Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

From Safety Documentation to Safety Knowledge Support: An Evidence-Grounded LLM Framework for Medical Devices

Source

arXiv

arxiv.org/abs/2608.12025v1

Read the source

Source

  • Source name: arXiv
  • Source kind: paper
  • Authors / organization: Tuhinangshu Gangopadhyay, Rasmus Adler, Peter Liggesmeyer, Jan Reich
  • Venue: arXiv: cs.SE
  • Published: 2026-08-12
  • Collected: 2026-08-13
  • URL: http://arxiv.org/abs/2608.12025v1
  • PDF: https://arxiv.org/pdf/2608.12025v1
  • DOI: UNKNOWN
  • Registry group: academic_apis_and_preprints
  • Evidence role: canonical_academic_index
  • Language: en
  • Country / region: UNKNOWN

Plain-Language Brief

The paper proposes a framework that uses large language models (LLMs) to support medical device safety work, which is document-based and costly due to the need for safety and domain experts. The framework connects device artifacts, provides controlled knowledge storage and retrieval, generates candidate safety items, performs critique and uncertainty checks, and records expert review. It does not decide on the safety of devices or provide regulatory approval.

What the Document Says

  • Medical devices are becoming more software-intensive, connected, and AI-enabled.
  • The development of medical devices requires risk-management evidence as per ISO 14971 and IEC 62304 for software.
  • The evidence must be kept consistent across various aspects of the device's life cycle.
  • Large language models (LLMs) may reduce parts of this effort due to the document-based nature of medical-device safety work.
  • Current LLM-based safety-engineering studies have limitations such as addressing isolated methods, relying on generic prompting or public examples, and providing limited support for source links, traceability, uncertainty handling, lifecycle updates, and recorded expert review.
  • The paper argues that the central research problem is not safety-text generation, but source-linked safety-knowledge support.
  • A proposed evidence-grounded framework connects device artifacts, controlled knowledge storage and retrieval, method-specific generation of candidate safety items, critique and uncertainty checks, and recorded expert review.
  • The framework prepares, links, checks, and updates candidate safety artifacts for expert decision-making.
  • It does not decide whether a device is safe or provide regulatory approval.
  • An evaluation strategy is outlined using non-public or newly built medical-device case studies and expert reference analyses to assess the framework's performance.

Key Concepts Mentioned

  • medical devices
  • software-intensive
  • connected
  • AI-enabled
  • risk management
  • ISO 14971
  • IEC 62304
  • large language models (LLMs)
  • evidence-grounded framework
  • device artifacts
  • controlled knowledge storage and retrieval
  • candidate safety items
  • critique and uncertainty checks
  • expert review
  • non-public case studies
  • expert reference analyses

Claims Made by the Document

  • LLMs can reduce parts of the effort required for medical-device safety work.
  • The central research problem is not safety-text generation, but source-linked safety-knowledge support.

Evidence Used

  • The document states that medical-device safety work is highly document-based and costly due to the need for experts.

Methods / Approach

  • Proposing an evidence-grounded framework
  • Outlining an evaluation strategy

Limitations Stated by the Authors

  • Current LLM-based safety-engineering studies have limitations such as those listed.
  • The proposed framework does not decide on the safety of devices or provide regulatory approval.

Limitations Noted by the Agent

Agent interpretation — not from the source.

The framework, as proposed, may still have limitations in terms of its ability to fully replace human experts and ensure device safety.

Open Questions

  • The actual performance of the proposed framework is yet to be evaluated.
  • How effective will the framework be in a real-world setting?

Notes for Future Q&A

May include interpretation. Labeled where it does.

The paper presents a theoretical framework but does not provide empirical evidence of its effectiveness.

Raw Abstract / Excerpt

Medical devices are becoming more software-intensive, connected, and AI-enabled. Their development requires risk-management evidence aligned with ISO 14971 and, for software, IEC 62304. This evidence must be kept consistent across requirements, design decisions, software changes, verification results, complaints, and post-market data. These tasks are costly and depend on scarce safety and domain experts. Large language models (LLMs) may reduce parts of this effort because medical-device safety work is highly document-based. However, current LLM-based safety-engineering studies often address isolated methods, rely on generic prompting or public examples, and provide limited support for source links, traceability, uncertainty handling, lifecycle updates, and recorded expert review. This limits their use in regulated medical-device development. This paper argues that the central research problem is not safety-text generation, but source-linked safety-knowledge support. We propose an evidence-grounded framework that connects device artifacts, controlled knowledge storage and retrieval, method-specific generation of candidate safety items, critique and uncertainty checks, and recorded expert review. The framework prepares, links, checks, and updates candidate safety artifacts for expert decision-making. It does not decide whether a device is safe and does not provide regulatory approval. We also outline an evaluation strategy using non-public or newly built medical-device case studies and expert reference analyses to assess coverage, correctness, relevance, traceability, duplicate rate, unsupported claims, and review effort.