Development and external validation of a LASSO-based parsimonious nomogram for predicting BPPV recurrence: a...

Benign Paroxysmal Positional Vertigo (BPPV) exhibits high recurrence rates post-repositioning. This study aimed to identify core risk factors and develop a parsimonious, externally validated nomogram to personalize recurrence risk stratification.

We conducted a multi-center retrospective cohort study involving BPPV patients from two medical centers (2020-2025). The Least Absolute Shrinkage and Selection Operator (LASSO) and stepwise multivariate logistic regression were employed to identify predictors and construct the model. Performance was evaluated using the Area under the Curve (AUC), calibration plots, Decision Curve Analysis (DCA), and Clinical Impact Curves (CIC).

Among the participants, the 1-year recurrence rate was 23.97%. A parsimonious model comprising three core variables: Diabetes Mellitus (OR = 11.42), Non-posterior canal involvement (OR = 4.17), and 25-hydroxyvitamin D deficiency [25 (OH) D] (OR = 3.76), was established. The model demonstrated good discrimination, with an AUC of 0.905 in the training set, 0.872 in the internal validation set, and 0.853 in the external validation set. Calibration curves showed excellent agreement between predicted and observed probabilities. DCA and CIC confirmed significant net clinical benefit across a wide range of threshold probabilities (0.05-0.76), outperforming default clinical strategies.

We developed and externally validated a highly accurate nomogram integrating metabolic, anatomical, and nutritional markers. This simplified tool facilitates early identification of high-risk patients and supports targeted secondary prevention in clinical practice.