JMIR Publications

ABSTRACT

Background:

: Spinal cord injury (SCI) causes substantial disability by disrupting spinal pathways, making functional independence a central rehabilitation goal. Traditional prognostic tools, notably the International Standards for Neurological Classification of Spinal Cord Injury (ISNCSCI), have limitations, including ceiling effects and misclassification during evaluation. Access to MRI, and neurophysiological measures is often restricted to specialized centers and early post-injury windows. Consequently, clinical prediction rules and conventional regression models—often focused on independent walking and less SCI-specific outcomes—may inadequately forecast overall independence, limiting goal setting and resource planning.

Objective:

To develop and validate machine learning (ML)-based rules that predict functional independence, as measured by SCIM-III, at 3, 6, and 12 months post-injury in individuals with SCI. The model combines clinical admission predictors, readily available to enhance predictive performance: age, sex, time since injury, upper and lower extremity strength (UEMS/LEMS), ISNCSCI data (ASIA Impairment Scale (AIS) grade and neurologic level of injury (NLI)), rehabilitation type and the trunk-control scale (ECT) score validated by the research team.

Methods:

Using retrospective electronic clinical records (2015–2026) from Mexico’s National Institute of Rehabilitation (INR LGII; a tertiary national rehabilitation reference center) for model development and a prospective independent cohort (January–December 2027) for external validation, the study will develop ML models to accurately and generally predict SCIM-III outcomes. Eligible participants are adults (≥18 years) with subacute or chronic SCI. Six architectures are compared: linear regression, CART, CatBoost, LightGBM, multilayer perceptron, and Gaussian process regression, using 10-fold stratified cross-validation for internal validation. Performance is assessed by RMSE, MAE, R², AUC, calibration plots, and decision curve analysis. The approach prioritizes reproducibility, interpretability, and clinical applicability, addressing prior models’ limitations (small sample sizes, limited validation, impractical input requirements). Key steps include data preprocessing, feature selection, model training with appropriate algorithms, and robust evaluation against existing prognostic benchmarks. External validation and ethical considerations are integrated, with commitments to data sharing where feasible.

Results:

The development cohort targets ≥500 retrospective records; the external validation cohort will prospectively register 100 records. Results will be reported upon study completion.

Conclusions:

This protocol describes the first Mexican study to develop a validated ML-based prediction rule for functional independence (SCIM-III) in SCI rehabilitation. A reliable, clinically actionable ML-based prediction tool that estimates SCIM-III trajectories to support goal setting, rehabilitation planning, and resource allocation in SCI care will be developed. Model performance will be benchmarked against published SCI prediction rules, and it will be considered clinically useful if it outperforms the mean-score baseline and existing rules on the prospective validation cohort. Clinical Trial: This study (SECIHTI CBF-2025-I-3891, INRLGII 128/25) is embargoed preregistered on the Open Science Framework (OSF; URL: osf.io/4fmy6), and on the Mexican national clinical registry (estudiosclinicos.salud.gob.mx No: INRLGII 128/25). Both registrations are completed prior to prospective data collection.