Renal scarring in lower urinary tract dysfunction, Misc.
| Article DOI | https://doi.org/10.28982/josam.691768 |
| Objective | To predict renal scarring in children with LUTD |
| AI Approach | 7 models were tested: ANN and XGB were superior |
| Data Source(s) | Single institutional series (75 patients) |
| Model Input | Clinical Features (episodes of symptomatic UTI, presence of VUR, bladder trabeculation, bladder wall thickness, catherization), VUDS Features (post-void residual volume, bladder volume, wall thickness, compliance, detrusor hyperactivity, reduced bladder capacity, detrusor leak point pressure, DMSA Features (differential function of less than 40%, presence of renal scarring and/or atrophy) |
| Model Outcome | Presence of renal scarring |
| Model Metrics | SMOTE data: AUROC 0.90 and 0.88 for ANN and XGB sensitivity 88% and 82% for ANN and XGB; specificty 94% and 93% for ANN and XGB. |
| Model Usability | NA |
AIPEDURO 
Leave a comment