Artificial Neural Networks Show Complex Interplay Among Risk Factors Related to Pregnancy, and Peri and Post Natal Period That May Contribute to Autism: A Pilot Study

Background: Autism Spectrum Disorder (ASD)  is a multi-factorial disease, where a single risk factor  unlikely can provide comprehensive information. Moreover, due to the non-linearity of biomarkers, traditional statistic is often unsuitable and underpowered to dissect their relationship. Recent epidemiological studies have pointed out a number of pregnancy and peri-post natal factors which, contributing to focal brain inflammation, predispose to ASD development.  

Objectives: The aim of the study is to assess the prevalence and natural relationships among 23 potential risk factors in pregnancy history and peri and post natal events in a group of 45 autistic children in comparison with 54 typical children. 

Methods: Traditional statistics (Principal Component Analysis-PCA) and  Artificial Neural Networks (Auto-CM system) were applied to highlight the  associations among variables under study. Auto-CM is a special kind of Artificial Neural Network developed at Semeion Research Institute( Rome) and successfully applied in many complex chronic degenerative diseases, able to find out consistent trends and associations among variables creating a semantic connectivity map. The matrix of connections, visualized through minimum spanning tree filter, takes into account nonlinear associations among variables and captures connection schemes among clusters.  

Results: An higher prevalence of potential risk factors was observed in 18 out of 23 risk factors in autistic group; for five of them the difference in prevalence was statistically significant (p<0.05) despite the relative small sample size: exposure to solvent or paints during pregnancy  (25% autism  vs. 3.8% Typicals), pregnancy complications( 50% autism vs 32% Typicals), perinatal complications ( 36.4 % autism vs 20.75% Typicals), stressful life events (mean number per woman: 0.49  autism vs. 0.06  Typicals), early antibiotic therapy after birth(25.04% autism vs 13.21% Typicals). Auto-CM system, at variance with PCA, was able to point out complex relationships among variable under study showing a convergence of branches of risk factors toward autistic outcome.

Conclusions: The general prevalence of  potential risk factors in pregnancy history and peri and post natal events is higher in autistic group in comparison with Typicals. According to univariate analysis exposure to solvent or paints during pregnancy, pregnancy complications, perinatal complications  stressful life events and early antibiotic treatment appear as key players. Artificial neural networks help to highlight  the underlying  interaction scheme among different factors on study showing a complex matrix of connections among them.