Postnatal outcomes from prenatal ultrasound in LUTO
| Article DOI | https://doi.org/10.1002/uog.29129 |
| Objective | To predict postnatal outcome in fetuses with lower urinary tract obstruction (LUTO) using prenatal ultrasound findings |
| AI Approach | Random forrest |
| Data Source(s) | Single institutional series (125 patients) |
| Model Input | Gestational age (GA) at initial visit, indication for referral, fetal sex, suspected prenatal diagnosis and confirmed postnatal diagnosis, ultrasound features, genetic anomalies, maternal demographics and prenatal interventions |
| Model Outcome | Need for transplant Need for dialysis Death |
| Model Metrics | Predict transplant: AUC 0.65, 77% accuracy, 50% sensitivity, 80% specificity Predict death: AUC 0.78, 72% accuracy, 83% sensitivity, 67% specificity Predict dialysis: AUC 0.69, 71% accuracy, 70% sensitivity, 71% specificity |
| Model Usability | GitHub repository (https://github.com/larunerdman/PrenatalLUTO) with example data and script used for this project. |
AI = Artificial intelligence, LUTO = Lower urinary tract obstruction, AUC = Area under the receiver operator characteristic
AIPEDURO 
Leave a comment