# Development of nomogram model based on LASSO-Logistic regression for predicting postoperative undernutrition to complex anorectal malformation: a pilot exploratory study

## TL;DR

This study creates a model to predict undernutrition after surgery for complex anorectal malformations, helping doctors intervene early.

## Contribution

A novel nomogram model using LASSO-Logistic regression is developed to predict postoperative undernutrition in complex ARM patients.

## Key findings

- Non-parental caregivers, preoperative undernutrition, systemic malformations, and artificial feeding are significant predictors of postoperative undernutrition.
- The nomogram model demonstrated high accuracy with an AUC of 0.877 and good calibration.
- The model provides net benefit for clinical decision-making when predicting malnutrition probability above 0.03.

## Abstract

Nutritional problems in patients with anorectal malformation (ARM) after anorectoplasty have not received sufficient attention. This study aimed to establish prediction model of postoperative undernutrition for patients with complex ARM.

Retrospective review of 104 patients with complex ARM was conducted with assessments of clinical data. Lasso-Logistic regression analysis was used to identify independent factors, and then establish a Nomogram model for predicting postoperative undernutrition. Harrell’s concordance index and calibration curves were applied to evaluate the performance of this model. R software was used for statistical analysis.

Of the 104 patients, the proportion of malnutrition was 28.85% (30/104). Lasso-Logistic regression analysis showed that non-parental caregivers (Odds ratio [OR]: 11.20), undernutrition at anorectoplasty (OR: 4.101), surgery for other systemic malformation (OR: 8.378), and feeding method (artificial, OR: 25.320) were independently associated with postoperative undernutrition (all P < 0.05), and Nomogram model was developed based on these determinants. The area under the receiver operator characteristic curve of this cohort was 0.877 and it still was 0.906 through bootstrapping validation (n = 1000). The H-L goodness-of-fit test showed that there was no significant difference between the predicted and actual incidence of postoperative undernutrition (χ2 = 7.55, P = 0.273). Meanwhile, the calibration curves indicated that the forecast was in good agreement with the actual situation. Decision curve showed that when the probability threshold of Nomogram model predicting malnutrition was >0.03, application of the model would add net benefit compared to either the treat-all strategy or the treat-none strategy.

This Nomogram model for predicting postoperative undernutrition of complex ARM patients showed desirable accuracy and discrimination for clinicians, aiding the early initiation of preventative interventions for undernutrition.

The online version contains supplementary material available at 10.1186/s12876-025-04202-5.