Online ICP forecast for patients with traumatic brain injury

Traumatic brain injury (TBI) endangers many patients and lays great burden on the neural intensive-care units in the whole world. To improve the outcome of TBI patients, it is desirable to forecast the intracranial Pressure (ICP) so to enable timely or early interventions to control the ICP level. Past research mainly focused on ICP pulse morphology analysis and ICP waveform forecast, but results were not satisfactory. In this paper, to forecast ICP continuous trends, we propose an autoregressive integrated moving average (ARIMA) ICP forecast online application with orders selection predicated on autocorrelation function (ACF) and partial autocorrelation function (PACF). Results show that the accuracy of ICP forecast improves significantly with our forecast model, compared with ARIMA based on Akaike information criterion (AIC) and artificial neural network approach. Besides, the forecast processing time of ARIMA model predicated on PACF and ACF is much shorter than ANN and ARIMA predicated on AIC.