EUVIMEDEUVIMED entdecken
Systematische Übersicht

Explainable Artificial Intelligence for Phishing Detection and Its Implications for Health Information System Security: A Systematic Literature Review

Habibah Alya

Al Makki Health Informatics Journal · 25.05.2026 · 4 (3)

DOI: 10.57185/a47tqt77 ↗

Kurz gefasst

Systematische Übersicht in der Zeitschrift Al Makki Health Informatics Journal, online erschienen am 25.05.2026. Verfasst von Habibah Alya.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Phishing attacks continue to grow in scale and sophistication, and health care is a vulnerable target because it holds high-value patient data and depends on available information systems. Deep learning and Transformer-based detectors are accurate but often opaque. This review aimed to synthesize the development of Explainable Artificial Intelligence (XAI) for phishing detection and interpret its implications for health information system security. Following a PRISMA 2020-guided protocol applied to Scopus, 365 records were retrieved, 71 full texts were assessed, and 48 peer-reviewed studies published from 2022 to 2026 were included by a single reviewer, alongside supporting cybersecurity XAI literature. Detection models evolved from traditional machine learning toward ensemble, Transformer, and large language model architectures, with ensemble learning the most frequently reported baseline, while SHAP and LIME dominated explainability. Explainability improved transparency, whereas accuracy gains came mainly from explainability-driven feature selection in a few studies rather than from XAI in general. However, no study integrated explainable feature selection, machine learning, ensemble learning, and deep learning with trustworthy cybersecurity principles in one framework, and none used health care data or settings. Real-time deployment, standardized evaluation of explanation quality, and human-centered validation remain open challenges. A conceptual framework and a research agenda with four directions (explainable AutoML, explainable foundation models, explainable federated learning, and human and AI collaborative security) are proposed toward explainable, adaptive, and privacy-preserving phishing detection. The health care implications are interpretive and require empirical validation in health facilities.

Zitieren

APA

Alya, H. (2026). Explainable Artificial Intelligence for Phishing Detection and Its Implications for Health Information System Security: A Systematic Literature Review. Al Makki Health Informatics Journal, 4(3). https://doi.org/10.57185/a47tqt77

Vancouver

Alya H. Explainable Artificial Intelligence for Phishing Detection and Its Implications for Health Information System Security: A Systematic Literature Review. Al Makki Health Informatics Journal. 2026;4(3). doi:10.57185/a47tqt77

BibTeX
@article{Alya2026,
  author = {Alya, Habibah},
  title = {Explainable Artificial Intelligence for Phishing Detection and Its Implications for Health Information System Security: A Systematic Literature Review},
  journal = {Al Makki Health Informatics Journal},
  year = {2026},
  volume = {4},
  number = {3},
  doi = {10.57185/a47tqt77}
}
Kostenlos starten →