Abstract
OBJECTIVE: Blood-based biomarkers are valuable for investigating Autism Spectrum Disorder (ASD). This study aimed to identify key blood-based risk factors for ASD and to develop and validate a clinical model for predicting disease risk. METHODS: We retrospectively analyzed data from 879 children, including patients with ASD and healthy controls. After using propensity score matching to balance for age and sex, we employed multivariate logistic regression to identify independent risk factors from routine blood parameters, vitamins, and trace elements. We then characterized dose-response relationships and a nomogram for ASD risk prediction was constructed and internally validated. RESULTS: Four independent risk factors were identified: calcium, serum iron, vitamin D, and platelet distribution width (PDW). Lower levels of serum iron and vitamin D, along with higher PDW levels, were significantly associated with an increased risk of ASD. Calcium exhibited a non-linear relationship, with risk peaking at a concentration of 1.13 mmol/L. The nomogram based on these four markers demonstrated strong predictive performance, with sensitivities of 84.6% and 78.6% and specificities of 73.4% and 79.0% for the training and validation sets, respectively. CONCLUSION: These findings highlight the feasibility of calcium, iron, vitamin D, and PDW as independent blood-based risk factors for ASD predicting and elucidated their specific patterns of association with ASD prevalence. The nomogram ASD risk prediction model, constructed utilizing these key markers, demonstrated commendable discrimination, calibration, and clinical utility. This model holds considerable promise as an efficacious tool for early ASD screening and risk assessment, with significant potential for clinical translation.