Transient and persistent small-bowel intussusception in children: a decision tree analysis model based on ultrasound and clinical findings.

PURPOSE

To develop a systematic and efficient decision tree analysis (DTA) model to improve the diagnostic accuracy of transient small-bowel intussusception (TSBI) and persistent small-bowel intussusception (PSBI) in children.

METHODS

From February 2019 to June 2022, ultrasound (US) features and clinical findings of pediatric patients with small-bowel intussusception (SBI)-including SBI diameter, outer bowel wall thickness, thickness of the head and body of the intussusceptum, length of the intussusceptum, and presence of pathological lead points (PLPs)-were recorded and analyzed. A classification and regression tree algorithm was then used to develop a DTA model, which was trained and validated by randomly categorizing the patients into training (60%, 200/331) and validation (40%, 131/331) datasets to assess diagnostic performance.

RESULTS

A total of 331 patients with SBI (270 with TSBI and 61 with PSBI) were included; the maximum age was 9 years. The initial diagnostic predictor in the DTA model was the detection of a PLP via US, followed by intussusceptum length (P < 0.001). The sensitivity, specificity, and accuracy of the DTA model were 98.2%, 100%, and 98.6%, respectively.

CONCLUSION

The DTA model developed in this study facilitated the differential diagnosis of TSBI and PSBI in pediatric patients with SBI, with a clinical concordance rate of 98.6%.