Examining AI-Powered ADHD Diagnosis: Current Trends, Key Challenges, and Future Directions in the Field
Attention deficit hyperactivity disorder (ADHD) is a type of neurodevelopmental disease affecting the mental health of children and adults. Individuals with ADHD show various symptoms such as inattention, hyperactivity, and impulsivity. Early diagnosis of ADHD helps to alter neural connections and r...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10988850/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Attention deficit hyperactivity disorder (ADHD) is a type of neurodevelopmental disease affecting the mental health of children and adults. Individuals with ADHD show various symptoms such as inattention, hyperactivity, and impulsivity. Early diagnosis of ADHD helps to alter neural connections and refine symptoms. The clinical practice to diagnose ADHD is through subjective measures and does not significantly capture the underlying structural and functional mechanisms of the brain. Therefore, it is crucial to explore other approaches such as Artificial Intelligence (AI) to improve the accuracy and efficacy of ADHD diagnosis. Consequently, in this article we systematically investigate various Machine Learning (ML) and Deep Learning (DL) approaches as well as different diagnostic tools or modalities employed for the identification of ADHD. Particularly, a Systematic Literature Review (SLR) is conducted to review and analyze 98 selected studies published from 2021 to 2024. Subsequently, the selected studies are grouped into five categories based on the modalities utilized in these studies: physiological signals (37), magnetic resonance imaging (31), questionnaires (11), motion data (8), and others (11). We also analyze AI models which indicates that 45 studies utilized ML models, 33 studies employed DL models, and 20 studies used both. However, there are still some gaps in current research such as a lack of publicly available datasets except MRI and EEG. Although datasets for MEG and actigraphy exist, but they are underexplored and have been utilized in only a few studies. While DL models like CNNs and ANNs have been increasingly applied in recent years for ADHD diagnosis, there is a shortage of advanced DL models, including transfer learning approaches like ResNet and VGG. Additionally, there is a lack of interpretability in AI models, particularly DL models. Furthermore, most studies focus on individual modalities for ADHD diagnosis, and despite many studies showing excellent results, there is a lack of implementation of AI-based tools in real-world clinical settings. These gaps highlight areas for further exploration and development. |
|---|---|
| ISSN: | 2169-3536 |