[EMAIL PROTECTED] wrote: > We have now launched the web page that converts > clinical notes to SCT codes which is available at: > > http://www.it.usyd.edu.au/~clinnote/gpnotes.cgi > > Pls use it to give us examples of the notes you record in your daily > practice. You can select to get instant reporting or delayed reporting via > email (reccommended for lengthy text > 5 full lines). > Please give us feedback on what we have missed which you think should be > describale in SCT - it may that SCT doesn't have it. > Our analysis is by no means perfect and any feedback will help us to > improve it.
This is fab! Still a long, long way to go - for example, it doesn't handle abbreviations or multiple qualifiers, and there is a need to capture the semantic relationships between the concepts identified (especially negation), but if anyone can solve such problems, it is likely to be a professor of language technology and his team... > You can play with it to scan what it does by inserting any medical related > text. We were using some published articles and discovered that SCT knows > "postmenopausal" but doesn't know "post-menopausal". We have lots of text pre-processing code which was written to help normalise emergency department triage desk free text (if anyone is interested, a description of the public health surveillance system in which it is used appears here: http://www.biomedcentral.com/1471-2458/5/141 ), including my hamfisted stab at an automated semi-Bayesian spelling corrector (somewhat smarter than your average word processor spell checker). We can share this code with your team if you like. Having seen your demo, I still think that the dream of a mechanism in which the GP types the first few characters of each term into a text box and the system provides an intelligent set of guesses about whcih SCT concept (or English language word) was intended and auto-completes it for him/her is feasible. And the next step beyond that, in whch said GP mutters notes into a microphone and the computer does voice recognition on the utterances. recent versions of Dragon Dictate or ViaVoice or similar voice recognition productions perform impressively well - just a matter of marrying the two together. Very exciting! Tim C _______________________________________________ Gpcg_talk mailing list [email protected] http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
