Early diagnosis of autism spectrum disorders (ASD) is crucial for improving interventions and support (and, therefore, the quality of life) for individuals with this condition. Healthcare professionals use various observed manifestations and behaviours to support their diagnosis. This exercise proposes the development of a Bayesian network to aid in diagnosing autism in children, utilising various clinical observations and data.

Clinical observations indicate that 15% of patients experience social interaction problems, particularly difficulties relating to other children and adults. Of this group, 70% also exhibit limited nonverbal communication, characterised by minimal or no gestures and facial expressions. If they do not suffer from social interaction problems, the likelihood of having limited nonverbal communication drops to 20%. Furthermore, patients with social interaction problems also experience motor stereotypies (the constant repetition of movements or sounds) 65% of the time; this number drops to 20% if they do not have social interaction problems.

Finally, if a child presents limited nonverbal communication and motor stereotypies, the probability of being diagnosed with autism is estimated at 99%. If only one of these two factors is present, autism is present in 85% of cases. In the absence of both, only 5% of children have autism.