Chaotic ECG analysis using combined models

Tools derived from chaos theory have proven useful in the analysis of medical data, especially in cardiology. These tools are particularly helpful in analyzing biomedical signals, such as electrocardiograms, electroencephalograms, and other time series data arising from applications such as hemodynamic studies. In the work described here, chaotic and clinical parameters are combined to develop multiple models. These models are then intersected to try to minimize false positives and false negatives. The chaotic parameters are derived from a new theoretical approach to chaotic analysis. The neural network Hypernet is used to combine chaotic and clinical variables. The methodology is illustrated in a model which attempts to identify the presence of congestive heart failure.