Device Software Verification

    Verification and validation for medical device software

    Device makers lose review time when test evidence does not trace back to requirements and risk controls. Agnotic Technologies plans and runs verification and validation for medical device software and SaMD, and produces the records IEC 62304 and an FDA premarket submission expect.

    IEC 62304 by safety classRequirement to test traceabilityBasic or Enhanced evidenceUsability validation

    Trusted by global innovators

    Benchmark
    Chibasco
    Fundency
    Lantimer
    Lauren
    Lera
    One Minute
    Pento Pix
    TAP
    Xtrium
    Healthevolve
    Benchmark
    Chibasco
    Fundency
    Lantimer
    Lauren
    Lera
    One Minute
    Pento Pix
    TAP
    Xtrium
    Healthevolve
    Benchmark
    Chibasco
    Fundency
    Lantimer
    Lauren
    Lera
    One Minute
    Pento Pix
    TAP
    Xtrium
    Healthevolve
    Benchmark
    Chibasco
    Fundency
    Lantimer
    Lauren
    Lera
    One Minute
    Pento Pix
    TAP
    Xtrium
    Healthevolve

    Verification and validation

    What is verification and validation for medical device software?

    Verification and validation for medical device software is the evidence that the software was built right and that it is the right software. Verification checks each design output against its requirements; validation checks the finished software against user needs and intended uses. Agnotic Technologies plans both activities and produces the test records a device maker submits.

    This page is about the evidence. Choosing the FDA pathway and classifying the device sit on our FDA SaMD compliance page, and premarket cybersecurity work sits on our medtech engineering page.

    How much evidence you need depends on two classifications that are easy to mix up: the IEC 62304 software safety class (A, B or C) and the FDA documentation level (Basic or Enhanced). We settle both at the start, because together they decide which tests are written down and how.

    Decision table

    Which verification activities does each IEC 62304 software safety class require?

    Our working summary of IEC 62304:2006 with Amendment 1:2015. Your quality team confirms the class and the final plan against the standard text.

    Activity and recordClass AClass BClass C
    Development plan and software requirementsRequiredRequiredRequired
    Architecture design, with SOUP requirements and segregation for risk controlNot requiredRequiredRequired
    Detailed design of each software unit and its interfacesNot requiredUnits identified; detailed design not requiredRequired and verified
    Unit verification against acceptance criteriaNot required; units are implementedRequiredRequired, with added criteria such as fault handling, data flow and boundary conditions
    Integration and integration testingNot requiredRequiredRequired
    Software system testingRequiredRequiredRequired
    SOUP recordsTitle, manufacturer and versionPlus functional and performance requirements and a review of published anomaly listsSame as class B

    Class C needs the most records, but class A still needs requirements, system testing and a controlled release. Low risk never means no evidence.

    FDA documentation level

    Which test evidence goes into an FDA premarket submission?

    FDA's device software guidance sets two documentation levels. The level follows the worst credible harm a software failure could cause before risk controls are applied, not the device class alone.

    01

    Basic Documentation

    Example: Software where a failure could not lead to a hazardous situation with a probable risk of death or serious injury.

    Approach: A summary of unit, integration and system testing, the system level test protocol and report, the tested version history and the list of unresolved anomalies with their impact.

    02

    Enhanced Documentation

    Example: Software where a failure could present a hazardous situation with a probable risk of death or serious injury.

    Approach: Everything at the Basic level, plus unit and integration test protocols and reports, and a software design specification that traces to the requirements.

    Traceability

    What does one row of a traceability matrix contain?

    Sample rows for an illustrative dose calculator feature. Every requirement links to the hazard it controls, the test that verifies it and the recorded result.

