Sudden infant death syndrome (SIDS) is the sudden death of an infant under one year old without any apparent warning signs. It is the leading cause of death of infants between the age of one month and one year in the developed countries. Previous studies have indicated that SIDS victims may suffer from cardiorespiratory failure, which reflects immaturity of autonomic nervous system (ANS) control. Heart variability (HRV) has been shown to be a noninvasive and inexpensive way to assess the ANS. Increased heart rate and decreased heart rate variability have been observed in SIDS victims. While it is unclear whether there is a relationship between SIDS and infants who have apparent life-threatening events (ALTE) like apnea and bradycardia, there is a desire to gain a better understanding of the nature of ALTE, their frequency and severity, and the ability of home monitors to detect events and provide an alarm. The Collaborative Home Infant Monitoring Evaluation (CHIME) study group, formed by NIH, sought to study these issues through the collection of a massive clinical database, which includes physiological, signals from 529 infants recorded in polysomnographic (PSG) studies. The infants were then studied several months using a home monitor which recorded life-threatening events. A classification is available regarding which infants had verified, apparent life-threatening events. Given the wealth of information linking SIDS and heart rate variability (HRV), we desire to investigate whether heart rate variability is effective predictor for life-threatening events. This research will investigate whether HRV can differentiate infants at risk for future apparent life-threatening events (FALTE). First statistical methods will be used to determine if there is significant difference in HRV between normal and FALTE infant groups. Next, the major goal is to design a method to classify infants using HRV, exploring various prediction models including logistic regression, decision tree, and neural networks. The relationship between HRV parameters and all the affected factors will also be studied to provide a clearer understanding of the parameters, and to help determine how to use HRV as a classifier. Successful prediction of infants with future apparent life-threatening events could have clinical significance and provide physicians with an opportunity for treatment.
