High-risk newborn infants in an intensive care unit often suffer significant morbidity and mortality because of infectious illnesses which elude early diagnosis and are present in advanced stages with circulatory shock. Earlier diagnosis and therapy might present in advanced stages with circulatory shock. Earlier diagnosis and therapy might prevent or reduce late complications and deaths. A strategy for early detection of impending catastrophic events should reduce mortality as well as hospital costs. Our proposed research is based on the fact that the time between successive heartbeats varies incessantly. In newborn infants, this heart rate variability is reversibly reduced during severe illness. During this program, this heart rate variability is reversibly reduced during severe illness. During this program, the heart rate variability of hospitalized, at-risk newborn infant will be monitored to test the hypothesis that a decrease in heart rate variability presages a catastrophic infectious illness. Preliminary research supports this hypothesis. heart rate variability data will be processed using an innovative multi-linear analysis technology: polynomial networks. Polynomial networks have been successfully applied to many data processing challenges in the fields of health care, financial modeling, and pattern recognition. This will result in clinically useful indices to quantify heart rate variability. The long- term objective of this effort is to test the hypothesis that monitoring heart rate variability will lead to earlier diagnosis and more effective therapy of catastrophic infectious illnesses in newborn infants, and thus be useful for a variety of clinical practices.
