Inferensys

Glossary

Design History File (DHF)

A Design History File (DHF) is a compilation of records that describes the design history of a finished medical device, demonstrating that it was developed in accordance with the approved design plan.
Modern WeWork hardware lab area with product team collaborating around AI device prototypes, 3D printer in background, dramatic industrial lighting with product sketches on glass walls.
REGULATORY DOCUMENTATION

What is a Design History File (DHF)?

A Design History File (DHF) is the formal compilation of records that proves a medical device was developed in accordance with an approved design plan and regulatory requirements.

A Design History File (DHF) is a compilation of records that describes the complete design history of a finished medical device, demonstrating that development followed the approved design plan and 21 CFR Part 820.30 design controls. It serves as the definitive evidentiary record that design outputs meet design inputs, encompassing everything from initial user needs to final design transfer.

The DHF contains or references all design review minutes, verification and validation protocols, risk management files per ISO 14971, and approved design changes. For Software as a Medical Device (SaMD), the DHF integrates software development lifecycle documentation aligned with IEC 62304, providing auditors with a traceable narrative of how safety and performance requirements were methodically translated into the released product.

DESIGN HISTORY FILE

Core Components of a DHF

A Design History File (DHF) is not a single document but a compilation of records proving a medical device was developed according to its approved design plan and regulatory standards. These are the essential artifacts auditors expect to find.

01

Design and Development Plan

The master schedule and organizational blueprint that defines who does what and when. This document outlines all design activities, assigns responsibilities, identifies interfaces between groups, and is continuously updated throughout development. It must align with ISO 13485 and 21 CFR Part 820.30(b).

02

Design Inputs

The physical and performance requirements that form the basis of the device. These must be unambiguous, verifiable, and complete. For SaMD, inputs include:

  • Intended Use Statement and clinical claims
  • Functional, performance, and safety requirements
  • Cybersecurity and interoperability specifications
  • Applicable regulatory standards (e.g., IEC 62304)
03

Design Outputs

The tangible results of the design effort that deliver against the design inputs. Outputs include software architecture diagrams, source code, compiled binaries, user manuals, and the Software Bill of Materials (SBOM). Each output must reference the specific design input it satisfies, creating a traceable, auditable link.

04

Design Review Records

Formal, documented assessments conducted at major milestones to evaluate the design's maturity. Records must identify the reviewers, date, issues found, and resolution actions. Reviews confirm that the design can proceed to the next phase and that Verification and Validation (V&V) plans are adequate.

05

Design Verification Records

Objective evidence confirming that design outputs meet design inputs (i.e., 'did we build the thing right?'). For software, this includes unit test results, integration test logs, static code analysis reports, and traceability matrices. Verification is a continuous, iterative process, not a final gate.

06

Design Validation Records

Objective evidence confirming that the final device meets user needs in the intended clinical environment (i.e., 'did we build the right thing?'). For diagnostic SaMD, this includes clinical validation study data, Human Factors Engineering (HFE) usability test reports, and evidence of diagnostic accuracy (sensitivity/specificity).

07

Design Transfer Records

Documentation proving the design was correctly translated into a production specification for manufacturing or, for SaMD, a reproducible release process. This includes build scripts, deployment configurations, and installation qualification records ensuring the software can be consistently distributed to end-users without corruption.

08

Design Change Records

A chronological log of all post-approval modifications, maintained under Corrective and Preventive Action (CAPA) procedures. Each record must describe the change rationale, risk assessment, verification/validation activities performed, and regulatory impact analysis—especially critical for determining if an Adaptive Algorithm update requires a new 510(k).

REGULATORY COMPLIANCE

Frequently Asked Questions

Essential questions about the Design History File (DHF), its contents, and its role in FDA clearance pathways for Software as a Medical Device (SaMD).

A Design History File (DHF) is a compilation of records that describes the complete design history of a finished medical device, demonstrating that it was developed in accordance with the approved design plan and 21 CFR Part 820.30. The DHF works as a chronological narrative of the design and development process, containing or referencing all records necessary to prove that the design was conducted according to established procedures. For Software as a Medical Device (SaMD), the DHF captures the evolution of software architecture, algorithm selection, and iterative testing cycles. It serves as the primary evidence during an FDA inspection or audit that the manufacturer followed a controlled, documented design process. The file is not a single document but a structured index linking to design inputs, design outputs, verification and validation protocols, design reviews, and risk management files. Critically, the DHF must be maintained for the lifetime of the device and be readily accessible for regulatory review.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.