A machine learning-based predictive framework for head impact injuries
Year: 2025
Type: article
Abstract: Modern technologies have rapidly developed in recent years, particular in the field of artificial intelligence (AI). Nascent technique has been adapted by researchers in their respective research fields to leverage the research outcome. Several research fields have achieved significance breakthroughs by embracing AI to address challenges. Machine learning (ML), a prominent subset of AI, is a data-driven method that enables the discovery and establishment of intricate relationship within datasets. Training ML models require large amounts of high-quality data to effectively capture and comprehend the underlying data relationships. Once adequately trained, these AI models can be deployed in real-life applications without the need for extensive computational power or resources. This research paper introduces the utilization of artificial neural networks (ANNs) to predict intracranial pressures using kinematic linear accelerations as inputs only. The proposed model incorporates three kinematic inputs (ax, ay, az) and time as inputs to the ANN. The developed ANN model exhibits the ability to predict intracranial pressure values for five distinct brain regions, representing the front, rear, left, right and rear sections of the brain. Both feedforward and time series ANNs were trained using 80 samples with 177 time-steps data. The ANN model successfully predicted intracranial pressure from head kinematics, with a correlation coefficient (R = 0.98) and root mean square error (RMSE = 0.008MPa). These findings highlight the efficacy of ANNs in accurately predicting intracranial pressure based on linear kinematic measurements.
Source: Results in Engineering
Authors: Jing Yi Heng, Rifai Chai, Saeed Mouloodi, Raj Das, Kwong Ming Tse
Language: English
Subfield: Neurology
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $1,940
OpenAlex ID: W7083027334