Infrastructure compliance enforcement is increasingly considered for agentic AI systems that can plan, act and self-correct over many steps. A structured examination of the literature reveals several significant gaps. Multi-agent compliance pipelines have been architecturally proposed in several works but none report a functioning prototype with measurable compliance outcomes. Theoretical discussions extensively cover the safety hazards arising from the granting of autonomous control to an LLM agent over remediation of infrastructure. However, there are no documented real-world cases of an LLM agent outputting an operationally dangerous output in a compliance setting. The concept of utilising cross-run memory for compliance agents has been recognised but no lightweight implementation has yet been demonstrated to change the behaviour of agents. This paper surveys the field along seven dimensions: agentic architectures, compliance automation, LLM output safety, anomaly detection, multi-agent coordination, statefulness, and cloud-native deployment drawing on 46 representative works. Six specific gaps are identified through structured analysis. A hybrid architecture is then proposed that integrates the Isolation Forest anomaly detection with a four-agent LLM pipeline consisting of an Analyser that interprets system state and prior run history, a Planner that generates remediation strategies, a Verifier that applies LLM safety constraints, and an Explainer that produces human-readable audit reports. The architecture further incorporates deterministic value-level validation and persistent SQLite-based cross-run memory. In controlled experiments, the system improved compliance scores from 60% to 100%. Notably, a concrete instance of LLM overreach was observed during testing: the Planner agent generated a remediation plan that would have locked out SSH access by closing all network ports, a failure mode not previously reported in empirical literature. The paper concludes with a feature-by-feature comparison across twelve prominent works, a discussion of open challenges, and a proposed hybrid cloud extension.
Saleha Soudagar, V. Rajpurohit, Arati Shahapurkar et al.· 2026 4th International Confe...· 0 citations
Ultrasound (US) remains one of the most widely used medical imaging modalities because it is real-time, radiation-free, portable, and comparatively inexpensive, but its diagnostic accuracy is strongly operator-dependent and subject to inter-observer variability. Over the past decade, artificial intelligence (AI) - and deep learning (DL) in particular - has been applied extensively to ultrasound image analysis in an effort to reduce this variability and support faster, more consistent diagnosis. This paper presents a comprehensive narrative review of current AI/DL applications in ultrasound diagnosis, synthesising evidence across four major clinical domains: breast lesion classification, thyroid nodule risk stratification, fetal cardiac screening for congenital heart disease, and abdominal/musculoskeletal applications. A structured review methodology is described, followed by a proposed conceptual framework that organises reviewed applications by deep learning task - classification, detection/localisation, and segmentation - and by clinical domain, situating an explainability layer and human-in-the-loop clinical review at the centre of responsible deployment. Reported diagnostic performance metrics (accuracy, sensitivity, specificity) from a representative set of primary studies and meta-analyses are synthesised and compared, showing that convolutional neural network (CNN)-based models frequently reach or approach expert-level diagnostic performance in constrained, retrospective evaluation settings, with reported accuracies generally in the 88-99% range and reported sensitivities in the 87-98% range across the domains reviewed. All performance figures reported in this paper are drawn directly from the cited primary studies and meta-analyses, not from a new experiment conducted by the authors. The review concludes by identifying recurring limitations across the literature - dataset heterogeneity, limited external/multicentre validation, and interpretability gaps - and outlines directions for future research toward clinically deployable, trustworthy AI-assisted ultrasound diagnosis.
R. Sivakumar, Arati Shahapurkar, J. G et al.· Adolescência e Saúde· 0 citations