Section: Software

Calicot : intelligent cardiac monitoring

Participant : René Quiniou [ Correspondant ] .

Calicot (Cardiac Arrhythmias Learning for Intelligent Classification of On-line Tracks) is a software that takes as input several signals coming from sensors and that delivers as output fault states or diseases that were diagnosed by recognizing characteristic temporal patterns on the monitored signals. Calicot is devoted to monitoring cardiac patients and diagnosing cardiac arrhythmias.The software is mainly implemented in Java with a few modules in Prolog. The main features of Calicot are:

• a base of signal processing algorithms for abstracting signals into time-stamped symbolic events,

• a base of chronicles that are used on line by the chronicle recognizer(CRS (Chronicle Recognition System) from France Telecom R & D.) Chronicles are discriminant temporal patterns related to arrhythmias. They are learned automatically off line from signal examples by using ILP techniques,

• a pilot that adapts the behavior of the system to the monitoring context: noise on signals, patient's state, relevant arrhythmias, etc,

• a graphical interface that displays the recognized patterns on the signal curves and shows the related diagnoses.

The software prototype has been completely redesigned this year and offers new functionalities. The following website gives many details about the goals, the conception and the implementation of Calicot: http://www.irisa.fr/dream/Calicot/ .