Artificial neural network based intracranial pressure mean forecast algorithm for medical decision support

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Medical Informatics

Background:

• Intracranial pressure (ICP) monitoring is vital for managing neurological conditions.
• Predicting future ICP mean is essential for proactive clinical intervention but remains an unaddressed challenge.
• Accurate ICP forecasting enables timely treatment adjustments to improve patient outcomes.

Purpose of the Study:

• To develop and evaluate a novel algorithm for forecasting the mean of intracranial pressure (ICP).
• To introduce the nonlinear autoregressive with exogenous input artificial neural network based mean forecast algorithm (ANN(NARX)-MFA).
• To compare the predictive performance of ANN(NARX)-MFA against a standard nonlinear autoregressive artificial neural network (ANN(NAR)).

Main Methods:

• Utilized a nonlinear autoregressive with exogenous input artificial neural network (ANN(NARX)-MFA).
• Incorporated features extracted from past ICP data and segmented sub-windows for prediction.
• Compared ANN(NARX)-MFA with a nonlinear autoregressive artificial neural network (ANN(NAR) without sub-window feature extraction).

Main Results:

• The ANN(NARX)-MFA algorithm demonstrated superior prediction accuracy compared to the ANN(NAR) algorithm.
• Feature extraction from finer segmented sub-windows significantly improved the capture of subtle ICP trend changes.
• Decomposition of data windows into sub-windows enhances predictive capabilities for future ICP trends.

Conclusions:

• The ANN(NARX)-MFA algorithm is effective for predicting future ICP mean.
• Sub-window feature extraction is a valuable technique for improving ICP forecasting accuracy.
• Advanced ICP forecasting facilitates proactive medical management, optimizing patient recovery and outcomes.