The research, published in Physiological Reports, tackled a problem that has long frustrated both clinicians and preclinical scientists. Standard practice after spinal cord injury prioritizes medical stabilization and catheterization, pushing comprehensive urodynamic testing weeks or months down the road. By the time detailed bladder measurements are taken, a critical therapeutic window may already have closed. Compounding the problem, the bladder is not a single organ behaving in a single way; it is a synchronized neuromuscular system whose output depends on coordinated interactions among pressure, volume, timing, and filling rate. Reading any one of those numbers in isolation, the study’s authors argue, misses the multivariate signature that actually carries prognostic information.

To capture that signature, the team, led by Zachary Danziger of Florida International University, built an unusually rigorous monitoring setup. Twenty adult female Sprague-Dawley rats received bladder catheters; ten of them then underwent complete transection of the spinal cord at the ninth thoracic segment. All animals were housed in custom metabolic cages fitted with laser-etched wire mesh floors that let urine pass instantly to digital scales while deflecting solid waste, allowing around-the-clock automated recording of every voiding event. Implanted catheters connected to pressure transducers sampled intravesical pressure at 40 hertz. The result was a continuous, uninterrupted record of bladder pressure and voided volume for roughly 26 days after injury, spanning more than 5,000 voiding events across the group, a dataset with no real precedent in preclinical neurogenic bladder research.

The longitudinal data told a story of gradual but incomplete recovery. After the initial spinal shock phase, spinally mediated voiding reflexes re-emerged, and average voiding efficiency crept upward while post-void residual volumes declined. But even after a month, the injured group as a whole remained significantly worse than intact controls, with lower voiding efficiency and higher residual urine. More striking was the spread between individual animals. When the researchers applied an unsupervised clustering algorithm called DBSCAN to chronic-phase measurements of voiding efficiency and residual volume, three distinct recovery phenotypes emerged from the ten injured rats: a good recovery group of four animals reaching efficiency near 0.71 with residuals around 0.21 milliliters, a moderate group of five animals at roughly 0.45 efficiency and 0.78 milliliters of residual, and a single poor-recovery animal at 0.34 efficiency with nearly two milliliters retained after each void.

Crucially, even the best-recovering animals had not returned to normal. Intact rats showed clear day-night differences in voiding, emptying more efficiently during their rest phase, a pattern partly governed by circadian regulation of vasopressin and urine production. In every injured rat, regardless of phenotype, that diurnal variation was abolished and never returned over the four-week observation window. The good-recovery animals had achieved efficient emptying, but they did so through an involuntary spinal reflex circuit that resembles the primitive neonatal pattern of micturition rather than the supraspinally controlled system that existed before injury. Efficient voiding, in other words, can mask a fundamental change in the mechanism producing it.

The heart of the study lies in its predictive analysis. The researchers asked whether measurements taken in the first week after injury, days 3 through 10, could forecast which chronic phenotype each rat would occupy. Using a generalized linear model with a logit link function, constrained to at most three terms to avoid overfitting, they searched across thousands of candidate predictor combinations, including squared terms and interactions among six acute-phase urodynamic parameters. The winning model combined two interaction terms: the product of acute voided volume and acute voiding efficiency, and the product of acute ureter flow rate and the square of acute voiding efficiency. On the fitted data, this compact equation explained more than 90 percent of the variance in chronic voiding efficiency, and leave-one-out cross-validation, in which the model was retrained ten times with one rat withheld each time, confirmed that it generalized, achieving an R-squared of 0.84 on unseen animals.

That predictive power came from interactions rather than any single measurement, and this may explain why conventional clinical assessment has struggled to prognosticate early. During the acute phase, the urodynamic parameters of rats destined for good, moderate, or poor recovery were statistically indistinguishable when compared one at a time. Yet a principal components analysis revealed that each animal’s starting position along the dominant axis of multivariate variation, a dimension dominated by residual volume, efficiency, and filling rate, correlated with its eventual phenotype. Individual parameters told no story; the constellation of parameters told nearly the whole one. The finding suggests that redundancy in the urodynamic parameter space is considerable, which is good news for translation: a carefully chosen subset of clinically accessible measures might suffice for prediction without exhaustive monitoring.

The study also uncovered a temporal split in how bladder function reorganizes. Storage parameters, bladder capacity and ureter flow rate, established phenotype-specific values within the acute phase and then remained essentially stable for the rest of the month. Voiding parameters, by contrast, evolved continuously across weeks, consistent with progressive reorganization of spinal reflex circuitry. The authors interpret this split as evidence that two processes on different timescales shape outcome: rapid structural and molecular changes in the bladder wall, including epithelial barrier failure and inflammatory remodeling that begin within hours of injury, and slower neural plasticity in the spinal cord. Acute ureter flow rate, which reflects renal filtration rather than bladder physiology, varied systematically across phenotypes, hinting that systemic physiological differences at the time of injury may also contribute to prognosis through mechanisms that remain to be determined.

Pressure dynamics added a further layer. Non-voiding contractions, transient pressure rises unaccompanied by urine output, separated the phenotypes at every phase of recovery. Good-recovery animals averaged fewer than one such contraction per filling interval in the acute phase, while moderate-recovery animals averaged more than eight, a difference that persisted through the intermediate and chronic phases. Maximum voiding pressure diverged later, with good-recovery animals showing significantly lower pressures by the intermediate phase. Because inter-void intervals did not differ between groups, the contraction-count differences reflect genuine differences in detrusor instability rather than differences in how long the bladder had to misbehave.

The authors are careful to frame the work as preliminary and hypothesis-generating. Only female rats were studied to limit biological variability, and known sex differences in post-injury bladder dysfunction mean that validation in male and mixed cohorts is essential. Continuous metabolic cage housing also precluded simultaneous recording of external urethral sphincter activity, leaving detrusor-sphincter coordination unverified electrophysiologically. Still, the conceptual advance is substantial: this is, to the authors’ knowledge, the first continuous longitudinal characterization of post-injury bladder reorganization in a rat model, and it demonstrates that chronic outcome is not determined by injury level alone but by patterns of multivariate urodynamic interaction established early in recovery. If those patterns can be captured with clinically practical measurements in human patients, the window in which the spinal cord and bladder are most adaptable could finally become a window in which clinicians act, directing antispasmodic therapy, catheterization strategy, or emerging neuromodulation approaches to the patients who need them most, before irreversible secondary damage takes hold.

Subject of Research: Prediction of chronic neurogenic bladder recovery after spinal cord injury using early longitudinal urodynamics and machine learning in a rat model

Article Title: Early urodynamic profiles predict lower urinary tract recovery after spinal cord injury in female rats

Article References: Afrashteh, B., Khan, S., Adury, R. Z., Angoli, D., & Danziger, Z. C. (2026). Early urodynamic profiles predict lower urinary tract recovery after spinal cord injury in female rats. Physiological Reports, 14(17), Article e71101. https://doi.org/10.14814/phy2.71101

Image Credits: AI Generated

DOI: 10.14814/phy2.71101

Keywords: spinal cord injury, neurogenic bladder, urodynamics, voiding efficiency, machine learning, generalized linear model, rat model, bladder recovery, micturition, DBSCAN clustering, principal components analysis, Physiological Reports