跳到正文
原文
Nature 健康护理主题·· 5 天前AI 评分44

GRACE-AI:面向高风险 AI 系统的治理导向型 MLOps 框架,实现风险分级控制与持续保证

GRACE-AI: a governance-centric MLOps framework for risk-tiered control, continuous assurance, and explainability in high-stakes AI systems

AI 导读

研究提出 GRACE-AI 框架,面向医疗与金融领域的治理导向 MLOps 架构,包含监管对齐、动态风险分级、持续监测和可解释性四大模块。该框架在三个基准数据集的四个预测任务中取得平均准确率96.38%、AUC达0.973,较Transformer基线提升2.78个百分点;可覆盖94.7%风险、减少38.6%合规负担,漂移检测平均延迟仅1.8秒。

正文

Scientific Reports (2026) Cite this article

We’re sharing this article early to provide faster access to peer-reviewed, accepted research. It is citable and carries a permanent DOI. This version is subject to further edits and will be replaced automatically by the final Version of Record. All legal disclaimers apply.

Abstract

The recent successful adoption of artificial intelligence (AI) in several important applications has made it imperative to integrate operational machine learning with governance frameworks to facilitate regulatory alignment, structured accountability mechanisms, and risk-based decision-making. Current MLOps frameworks are good at controlling the technical lifecycle of AI models, but are less likely to support governance needs in regulated environments. This paper introduces a governance-oriented MLOps framework to support healthcare and financial AI systems, namely GRACE-AI (Governance, Risk, Assurance, Compliance, and Explainability for Artificial Intelligence). The proposed framework is composed of four interconnected layers, (i) Regulatory Alignment Layer: mapping regulatory requirements into pipeline constraints that incorporate information about governance; the Regulatory Alignment Layer is presented as a proposed architectural component for automated obligation extraction; its clause-level classification performance was not independently benchmarked in this study (ii) Dynamic Risk-Tiering Engine: assigning adaptive governance risk levels based on composite risk scores; (iii) Continuous Assurance Pipeline: drift detection, monitoring, automated recalibration, and anomaly tracking; and (iv) Explainability-by-Design Module: generation of stakeholder-oriented and interpretable explanations. The proposed framework was evaluated using three benchmark datasets supporting four prediction tasks across the healthcare and financial domains, achieving an average accuracy of 96.38% and an unweighted cross-task mean AUC of 0.973; this is 2.78 percentage points above the strongest evaluated standalone baseline (Transformer, 93.60%) and 6.06 percentage points above the mean of the five evaluated standalone baselines. In addition, GRACE-AI could cover 94.7% of the risk, reduce compliance overhead by 38.6%, and detect drift with an average latency of 1.8 s. The findings suggest that the proposed governance model is effective in a controlled experimental setting and could be used to enable trustworthy, governance-aware deployment of AI in regulated environments. In the future, the framework will be validated under operational, real-world conditions with human involvement and at production scale.

Materials availability

Materials used in this research are available with corresponding author and given on request.

Funding

No financial support was received by the authors in this research.

Author information

Authors and Affiliations

  1. Hyderabad, 500038, India

    Rakesh Reddy Kamireddy, Pavan Kumar Reddy Bijjam & A Venkata Sudhakara Reddy

  2. London, Coventry, CV1 4AW, UK

    Sunilkumar Reddy Eraganeni

Authors

  1. Rakesh Reddy Kamireddy
  2. Pavan Kumar Reddy Bijjam
  3. Sunilkumar Reddy Eraganeni
  4. A Venkata Sudhakara Reddy

Corresponding author

Correspondence to Rakesh Reddy Kamireddy.

Ethics declarations

Competing interests

The authors whose names are listed immediately below certify that they have NO affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants; participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript.

Ethical approval

This research does not involve humans or animals, so no ethical approval is required.

Consent for publication

The authors give consent for their publication.

Clinical trial number

Not applicable.

Additional information

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kamireddy, R.R., Bijjam, P.K.R., Eraganeni, S.R. et al. GRACE-AI: a governance-centric MLOps framework for risk-tiered control, continuous assurance, and explainability in high-stakes AI systems. Sci Rep (2026). https://doi.org/10.1038/s41598-026-74834-4

Download citation

  • Received: 20 May 2026

  • Accepted: 01 October 2026

  • Published: 15 November 2026

  • DOI: https://doi.org/10.1038/s41598-026-74834-4

Keywords

来源:Nature 健康护理主题 · nature.com