Determination of Diagnosis and Prognosis in Spinal Cord Injury Using Machine Learning.

Area of Science:

• Neurology and Biomedical Engineering
• Artificial Intelligence in Medicine

Background:

• Traumatic spinal cord injury (tSCI) results in significant long-term disability and economic burden.
• Accurate diagnosis and prognosis are critical for effective tSCI management and patient rehabilitation.
• Limited treatment options highlight the need for advanced diagnostic and prognostic tools.

Purpose of the Study:

• To evaluate the potential of artificial intelligence (AI) and machine learning in improving the diagnosis and prognosis of tSCI.
• To explore the application of AI-based models in predicting clinical outcomes for tSCI patients.

Main Methods:

• Convolutional neural networks (CNNs) trained on magnetic resonance imaging (MRI) and diffusion tensor imaging (DTI) for diagnosis.
• Various prognostic models, including logistic regression, neural networks, and deep learning-based radiomics, were applied.
• Analysis of AI model performance in detecting cord damage, classifying injury severity, and predicting functional recovery.

Main Results:

• AI models, particularly CNNs, demonstrated high accuracy in diagnosing cord damage and injury severity from MRI and DTI.
• AI-based prognostic models showed improved prediction accuracy for functional recovery, ambulatory status, and survival compared to conventional methods.
• Identified limitations include small dataset sizes, study heterogeneity, and a lack of external validation.

Conclusions:

• AI and machine learning show significant promise for enhancing diagnostic accuracy and prognostic capabilities in tSCI.
• Addressing limitations through multicenter collaborations and multimodal data integration is essential for clinical generalizability.
• AI is poised to become a crucial tool supporting clinical decision-making and rehabilitation strategies for tSCI patients.