Sukjoon Kim

Prediction of the oculocardiac reflex from pre-operative linear and nonlinear heart rate dynamics in children.

This study was aimed to determine whether pre‐operatively measured linear and nonlinear analysis of heart rate variability might predict the occurrence of the oculocardiac reflex (OCR) or other arrhythmia during strabismus surgery in children (n = 185, mean (SD) age = 8.09 (3.31) years). We compared time‐ and frequency‐domain, and nonlinear dynamic indices of pre‐operatively measured RR interval data between the OCR‐positive group (maximum heart rate decrement = 20 beat.min−1 during the traction of extraocular muscle, n = 54), OCR‐negative group (< 20 beat.min−1, n = 78) and arrhythmia‐positive group (all other arrhythmias; n = 53). pNN50, rMSSD, high‐frequency power and nonlinear prediction error were significantly lower in the OCR‐positive and arrhythmia‐positive groups than in the OCR‐negative group. Discriminant analysis using these indices could correctly identify 39/54 (72.2%) OCR‐positive patients. Some pre‐operatively measured indices of linear and nonlinear heart rate variability, especially when used in combination, are valuable for predicting significant bradycardia during strabismus surgery in children.

Combining locally trained neural networks by introducing a reject class

This paper presents a new strategy for building and combining a local committee when a dataset is given. Training local committees is performed in two stages: active data partitioning and recombination by introducing an additional reject class. Active data partitioning is a preprocessing step that partitions the given dataset into several similar subsets using active learning. Additional reject class in this strategy plays an important role in assigning a focused area to each individual network of the committee. For combining the outputs of each individual network, we use a kind of sum rule criteria, assuming that the outputs of the individuals are equivalent to a posteriori Bayesian probabilities. All the learning procedures are based on the active learning paradigm. Experiments are performed on the two real-world datasets from the UCI machine learning database. The results show that the active data partitioning and recombining strategy is very successful for building a local committee and the combined result outperforms other algorithms, but the combined result can be affected by the training error level /spl epsiv/.