Neural Net Applications in Medicine — Exploration of Dynamical Thermoprofiles —

Summary

The possibility of automatical or semi-automatical judgement and exploration of thermoprofiles from thermoregulation diagnostics developed by Schwamm & Rost /1/ with the aid of computer-simulated so-called neural nets, which work as pattern classifiers, will be presented and discussed. The simulation algorithm which encodes and sorts patterns due to its similarities into a two-dimensional layer of neurons will be presented with examples of thermoregulation diagnostic diagrams. With a few limitations it is possible to use the self-organizing feature map algorithm developed by Kohonen as an investigation tool to unlock the problem of mapping onto each other between thermoregulation diagnostic classes and clinical diagnostic classes.