Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!watmath!clyde!floyd!harpo!seismo!hao!hplabs!sri-unix!FISCHER@RUTGERS.ARPA From: FISCHER@RUTGERS.ARPA Newsgroups: net.ai Subject: Re: computer ECG, FDA testing of AI programs Message-ID: <17229@sri-arpa.UUCP> Date: Fri, 2-Mar-84 20:54:42 EST Article-I.D.: sri-arpa.17229 Posted: Fri Mar 2 20:54:42 1984 Date-Received: Fri, 9-Mar-84 01:23:15 EST Lines: 34 From: RonApparently because of fierce competition, much current information, particularly with regard to algorithms, is proprietary. Worst in this regard (a purely personal opinion) is HP who seems to think nobody but HP needs to know how they do things and physicians are too dumb to understand anyway. ... They offer an advantage to small hospitals by offering verification of the analysis by a Cardiologist (for an extra fee). What the latter seems to say is that the responsibility for accepting the diagnosis is that of the local cardiologist. I cannot see a responsable doctor examining a few runs of a program's output and proclaiming it "correct." A hedge against complaints of computers taking over decision making processes from human has been that we can look at the algorithms ourselves or examine the reasons that a system concluded something. If this information becomes proprietary the government will probably license software for medical purposes the way the FDA does for new drugs. Imagine a testing procedure for medical diagnostic AI programs that is as expensive and complicated as that for testing new drugs. (ron) [Ron makes a good point. As a side issue, though, I would like to mention that H-P has not been entirely secretive about its techniques. On March 8, Jim Lindauer of H-P will present a seminar at Stanford (MJH 352, 2:45PM) on "Uses of Decision Trees in ECG Analysis". -- KIL]