Date(s) - 25/06/2013
2:00 pm - 3:00 pm
seminar room 5, IoP
Extracting information for research from free text in electronic health records
Electronic health records are extremely useful for research, but much of the information is entered as unstructured free text rather than in coded form. Free text may contain clinical information which can better characterise participants in a research study or identify outcomes which are not coded. However, because of concerns over confidentiality, free text is not readily available to researchers using health record databases, and it can be time-consuming to analyse without automated tools.
Natural language processing systems to analyse clinical text may overcome these research difficulties, and could also assist patient care by facilitating decision support and data entry by clinicians. This talk will focus on the Freetext Matching Algorithm, an open source program which converts free text into Read codes (the coding system used by general practitioners in the UK), and can also extract dates and selected test results. It was developed and tested in the Clinical Practice Research Datalink (formerly the General Practice Research Database), a research database of anonymised UK general practice records.