Using autoregressive spectral analysis to investigate weaning index

Mechanical ventilation is widely used in lifesaving. But it is associated with numerous complications. Patients having weaning failure are about 7% to 19% and are subject to high mortality. To investigate this problem, we use a digital signal process approach. We record breathing data before and after weaning. The patients´ weaning success and weaning failure are also recorded. The autoregressive power spectral density method is used to quantify the breathing data and investigate its rhythm changes. From spectral analysis, the power ratio of the first two peaks in the low frequency regions is calculated. We observed that patients with weaning failure have this ratio less than 1. This ratio may be a good weaning index.