Background: Several factors like gender of newborn, families' education and economic status together with cultural perceptions of autism can influence the parents' recognition of the first signs of autism in their children and influence their decision to seek medical help for ASD suspicion. These factors suffer from non-linearity, complexity, fuzzy interaction which need a machine learning approach. ??

Objectives: The aim of this pilot study is to disentangle the association between age at first medical contact for autism diagnostic suspicion and parental characteristics in Lombardy Region, the most populous and rich Region in Italy, with an innovative machine learning approach.

??Methods: Fifty-eight ASD cases referred to our Institute for Rehabilitation from 2022 to 2025 coming from Lombardy municipalities were included in the study. The study group was composed of 41 males and 17 females diagnosed with autism according to DSM V criteria. The age at first medical contact for parental concern, determined from parents' reports, child gender, mother, and father education achievement level, their occupation level, family Socio-Economic Status (SES) and their place of birth (North, Centre, South Italy, Foreign country) were used for input data. The parameters were processed using a new data mining method based on a particular artificial adaptive system, Activation and Competition System (ACS), developed at Semeion Research Centre (Rome, Italy). ACS is an auto-associative neural network, able to integrate the weight matrices other algorithms. ACS once trained, works as a dynamic nonlinear associative memory. Whenever any input is set on (in our case early age at first visit), ACS will activate all its units in a dynamic, competitive, and cooperative process at the same time. This process ends once the evolutionary negotiation among all the units will find its natural attractor. The data set was trained using the four types of algorithms: Pearson linear correlation algorithm, prior probability algorithm, Self organizing maps and A-Temporal Diffusion Model algorithm, thus using simultaneously four different weight matrices.

??Results: The age at first medical contact for parental concern ranged between 6 months and 8.5 years (mean 40 months). ACS activation with the query “very early age at first visit” resulted in a dynamical spread of variables values (figure 1). Being female and living in southern Italy were the first responding variables, negatively associated to the query, while Living at North Italy, then male, higher family SES were the first responding variables positively associated to the query. Family SES variables were later positively responsive to the query. ??

Conclusions: The age at first medical contact for parental concern ranged between 6 months and 8.5 years (mean 40 months). ACS activation with the query “very early age at first visit” resulted in a dynamical spread of variables values (figure 1). Being female and living in southern Italy were the first responding variables, negatively associated to the query, while Living at North Italy, then male, higher family SES were the first responding variables positively associated to the query. Family SES variables were later positively responsive to the query.

Notes:

Fondazione VSM di Villa Santa Maria