home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!pipex!unipalm!uknet!edcastle!kens
- From: kens@castle.ed.ac.uk (K G N Sutherland)
- Newsgroups: comp.ai.neural-nets
- Subject: Request: Refs on NNs for medical diagnosis
- Message-ID: <30720@castle.ed.ac.uk>
- Date: 25 Jan 93 11:13:39 GMT
- Distribution: comp.ai.neural-nets
- Organization: Edinburgh University
- Lines: 29
-
- Hi, I wonder if anybody out there knows anything
- about the problems of using NNs for medical diagnosis?
- I'm keen to try to use a conventional MLP to establish
- whether a particular set of clinical symptoms can be
- reliably related to a particular diagnosis. However,
- a number of problems are apparent.
-
- I'm particularly interested in the problems of:
-
- - subjectivity of the data.
- - 'fuzzy' nature of some symptoms (ie. fairly XXX).
- - sample size in relation to number of symptoms used.
-
- Any refs/suggestions ?
-
- Thanks in advance.
-
- Ken
-
- -------------------------------------------------------------------
-
- Dr K Sutherland,
- University of Edinburgh,
- Western General Hospital,
- Crewe Road,
- Edinburgh EH4 2XU
- UK
- email: K.G.N.Sutherland@uk.ac.ed.castle
- phone: (+44) 31 332 2117
-