Early identification of Attention Deficit Hyperactivity Disorder (ADHD) is often limited by subjective methods and complexities related to symptom variability and comorbidities. This study presents DETEC-ADHD, a web-based application based on machine learning (ML) and electroencephalography (EEG) to improve the diagnostic accuracy of ADHD and identify its subtypes (inattentive, hyperactive-impulsive, and combined).
The system integrates personal, medical and psychological data with EEG signals, using the theta/beta wave ratio as a biomarker. Among the algorithms tested, Logistic Regression demonstrated high efficacy, achieving 100% accuracy in children and 90% in adults. In addition, DETEC-ADHD uses portable devices for real-time data collection, such as the Muse Band S (Gen 2), optimizing accessibility and scalability in resource-limited contexts.
Validation tests included cognitive and relaxing activities, focusing on beta and theta brainwave patterns, associated with attention processing and relaxation states, respectively. The model was developed and validated with the HYPERAKTIV dataset, which included EEG recordings, physiological parameters and behavioral data, and demonstrated robustness in ADHD classification.
The system’s modular architecture is organized into presentation, integration, services and storage layers, enabling integration with smart devices and continuous monitoring. This design enables real-time analysis, automatic classification and provision of accurate diagnoses aligned with clinical needs.
The results indicate that DETEC-ADHD represents an innovative and effective approach for early diagnosis and identification of ADHD subtypes, offering an accessible and evidence-based solution for clinical and laboratory settings.
Reference :
SANTARROSA-LÓPEZ, Ismael; ALOR-HERNÁNDEZ, Giner; BUSTOS-LÓPEZ, Maritza; HERNÁNDEZ-CAPISTRÁN, Jonathan; SÁNCHEZ-MORALES, Laura Nely; SÁNCHEZ-CERVANTES, José Luis; MARÍN-VEGA, Humberto. DETEC-ADHD: A Data-Driven Web App for Early ADHD Detection Using Machine Learning and Electroencephalography. Big Data and Cognitive Computing, v. 9, n. 3, 2025. Available at: https://doi.org/10.3390/bdcc9010003. Accessed on: Jan. 27, 2025.
