Key Takeaways
- Emerging technologies—cloud computing, AI/ML, and post‑quantum cryptography—are expanding the attack surface of medical devices and can directly affect patient safety.
- Cybersecurity must be treated as a core design consideration, not an afterthought, with risk management shared among manufacturers, healthcare delivery organizations, and third‑party providers.
- Device constraints (limited power, memory, legacy software) and the shift to home‑based, patient‑managed use complicate traditional security controls.
- Effective mitigation relies on clear role definition (via SLAs and contracts), DevSecOps practices, threat modeling, Software Bills of Materials (SBOMs), resilient architecture (multi‑region backups, offline modes), and ongoing validation of AI/ML and cryptographic components.
- Transitioning to post‑quantum cryptography faces technical hurdles (greater memory/processing needs), interoperability with legacy systems, and variable timelines, especially for long‑lived implantables.
Overview of MITRE’s Analysis on Evolving Technologies
MITRE’s discussion paper, Cybersecurity Risk Analysis for Medical Devices in the Era of Evolving Technologies, warns that the rapid integration of cloud services, artificial intelligence/machine learning (AI/ML), and post‑quantum cryptography is reshaping the cybersecurity risk landscape for medical devices. Drawing on interviews with manufacturers, healthcare providers, regulators, and security vendors, the paper emphasizes that these innovations introduce distinct and evolving attack surfaces that can impair device functionality and, in worst cases, jeopardize patient safety. Consequently, cybersecurity must be elevated from an after‑thought to a fundamental design consideration, with risk management viewed as a shared, multi‑stakeholder responsibility that adapts to distributed care models and continuously advancing technologies.
Scope of Medical Devices and Shifting Operational Context
The analysis notes that today’s medical device ecosystem spans everything from implantable pacemakers and portable infusion pumps to large hospital systems such as MRIs and CT scanners. Many of these devices operate under tight constraints—limited power, memory, and processing capacity—which restrict the cybersecurity controls that can be applied. At the same time, larger devices often run outdated hardware and software due to their long operational lifetimes. Moreover, the operational context is moving beyond tightly managed hospital environments into ambulatory and home settings, where patients themselves manage devices through connected apps and wearables. This shift reduces the oversight that healthcare delivery organization (HDO) staff can exercise, thereby complicating risk ownership and accountability.
Resource Limitations, Legacy Systems, and Interconnectivity Challenges
MITRE highlights that the expanding ecosystem introduces dependencies on third‑party vendors and blurs accountability lines. Devices are increasingly interconnected with other medical products and health IT systems, creating a web of dependencies that can amplify the impact of a single vulnerability. Legacy software and hardware further impede the adoption of modern security measures, while long device lifecycles mean that security patches may be difficult or costly to deploy. The combination of constrained resources, outdated components, and extensive interconnectivity makes traditional, siloed cybersecurity approaches inadequate for protecting patient safety in this complex landscape.
Cloud Computing: Risks and Mitigation Strategies
When manufacturers adopt cloud computing, they face cybersecurity challenges in both acquisition/deployment and software development. While cloud services can lower costs and improve efficiency, they shift control away from both manufacturers and HDOs, introducing third‑party risks that are not always fully understood or regulated. Cloud service providers can become systemic points of failure; a disruption—such as a ransomware attack on Elekta’s cloud services that halted cancer treatment at over 170 sites—can simultaneously affect hundreds of facilities. To mitigate these risks, MITRE recommends:
- Defining clear roles and responsibilities in contracts and service level agreements (SLAs), referencing frameworks like ISO 13485:2016 and NIST cloud security guidance.
- Embedding DevSecOps practices, including secure CI/CD pipelines, container and orchestration security, throughout the product lifecycle.
- Designing resilient architectures that incorporate cloud components into Software Bills of Materials (SBOMs), map trust boundaries across the full cloud stack, and leverage tools such as MITRE ATT&CK cloud matrices and CSP‑native security solutions.
- Ensuring preparedness through multi‑region provisioning, local caching, offline operation modes, and geographically distributed backups to maintain care continuity during a cloud‑targeted cyberattack.
AI/ML Integration: Distinct Risks and Defensive Measures
Incorporating AI/ML into medical devices introduces a distinct and serious set of cybersecurity risks that span the entire device lifecycle. Attacks can poison training data to cause misdiagnoses, manipulate model inputs via adversarial examples or prompt injections, or exploit AI‑generated code that harbors hard‑to‑detect bugs. Membership inference attacks may reveal whether individuals were part of training datasets, raising HIPAA‑related privacy concerns. Because AI/ML outputs are stochastic and non‑deterministic, traditional debugging techniques often fail, and emergent defenses such as guardrails and Retrieval Augmented Generation remain immature.
To address these challenges, manufacturers should:
- Secure the learning environment by protecting training data, parameters, models, prompts, and associated systems, and maintain strict separation between training and test datasets to prevent data poisoning.
- Implement guardrails, robust testing mechanisms, and techniques such as prompt engineering and hyperparameter tuning, complemented by automated testing and manual red‑team exercises by subject‑matter experts.
- Fully embed AI/ML security within the broader software security program through threat modeling of AI/ML components, least‑privilege access controls, trust‑boundary enforcement (especially for cloud‑ or third‑party AI), and integrity validation of outputs.
- Conduct diligent risk and liability analysis when acquiring AI/ML components, weighing cybersecurity risk against potential impacts on patient safety, product correctness, and data confidentiality.
Post‑Quantum Cryptography: Transition Hurdles and Considerations
MITRE urges manufacturers and healthcare organizations to begin planning a transition to post‑quantum cryptography (PQC) to counter the future threat posed by quantum computers to current encryption and digital signatures. While viable PQC alternatives exist for confidentiality and authentication, migrating away from widely used legacy algorithms is complex. Three interconnected technical challenges arise:
- PQC algorithms generally require more memory, code, processing power, and larger message sizes, translating into higher hardware demands and costs.
- Interoperability with legacy devices remains a vulnerability; new PQC‑enabled systems must interface with older equipment lacking post‑quantum controls, forcing endpoints to handle multiple cryptographic standards simultaneously and increasing complexity while quantum risk persists as long as legacy systems remain.
- Transition timelines vary widely based on engineering constraints, financial considerations, legal mandates, device lifespan, and the feasibility of updating versus replacing existing devices; implantable devices pose particular difficulty because altering cryptography may necessitate a medical procedure that is justified only if it serves the patient’s broader clinical interests.
MITRE advises a dual‑track strategy: securing newly built devices against quantum threats while developing an organizational roadmap that protects operations and patients across all cryptographic systems—including those not directly controlled—during the migration period.
Conclusion and Recommendations
In closing, MITRE acknowledges that emerging technologies promise meaningful advances in patient care but simultaneously introduce new cybersecurity risks. Managing these risks does not require an entirely new paradigm; rather, it builds on existing practices such as maintaining SBOMs, conducting threat modeling, and applying incremental adaptations that encompass cloud, AI/ML, and cryptographic components. Governance is central: clear contracts and SLAs delineate cybersecurity expectations, while roadmaps for adoption and migration enable smooth transitions. The only constant is change, and manufacturers must develop medical devices that remain updatable throughout their lifetimes, ensuring they can continue to deliver needed care without becoming vulnerabilities that threaten patients or the care environment. By treating cybersecurity as a core, shared responsibility and continuously evolving defenses alongside technology, the healthcare sector can harness innovation while safeguarding patient safety.