Abstract: Objective To develop an auxiliary assessment model for true umbilical cord knots based on three-dimensional high-definition flow volume imaging(3D HD-Flow)and to preliminarily evaluate its complementary discriminative value in singleton pregnancies with suspected umbilical cord knots on two-dimensional ultrasound. Methods This prospective study included 44 singleton pregnant women with suspected umbilical cord knots detected by two-dimensional ultrasound at Jinan Maternal and Child Care Hospital from January 2022 to August 2025. Using 3D HD-Flow, 15 sonographic features were extracted, including closed-loop integrity, vascular traversing, umbilical cord coiling index and hemodynamic parameters.Variables were screened through univariate analysis, and a diagnostic model was constructed using Firth penalized likelihood logistic regression. Model performance was evaluated using receiver operating characteristic curve analysis with bootstrap-corrected area under the curve(AUC). Calibration curve was plotted to assess model fit, and leave-one-out cross-validation(LOOCV)was performed for internal validation. Results Among the 44 cases, 18(40.9%)were confirmed as true knots and 26(59.1%)as false knots. Univariate analysis identified four significant features: closed-loop integrity score, vascular traversing score, umbilical cord coiling index, and umbilical artery pulsatility index at the knot. The umbilical artery pulsatility index at the knot showed significant intergroup difference in univariate analysis(true knot group: 1.12±0.45 vs. false knot group: 0.90±0.15, P=0.024). Multivariable Firth logistic regression revealed that closed-loop integrity(OR=3.49, 95%CI: 1.54-13.76, P=0.001 8)and vascular traversing(OR=14.39, 95%CI: 1.75-456.33, P=0.010 7)were independent predictors. The model achieved an AUC of 0.988(bootstrap-corrected AUC=0.985, 95%CI: 0.961-0.999), with LOOCV-derived AUC of 0.981(95%CI: 0.950-1.000). The calibration curve demonstrated that the predicted probabilities were higher than the observed probabilities, indicating an overestimation tendency. The sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were 94.4%, 100%, 97.6%, 100%, and 95.8%, respectively. Inter-observer agreement for both core features was excellent(intraclass correlation coefficient: 0.92 and 0.88, respectively). Conclusion The 3D HD-Flow-based imaging features demonstrate good auxiliary assessment value in singleton pregnant women with suspected umbilical cord knots on two-dimensional ultrasound, and can serve as a complementary assessment tool for this population.

Key words: True umbilical cord knot, Three-dimensional high-definition flow imaging, Auxiliary assessment model, Prenatal ultrasound, Logistic regression