    RequirementHazard and risk controlVerificationEvidence
    SRS-112: reject a dose entry above the configured maximumHAZ-07 overdose from a typing error; control RC-07 input limit checkUnit tests at, below and above the limit; system test ST-31Pass in build 2.4.0, linked to the CI run and the signed test report
    SRS-118: show the unit of measure beside every dose valueHAZ-09 unit confusion; control RC-09 labelled displayUsability validation task with intended users; system test ST-36Task results and observed use errors in the usability report
    SRS-140: keep the last confirmed dose if the sync service is downHAZ-12 stale data shown as current; control RC-12 staleness bannerIntegration test with the sync service stoppedPass, with the related anomaly closed in build 2.4.1

    We generate this matrix from the links stored on each requirement and test case, so it is never retyped by hand before a submission.

    Failure points

    Where does verification and validation evidence fall apart?

    Problems we see when device teams prepare a submission or an audit, and the control we put in place.

    Challenge

    Tests pass, but nobody can show which requirement or risk control each test covers.

    Agnotic approach

    Every test case carries the IDs of its requirement and risk control, and the trace matrix is generated from those links.

    Challenge

    The software safety class was assumed, not argued, and changes late.

    Agnotic approach

    We write the class rationale from the hazard analysis in the first sprint and revisit it when the architecture changes.

    Challenge

    Third-party libraries ship in the build but are missing from the SOUP list.

    Agnotic approach

    The SOUP list is produced from the dependency manifest in CI, so a new library cannot ship without a record.

    Challenge

    Automated tests run in CI, but the evidence lives only in pipeline logs that expire.

    Agnotic approach

    Results are exported per build, reviewed, signed off and stored with the version they tested.

    Challenge

    A small fix ships without regression evidence.

    Agnotic approach

    Change impact analysis names the affected requirements, and their regression tests run before release.

    Challenge

    Validation is a demo for the product team, not a study with users.

    Agnotic approach

    Usability validation runs with representative intended users in realistic use environments, following IEC 62366-1 and FDA human factors guidance.

    What do we deliver in a device software V&V engagement?

    Verification and validation services that fit your quality system instead of running beside it.

    Verification and validation plan

    Scope, test levels, environments, entry and exit criteria and sign-off roles, mapped to your QMS procedures.

    Trace matrix

    Requirement to risk control to test case to result, generated from your tools and kept current on every build.

    Unit and integration test suites

    Automated tests written to the acceptance criteria your safety class needs, run on every merge.

    System test protocols and reports

    Protocols with expected results, executed with observed results and pass or fail, ready for the submission.

    SOUP and off-the-shelf software records

    Version, purpose, known anomaly review and the requirements each item must meet.

    Usability validation support

    Use-related risk inputs, task scripts and summative study support with intended users.

    Regression evidence per release

    Change impact analysis, regression runs, and an updated version history and anomaly list for each release.

    CI evidence pipeline

    Build-tagged test results, coverage and static analysis output exported as controlled records.

    Who it is for

    Who do we run device software verification and validation for?

    Device and digital health teams whose software needs evidence a reviewer can follow.

    SaMD startups

    Teams preparing a first premarket submission who need the evidence built alongside the product.

    Device makers with embedded software

    Firmware and companion app teams whose software is part of a physical device.

    Teams changing released device software

    Changes that need change impact analysis and regression evidence before release.

    Quality and regulatory teams before an audit

    Trace gaps and SOUP records closed before an inspection or a notified body audit.

    Digital health products adding regulated features

    Products crossing from wellness into device functions, which need a V&V process for the first time.

    Standards

    Which standards shape device software verification and validation?

    The standards and rules behind the tables above.

    IEC 62304ISO 14971IEC 62366-121 CFR Part 820ISO 13485FDA software guidance

    Compliance-First Healthcare App Development Services Backed by Global Standards

    01HIPAA logo

    HIPAA

    Health Insurance Portability and Accountability Act

    Protect PHI with privacy-first architecture, encrypted storage and transmission, strict access controls, and traceable audit logs.

    02GDPR logo

    GDPR

    General Data Protection Regulation

    Implement lawful consent flows, data minimization, retention controls, and secure processing for sensitive health data.

