Automated Patient History

King’s College Hospital (KCH) faced a monumental task during the deployment of its new Electronic Health Record system, Epic. To ensure accurate and comprehensive patient records, the problem lists of 2 million patients needed to be back-populated into the system. Manually reviewing patient histories across 5 million documents would have required 40 staff members working manually for approximately 3.5 years.

Using Natural Language Processing (NLP) models, a single engineer was able to efficiently identify and extract common conditions across all 2 million patients. This enabled the pre-population of Epic problem lists in just one month.

This innovative approach allowed KCH to review 10 times the amount of historical data compared to manual methods, saving the Trust approximately £550,000. Manual review would have been limited to just two years of historical data, whereas this solution provided a more comprehensive view, significantly improving patient record accuracy and reducing resource demands.