New Paper: Real-time clinician text feeds from electronic health records

Doctors record patient symptoms and tests in electronic health records, and this data is collected manually using electronic forms and then submitted to national databases to track disease. During the Covid pandemic, hospitals collect these figures manually from the record and then submit them to Public Health England to track trends, which has allowed the NHS to provide detailed charts.

Teams at KCH and GSTT instead used a different technique used in social media. Instead of manually finding and collecting cases, the text of what doctors were recording was analysed to detect word trends or phrases trending in the NHS health records. This produced real-time trending maps which correlated strongly manually collected data and also regional tracking of Covid cases in hospitals. The team also showed that as new symptoms from Covid were discovered, doctors recorded them more often. This would be analogous to hashtags in internet social media but instead confined to inside the NHS.

This tracking of word trends in electronic health records could form an alternative way of studying disease and healthcare, which is very easy and cost-effective to run at a single hospital level or even a whole region level with appropriate privacy safeguards. 

This work was made possible through Kings College London researchers and software engineers behind the Cogstack platform and the Health Data Research UK Text Analytingics workstream, and the NHS IT departments having to adapt quickly to a rapidly changing pandemic.

The full paper can be viewed online at https://www.nature.com/npjdigitalmed/.

For more information on the HDR UK Text Analytics Programme please visit https://hdruk-text.org/about/