Machine Learning Introduces Electrophysiology Assessment as the Best Predictor for the Recovery Prognosis of Spinal Cord Injury Patients for Personalized Rehabilitation Approaches

The pronounced association between Evoked Potentials (EPs) and American Spinal Injury Association (ASIA) scores in individuals with Spinal Cord Injury (SCI) indicates that EPs may serve as dependable predictive markers for the progression of rehabilitation. Numerous studies have confirmed that variations in Somatosensory Evoked Potentials (SSEPs) demonstrate a relationship with ASIA scores, particularly during the early stages of the disease. Machine learning has witnessed a notable increase in significance within the medical field, primarily due to the increasing availability of health-related data and progressive enhancements in machine learning algorithms. It can be utilized to formulate predictive models that aid in disease diagnosis, anticipate disease progression, tailor treatment to fulfill individual patient needs, and improve the operational efficiency of healthcare systems. The strategic utilization of data can considerably elevate the quality of patient care, reduce healthcare costs, and promote the formulation of personalized and effective medical interventions. The healthcare industry reaps considerable benefits from the meticulous analysis of medical data, as it plays an integral role in promptly identifying patient diseases. Timely detection of a disease could contribute to effective symptom management and guarantee that appropriate treatment is provided. The present study aims to apply artificial intelligence techniques to identify predictors linked to the progression of SCI as assessed by the disability index, ASIA Impairment Scale (AIS), and final motor recovery. It is essential to clarify the role of Electrophysiological testing including SSEPs, MEPs and Nerve Conduction Studies (NCSs) in the prognostication of SCI. We analyzed empirical data obtained from a medical database consisting of 123 records. We developed an intelligent system that predicts the recovery of SCI utilizing machine learning algorithms, based on ensemble algorithms. More specifically it convolutes Decision Trees and Neural Network approaches usually resulting in better prediction accuracy. Throughout our experimental evaluation, SEPs achieved accuracies of 90%, which are comparable to full electrophysiology evaluation that obtained accuracies of 93%, and mostly better than MEPs and NCSs results for motor recovery prediction. Additionally, SEPs achieved an accuracy of 80%, which is close to the full electrophysiology evaluation that obtained an accuracy of 89%, and mostly better than MEPs and NCSs results for AIS scale determination. According to the previous results EPs could be established as the best predictors comparable to global electrophysiology assessment of SCI, resulting in more accurate efficacy than other diagnostic findings. Consequently, electrophysiology assessment should always be included, when available, as it elevates the total accuracy from only clinical investigation up to 93% (from at most 75%) for final motor recovery prediction, even for ASIA score determination, and consequently disease follow-up, up to 89% (from at most 66%). More data is needed to certify the above results. Further investigation is necessary to validate sensory electrophysiology assessment, which is significantly less expensive, portable, and simpler to administer than other prognostic tests, and more effective than clinical assessment methods such as AIS, in functioning as biomarkers for SCI scaling and prediction of recovery possibility as well as personalized rehabilitation planning. According to such findings, a Decision Support system can be developed as the objective alternative to the ASIA scale, using only sensory electrophysiology assessment.