EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Clinical artificial intelligence as a cognitive intervention: cognitive algovigilance, adverse events, and post-market surveillance

Daniele Angioni, Fabio Di Bello

AboutOpen · 29.07.2026 · 13 (1) · 47-53

DOI: 10.33393/ao.2026.3940 ↗

Kurz gefasst

Fachartikel in der Zeitschrift AboutOpen, online erschienen am 29.07.2026. Verfasst von Daniele Angioni und 1 weiteren Autor:innen. Die Arbeit nennt 30 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

robustness, bias, drift, transparency, and regulatory compliance. These dimensions remain essential, but they do not capture the full safety problem created when AI enters clinical reasoning. The literature has already described automation bias, overtrust, alert fatigue, trust miscalibration, deskilling, and failure modes in human–AI collaboration. The unresolved problem is that these effects are often treated as scattered human-factors concerns rather than as observable safety events. This Point of View proposes clinical AI as a cognitive intervention. Its primary safety unit is the human-machine cognitive system, not the isolated model. We define cognitive algovigilance as the systematic surveillance of cognitive effects arising from human–AI interaction, and Cognitive Adverse Events as measurable AI-mediated changes in clinical decision-making that increase the probability of suboptimal decisions, including when the algorithmic output is technically correct. We distinguish technical/software surveillance as an enabling layer from cognitive algovigilance proper and define cognitive red flags as pre-event signals, including those elicited by risk-triggered or randomly sampled pre-output reasoning probes. We identify AI-induced diagnostic narrowing as a sentinel phenotype and propose Cognitive Post-Market Surveillance as a complementary post-deployment layer for collecting cognitive near misses, comparing pre-AI and post-AI differentials, auditing overrides and high-risk non-overrides, and monitoring verification decay, cognitive dependence, and loss of the diagnostic tail. The aim is not to slow AI adoption, but to distinguish cognitive augmentation from cognitive toxicity and make clinical AI safety a verifiable cognitive discipline.

Zitieren

APA

Angioni, D., & Bello, F. D. (2026). Clinical artificial intelligence as a cognitive intervention: cognitive algovigilance, adverse events, and post-market surveillance. AboutOpen, 13(1), 47-53. https://doi.org/10.33393/ao.2026.3940

Vancouver

Angioni D, Bello FD. Clinical artificial intelligence as a cognitive intervention: cognitive algovigilance, adverse events, and post-market surveillance. AboutOpen. 2026;13(1):47-53. doi:10.33393/ao.2026.3940

BibTeX
@article{Angioni2026,
  author = {Angioni, Daniele and Bello, Fabio Di},
  title = {Clinical artificial intelligence as a cognitive intervention: cognitive algovigilance, adverse events, and post-market surveillance},
  journal = {AboutOpen},
  year = {2026},
  volume = {13},
  number = {1},
  pages = {47--53},
  doi = {10.33393/ao.2026.3940}
}
Kostenlos starten →