Utilizing Predictive Analytics to Understand Neurogenic Bladder Symptom Score (NBSS) Variations in Adults With...

Area of Science:

• Urology
• Neurology
• Data Science

Background:

• Spinal cord injury (SCI) leads to varied bladder health outcomes.
• Predicting urinary symptom trajectories post-SCI is crucial for patient management.

Purpose of the Study:

• To utilize machine learning to identify key variables influencing urinary symptoms in individuals with SCI.
• To assess the predictive capability of these models for bladder health after SCI.

Main Methods:

• A Decision Tree analysis (eCHAID) was performed using 238 variables from the Neurogenic Bladder Research Group SCI registry.
• Primary outcomes included baseline Neurogenic Bladder Symptom Score (NBSS) and 1-year change in NBSS.

Main Results:

• Baseline NBSS was predicted by bladder management method and quality of life (QOL); suprapubic tube/urostomy users with good bowel QOL had the lowest scores.
• Female patients with spontaneous voiding had the highest baseline NBSS.
• Improvement in NBSS at 1 year was associated with specific medication use, infection history, and management methods in subgroups of male and female patients, with 57% predictive capacity.

Conclusions:

• Decision tree models can reveal correlations between patient characteristics and urinary symptoms post-SCI.
• The predictive accuracy of these models for future bladder symptoms remains limited.