Mon, Sep 14, 2026
Volume 21, Issue 2 (Summer & Autumn- In Press 2024)
ASJ 2024, 21(2): 0-0 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Saadati M, Saadati M. Artificial Intelligence in Clinical Scalp EEG: A Systematic Evidence Map of PubMed-Indexed Studies, 2019–2026. ASJ 2024; 21 (2)
URL: http://anatomyjournal.ir/article-1-656-en.html
URL: http://anatomyjournal.ir/article-1-656-en.html
1- Department of Health Information Technology, School of Allied Medical Sciences, Mazandaran University of Medical Sciences, Sari, Mazandaran, Iran.
2- Department of English Language and Literature, Faculty of Humanities and Social Sciences, Golestan University, Gorgan, Iran.
2- Department of English Language and Literature, Faculty of Humanities and Social Sciences, Golestan University, Gorgan, Iran.
Abstract: (7 Views)
Artificial intelligence is being increasingly used for the analysis of clinical scalp electroencephalography data. However, current studies differ widely concerning disorders, data representations, architecture of models, validation, and reporting. In our work, we performed an evidence mapping of the available literature on the shift from domain-specific deep learning to generalizable representations and clinically portable systems.
An evidence map was carried out on English-language papers that were indexed in PubMed. The evidence map covered studies published between 1 January 2019 and 29 August 2026. Only papers that could be freely accessed either, through PubMed Central or other open-access PDF sources were considered. Every chosen paper had to use artificial intelligence or representation learning techniques applied to clinically relevant human scalp electroencephalography. The records were checked quickly by one reviewer. Each identifier was verified.
We first found 3,527 records in PubMed. Of those we screened 3,506 records. After looking at titles and abstracts we sought 2,059 reports for retrieval. Out of them 996 could not be found publicly 1,063 were checked for eligibility. 66 Were excluded. Finally we included 997 studies that we could get publicly. The topics common were epilepsy or seizure, with 390 studies (39.1%) and sleep, with 185 studies (18.6%). Convolutional networks appeared in 486 studies (48.7%). Recurrent models were in 235 studies (23.6%). Transformer or attention models were in 176 studies (17.7%). We used subject‑independent or external or cross‑dataset validation in 274 studies (27.5%). Code availability was present in 135 studies (13.5%). Explainability was present, in 436 studies (43.7%).
In the publicly retrievable PubMed literature, clinical scalp electroencephalography research has been slower to adopt expressive architectures than to produce transferable evidence. Independent validation, leakage-free evaluation, clear reporting, resource sharing, and clinically interpretable spatiotemporal representation continue to be important. This evidence map does not provide an estimate of evidence from the 996 reports that could not be publicly retrieved, and it is not meant to cover all clinical.
An evidence map was carried out on English-language papers that were indexed in PubMed. The evidence map covered studies published between 1 January 2019 and 29 August 2026. Only papers that could be freely accessed either, through PubMed Central or other open-access PDF sources were considered. Every chosen paper had to use artificial intelligence or representation learning techniques applied to clinically relevant human scalp electroencephalography. The records were checked quickly by one reviewer. Each identifier was verified.
We first found 3,527 records in PubMed. Of those we screened 3,506 records. After looking at titles and abstracts we sought 2,059 reports for retrieval. Out of them 996 could not be found publicly 1,063 were checked for eligibility. 66 Were excluded. Finally we included 997 studies that we could get publicly. The topics common were epilepsy or seizure, with 390 studies (39.1%) and sleep, with 185 studies (18.6%). Convolutional networks appeared in 486 studies (48.7%). Recurrent models were in 235 studies (23.6%). Transformer or attention models were in 176 studies (17.7%). We used subject‑independent or external or cross‑dataset validation in 274 studies (27.5%). Code availability was present in 135 studies (13.5%). Explainability was present, in 436 studies (43.7%).
In the publicly retrievable PubMed literature, clinical scalp electroencephalography research has been slower to adopt expressive architectures than to produce transferable evidence. Independent validation, leakage-free evaluation, clear reporting, resource sharing, and clinically interpretable spatiotemporal representation continue to be important. This evidence map does not provide an estimate of evidence from the 996 reports that could not be publicly retrieved, and it is not meant to cover all clinical.
Keywords: Electroencephalography, Artificial Intelligence, Deep Learning, Machine Learning, Neurologic Diseases
Type of Study: Review |
Subject:
Neuroanatomy
Received: 2026/09/6 | Accepted: 2026/09/13 | Published: 2024/08/30
Received: 2026/09/6 | Accepted: 2026/09/13 | Published: 2024/08/30
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
Copyright © The Author(s);
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC-By-NC), which permits use, distribution, and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.
Contact Information
