PECG

Researchers: Tibor Gyimóthy Ph.D; Zoltán Alexin; Gabriella Kókai

Overview

The PECG system integrates an ECG waveform classifier with an interactive learner (called IMPUT).

The PECG system contains three main parts: the ECG processing, the IDT and graphic viewer modules. It is based on an attribute grammar approach of ECG waveform analysis published by Skordalakis.

Skordalakis's system can classify the recognized ECG waveforms according to certain synthesized attributes. It contains a preprocessing part which produces a linguistic representation from an ECG signal. This linguistic representation is used as the input for the PECG system.

The basic idea of PECG is to integrate the ECG classifier program (implemented in Prolog) with the IDT (Interactive Diagnosis and Testing) algorithmic debugger module and a graphic viewer. This integrated tool can recognize if any modification in the classifier program is needed. If the system cannot analyze the input then the user is helped by the built-in IDT debugger to find the false clause.
The graphic viewer module helps the users to decide whether the program with a given input works correctly or not, by displaying the steps of debugging phase.

The learning part of the PECG based on an Inductive Logic Programming (ILP) method called IMPUT.

The new method involves an integration of PECG with the IMPUT interactive learning system. With the help of this system not only the place of the error can be located in the program but in many cases starting from a {/em buggy} Prolog program we can infer a {/em correct} program using an unfolding transformation and the IDT debugger. Applying the IMPUT method, the extended PECG system is able to suggest a correct solution to the user to replace the buggy clause recognized during the debugging process.

References

Kókai, G., Alexin, Z., Gyimóthy, T.:
Classifying ECG Waveforms in Prolog
in Proc. of the Fourth International Conference on the Practical Application of Prolog London, UK. 23rd-25th April 1996 173-201
Kókai, G., Alexin, Z., Gyimóthy, T.:
Analyzing and Learning ECG Waveforms
in Proc. of The Sixth International Workshop on Inductive Logic Programming (ILP'96) Stockholm, Sweden 28-30 August, 1996 152-171 and in Lecture Notes in Artificial Intelligence subseries of Lecture Notes in Computer Science series Vol 1314 127-145 (1997) Springer Verlag
Kókai, G., Alexin, Z., Gyimóthy, T.:
Learning Biomedical Patterns
INAP-96 The 9th Exhibition and Symposium on Industrial Applications of Prolog 16-19 October 1996 Hino, Tokyo Japan 159-168
Kókai, G., Alexin, Z., Gyimóthy, T.:
Application of Inductive Logic Programming for Learning ECG Waveform
AIME'97 The 6th Conference on Artifical Intelligence in Medical Europa, 23rd - 26th March 1997 Grenoble, France, 126-130 proceeding was published in Lecture Notes in Artificial Intelligence 1211 subseries of Lecture Notes in Computer Science
Kókai, G., Alexin, Z., Gyimóthy, T.:
Learnig ECG Grammar
The Fifth Symposium on Programming Languages and Software Tools Jyvaskyla, Finland, June 7-8, 1997, 84-95
Kókai, G., Csirik, J., Gyimóthy, T.:
Learning the Syntax and Semantic Rules of an ECG Grammar
AI*IA'97 Fifth Congress of the Italian Association for Artificial Intelligence Roma 16-19 September 1997 proceeding will be published in Lecture Notes in Artificial Intelligence subseries of Lecture Notes in Computer Science
Nilson, J., Kókai, G.:
A Graphical User Interface for an ECG Classifier System
21. Deutsche Jahrestagung für Künstliche Intelligenz Freiburg, 9-12 September 1997 proceeding will be published in Lecture Notes in Artificial Intelligence subseries of Lecture Notes in Computer Science, LNAI 1303, pages 397-400, Springer-Verlag, September 1997.