Combining ROC Curves to Improve Diagnostic Values

Afaf El-Ansary,Manan Alhakbany,Hanan A. Alfawaz,Laila Y. Al-Ayadhi
DOI: https://doi.org/10.3390/jcm13226695
IF: 3.9
2024-11-08
Journal of Clinical Medicine
Abstract:Background/Objectives: Autism spectrum disorder (ASD) is a neurodevelopmental disorder that is increasing throughout the world. Nevertheless, no specific diagnostic or even risk assessment marker is available. Combining more than one marker can improve the diagnostic value of each marker alone and maximize the AUC for ROC curve. Hedgehog (Hh) signaling modulates both intestinal inflammation and immunity. Cyclooxygenase-2 (COX-2) is required for inflammation, and it has been related to epileptic seizures. COX-2 generates prostaglandins-EP2 (PGE2) in the brain, which plays a major role in neuropathology and epilepsy. This study aims to understand the pathophysiology of ASD by investigating the diagnostic value of COX-2 and IHh using independent and combined ROC curves. Methods: COX-2 and IHh were measured in 40 children with ASD and 41 age and sex-matched controls using ELISA. Statistical analyses were performed using ROC curves, odds ratios, and multiple logistic regression models. Results: Higher levels of COX-2 and IHh were observed in the plasma of patients with autism than in controls. The ROC curve of HIh and COX-2 independently showed poor and fair AUCs of 0.662 and 0.776, respectively, while the combined ROC for both variables in relation to the control group increased the AUC = 0.831 significantly (p < 0.001). Conclusions: Thus, combining these variables could be a useful diagnostic marker for ASD.
medicine, general & internal
What problem does this paper attempt to address?