The clinical profile of a selected group of 26 young competitive athletes who survived a documented cardiac arrest (CA) was presented by the Arrhythmologic Centre of Trento in (Bertoldi et al, 1995). Computer simulations on the Trento database were performed to estimate a classification tree (CART) model. The goal of this preliminary study is the development of a method to identify high risk patients (those who may undergo a sudden cardiac death (SCD) in the follow-up period). Model construction was based upon an automatic screening of the most relevant predictor variables in terms of misclassification error and tree complexity. The data used up to 12 response variables to estimate the probability of adverse outcome (non-fatal recurrency or SCCD) in the follow-up period. To evaluate the accuracy of the model for novel data, error rate was estimated by computer simulations based on a novel bootstrap procedure. The model indicated the type of silent underlying cardiopathy (ARVD, DCM, IHD, MVP) and the therapy configuration as predictive of recurrency risk after a first CA
Young Competitive Athletes Resuscitated from Cardiac Arrest on Field: Exploratory Data Analysis and Computer Simulations
Furlanello, Cesare;Merler, Stefano;
1997-01-01
Abstract
The clinical profile of a selected group of 26 young competitive athletes who survived a documented cardiac arrest (CA) was presented by the Arrhythmologic Centre of Trento in (Bertoldi et al, 1995). Computer simulations on the Trento database were performed to estimate a classification tree (CART) model. The goal of this preliminary study is the development of a method to identify high risk patients (those who may undergo a sudden cardiac death (SCD) in the follow-up period). Model construction was based upon an automatic screening of the most relevant predictor variables in terms of misclassification error and tree complexity. The data used up to 12 response variables to estimate the probability of adverse outcome (non-fatal recurrency or SCCD) in the follow-up period. To evaluate the accuracy of the model for novel data, error rate was estimated by computer simulations based on a novel bootstrap procedure. The model indicated the type of silent underlying cardiopathy (ARVD, DCM, IHD, MVP) and the therapy configuration as predictive of recurrency risk after a first CAI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
