more Analytics news

Return to eHealthNews.nz home page