Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10 5/3/83; site eosp1.UUCP Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!eagle!allegra!princeton!eosp1!mcmillan From: mcmillan@eosp1.UUCP (Tobias D. Robison) Newsgroups: net.ai Subject: Re: computer ECG Message-ID: <621@eosp1.UUCP> Date: Mon, 27-Feb-84 10:13:03 EST Article-I.D.: eosp1.621 Posted: Mon Feb 27 10:13:03 1984 Date-Received: Tue, 28-Feb-84 07:45:30 EST References: <2043@ecsvax.UUCP> Organization: Exxon Office Systems, Princeton, NJ Lines: 39 I worked in the field of ECG analysis in the mid 1970's. ECG analysis programs (such as Bonner's) have a very hard time doing the necessary pattern recognition on ECG wave forms. Once they have approximately recognized wave types, they do MUCH better than any human being at quantifying the characteristics of abnormal wave forms. In 1974, the Bonner program was already considered more accurate than good human cardiologists in certain respects, and less likely to pronounce a good heart bad (or vice versa) athan average cardiologists. Obviously when there is an abnormality, the ECG program output will be carefully overread by a cardiologist. In this process, the ability of the human to recognize patterns, plus the quantifying data produced by the software and hardware produce excellent results. ECG systems can be seriously effected by signal noise (even 60 hertz AC noise). In the 1970's I was aware of at least one case where an ECG bureau really should have shut down, rather than supply ECg reports that it knew were seriously compromised by noise-sensitive ECG hardware. ECG systems developed at a culturally suitable time. The middle-aged cardiologists still like to make extra money by overreading routine ECG charts. Younger cardiologists usually don't care to spend their time this way and appreciate the time saved by ECG systems. In general, a cardiologist is always required to check an ECG system's report, to avoid gross machine error. In the 1970's, no company really made money from ECG analysis systems. The market was extremely sensitive to new technologies. No manufacturer could simply deliver working systems. All were required to adopt larger storage devices, faster minicomputers, improved analysis software, and ancillary software for report editing, databases of reports, comparing ECG's on the same patient, etc. The required rate of development in order to do business killed almost everyone financially. - Toby Robison allegra!eosp1!robison decvax!ittvax!eosp1!robison princeton!eosp1!robison (NOTE! NOT McMillan; Robison.)