    03FHIR logo

    FHIR

    Fast Healthcare Interoperability Resources

    Enable standardized health data exchange across apps, care teams, and systems through standards-based FHIR APIs.

    04HL7 logo

    HL7

    Health Level Seven International

    Support enterprise-grade interoperability with HL7-based integrations for records, events, and clinical messaging workflows.

    05HITRUST logo

    HITRUST

    Health Information Trust Alliance

    Align security programs to healthcare-specific control and risk management practices trusted by providers and ecosystem partners.

    06HITECH logo

    HITECH

    Health Information Technology for Economic and Clinical Health Act

    Design with breach notification readiness, digital record safeguards, and operational controls that support regulated care programs.

    07SaMD logo

    SaMD

    FDA Software as a Medical Device

    Plan software quality, traceability, and documentation pathways for products that may require SaMD review and submission.

    08MDR (EU) logo

    MDR (EU)

    Medical Device Regulation (European Union)

    Prepare EU market-ready processes for risk classification, evidence tracking, and lifecycle governance under MDR expectations.

    09SAMHSA logo

    SAMHSA

    Substance Abuse and Mental Health Services Administration

    Apply confidentiality controls and consent-aware sharing models for behavioral and mental health data experiences.

    Facts and sources

    What do FDA and IEC sources say about device software V&V?

    Rules and guidance behind this page, each linked to its primary source.

    • FDA recommends Enhanced Documentation in a premarket submission when a failure or flaw of any device software function could present a hazardous situation with a probable risk of death or serious injury, judged before risk control measures are applied.

      Source: FDA, Content of Premarket Submissions for Device Software Functions (June 2023)
    • At the Basic Documentation level FDA asks for a summary of unit, integration and system testing plus the system level test protocol and report; the Enhanced level adds unit and integration test protocols and reports with expected results, observed results and pass or fail determinations.

      Source: FDA, Content of Premarket Submissions for Device Software Functions, Table 1
    • FDA describes software verification as objective evidence that the design outputs of a life cycle phase meet the requirements specified for that phase, and software validation as confirmation that software specifications conform to user needs and intended uses.

      Source: FDA, General Principles of Software Validation
    • Under the Quality Management System Regulation, manufacturers of class II and class III devices, and of class I devices automated with computer software, must comply with the design and development requirements in clause 7.3 of ISO 13485.

      Source: 21 CFR 820.10(c) (eCFR)
    • IEC 62304 sets life cycle requirements for medical device software, whether the software is itself a medical device or part of one, and does not cover validation and final release of the medical device.

      Source: IEC 62304:2006 (IEC webstore)
    Engagement

    How does a verification and validation engagement run?

    Three stages, each with a clear output for your quality team.

    1.

    Gap review

    We review your requirements, hazard analysis, existing tests and records, and agree the safety class and documentation level.

    Scope
    2.

    Plan and build

    We write the V&V plan, add trace links, build the automated suites and draft the protocols.

    Build
    3.

    Execute and report

    We run the protocols on the release candidate and hand over reports, the trace matrix and the anomaly list.

    Evidence

    Scope and cost

    What drives the cost of verification and validation?

    We do not publish prices. These are the things that change the size of the work, priced after the scoping call.

    Safety class and documentation level

    Class C software at the Enhanced level needs unit and integration protocols as well as system tests.

    State of your records

    Whether requirements, hazard analysis and tests already exist and link to each other, or need to be built.

    Hardware and interfaces

    Test rigs, device simulators and the connected systems the software talks to.

    Usability validation scope

    How many user groups and use environments need a summative study.

    Release cadence

    How often you release, which sets how much regression evidence is produced.

    Your tools

    The requirements, test management and ALM tools your QMS already uses.

    Frequently Asked Questions

    Verification asks whether the software meets its specified requirements at each stage, through reviews, analysis and tests. Validation asks whether the finished software meets user needs and intended uses, which is why it involves intended users and the real use environment.