Diagnosis of ADHD in Children with EEG and Machine Learning: Systematic Review and Meta-analysis

Background/Objective:ADHD impacts about 5.9% of infant populations worldwide. Its diagnosis is traditionally based on clinical assessment and psychometric test results. Actually, machine learning (ML) has begun to be used to identify various neurodevelopmental conditions, favoring diagnostic accuracy. This research aimed to develop a meta-analysis and systematic literature review (SLR) on the application of ML techniques for the identification of ADHD in children using electroencephalograms (EEG).Method: Using the PRISMA methodology, 30 studies were selected from Scopus and Web of Science (WoS) that met the selection criteria.Results:The most commonly used ML technique was support vector machines. The models evaluated achieved accuracy rates of up to 94.92%, indicating their ability to diagnose ADHD.Conclusions: Predictor variables such as frequency power, neuronal connectivity, and entropy are useful for identifying characteristic patterns in EEG signals, that reflect neuronal dysfunction associated with ADHD.