CyberMed
SAMPLE
SAMPLE REPORT

FDA 524B Cybersecurity Readiness Report

Fictional company profile: MedFlow Health, cloud-connected insulin delivery system with embedded firmware, iOS/Android companion app, and BLE/Wi-Fi connectivity.

Readiness Tier

Moderate Risk

75 / 140

High Risk

Moderate

Lower Risk

Product Profile Summary

Cloud-connected insulin delivery system with embedded firmware, iOS/Android companion app, Bluetooth LE and Wi-Fi connectivity. Targeting 510(k).

This report is confidential and personalized to your product architecture and self-assessment responses.

Critical Gap: Cybersecurity Risk Assessment

Your self-assessment indicates you don't have a cybersecurity-specific risk assessment in place yet. For a cloud-connected insulin delivery system with BLE and Wi-Fi connectivity, this is the single most important gap to close before submission.

FDA's cybersecurity risk assessment is fundamentally different from your ISO 14971 safety risk analysis. Where ISO 14971 uses probability × severity, FDA requires exploitability-based scoring because cyber attackers are intentional actors, not random events. For your device specifically, this means scoring each threat based on how accessible the attack surface is (your BLE and Wi-Fi interfaces are remotely accessible, which elevates exploitability scores significantly compared to a device with only physical interfaces).

Your risk assessment needs pre-mitigation and post-mitigation scores for every threat identified in your threat model. Given your architecture (embedded firmware communicating via BLE to a mobile app, which syncs patient insulin delivery data to a cloud platform) you'll need to assess threats at each trust boundary: device-to-phone, phone-to-cloud, and cloud-to-clinician dashboard. Each of these boundaries has distinct threat profiles. The device-to-phone BLE link is susceptible to replay attacks and eavesdropping; the phone-to-cloud channel faces man-in-the-middle risks; the cloud platform has API authentication and data integrity concerns.

FDA reviewers typically start with this document. For insulin delivery devices, where a dosing error has direct patient safety implications, reviewers will scrutinize the connection between cyber risks and clinical hazards with particular care. We recommend using CVSS environmental scoring adjusted for your specific deployment context, and explicitly mapping each cyber risk to specific hazardous situations in your 14971 risk management file.

Critical Gap: Security Testing Evidence

Your self-assessment indicates some testing has been done but with significant gaps in coverage. For a 510(k) submission, the security testing eSTAR attachment often contains 10 or more individual test reports. A single penetration test is not sufficient.

Given your device architecture, here's what FDA will expect to see in your testing package. For the embedded firmware: static analysis (SAST) covering all custom code with CWE-based rules, and fuzz testing of the BLE GATT services. Every characteristic and descriptor should be fuzzed with malformed data to verify the firmware handles it gracefully without crashing, corrupting state, or accepting unauthorized commands. For the mobile companion app: OWASP Mobile Top 10 testing for both iOS and Android builds, including local data storage review (is patient insulin data encrypted at rest on the phone?), certificate pinning verification, and reverse engineering resistance.

For the cloud platform: API penetration testing covering authentication, authorization, injection, and business logic flaws. Automated vulnerability scanning (SCA) of all SBOM components against current CVE databases. And robustness testing of the full system under adversarial conditions: what happens to insulin delivery if the cloud goes down, if the BLE connection drops mid-command, or if the mobile app receives malformed data from a spoofed device?

Each test type needs a complete report: methodology, tools used, findings with severity ratings, and remediation evidence showing that identified issues were fixed and retested. All findings must trace back to risks in your cybersecurity risk assessment.

Partial Gap: Threat Modeling

You indicated you have a partially complete threat model, likely based on a generic template or covering only some interfaces. For a device with this many connectivity vectors (BLE, Wi-Fi, cloud APIs, mobile app) a comprehensive threat model is essential and needs to be fully device-specific.

Your threat model should use STRIDE (or attack trees) applied to every external interface: the BLE connection between device and phone (spoofing, replay attacks, unauthorized pairing), the Wi-Fi connection for cloud sync (MITM, network-level attacks, DNS poisoning), the cloud API surface (authentication bypass, injection, privilege escalation), and the OTA update channel (firmware tampering, rollback attacks, update spoofing). Don't forget physical attack vectors. For an insulin delivery device, an attacker with physical access could potentially access debug interfaces (JTAG, SWD) or extract firmware for reverse engineering. Each threat needs to trace back to a specific point in your security architecture views and forward to specific risk mitigations.

Partial Gap: SBOM

You have a component list but it's either manual or incomplete. For a multi-ecosystem device like yours (embedded C/C++ firmware, mobile apps in Swift/Kotlin, and a cloud backend) producing a complete, machine-readable SBOM requires a different toolchain for each layer.

For the embedded firmware, automated SBOM tools (like cdxgen) often miss SOUP components: your RTOS, vendor HAL libraries, and any proprietary protocol stacks. These need manual declaration with complete metadata (supplier, version, cryptographic hash, and purl identifier). For the mobile apps, your iOS CocoaPods/Swift PM and Android Gradle dependency trees should generate cleanly, but make sure you're also capturing platform SDK components. For the cloud backend, use Syft or equivalent to capture OS-layer components in your containers, not just application-level npm/pip packages. All three SBOMs need to merge into a unified device-level SBOM in CycloneDX or SPDX format. A manual spreadsheet won't meet FDA's machine-readable requirement.

Strong: Security Architecture

Your security architecture views are current and submission-ready. Given the complexity of your system (embedded device, mobile app, cloud platform, and two wireless protocols) maintaining all four FDA-required views (global system, multi-patient harm, updateability, and security use case views) is significant work, and it sounds like you've done it.

One recommendation: for your 510(k) filing, make sure your global system view explicitly shows all trust boundaries, particularly between the BLE link (device-to-phone) and the Wi-Fi/cellular link (phone-to-cloud). Reviewers will want to see that security controls are documented at each boundary crossing.

Company Context

Based on publicly available information on medflowhealth.com, MedFlow Health develops a connected insulin delivery platform targeting Type 1 and Type 2 diabetes management. The system consists of a wearable delivery device, an iOS/Android companion app ('MedFlow Connect'), and a cloud dashboard for healthcare providers ('MedFlow Clinic'). The website mentions BLE 5.0 connectivity, HIPAA-compliant cloud infrastructure on AWS, and an over-the-air firmware update capability. No current FDA clearances are listed, suggesting this is a first-time 510(k) submission. The attack surface is substantial: patient insulin dosing data traverses four trust boundaries (device to phone to cloud to clinician), and a compromise at any point could affect dosing accuracy or patient data confidentiality.

Want help closing these gaps in 30 days?

Our Cybersprint team can help you complete documentation, testing evidence, and submission-ready artifacts on an accelerated timeline.

This is a sample report generated from a fictional self-assessment. Your actual report will be personalized to your specific device architecture, regulatory pathway, and gap areas.

Take the assessment →