AI Architecture for Medical Devices: Ensuring Safety and Compliance in an Emerging Field

ai-architecture-for-medical-devices.md

AI Architecture for Medical Devices: Ensuring Safety and Compliance in an Emerging Field

Abstract

Artificial intelligence (AI) is revolutionizing the medical device industry, offering unprecedented opportunities for improved diagnostics, treatment, and patient outcomes. However, the integration of AI into medical devices presents significant challenges related to regulatory compliance, data privacy, and ethical considerations. This report examines the current landscape of AI architecture in medical devices as of March 2025, focusing on regulatory frameworks, industry leaders, implementation strategies, and ethical governance. We analyze the roles and responsibilities of AI architects in navigating these complexities while ensuring that AI-enabled medical devices remain safe, effective, and compliant with applicable regulations. Drawing from authoritative sources including FDA guidance documents and industry analyses, we provide a comprehensive overview of AI lifecycle management, HIPAA compliance requirements, and ethical AI governance frameworks. The findings highlight the critical need for robust AI architecture that addresses the unique challenges posed by AI in healthcare settings, providing valuable insights for medical device developers, regulators, and healthcare providers.

1. Introduction

The integration of artificial intelligence into medical devices represents one of the most significant technological advancements in healthcare over the past decade. AI-enabled medical devices can analyze vast amounts of data, recognize patterns, and make predictions with increasing accuracy, often exceeding human capabilities. These capabilities are transforming various areas of healthcare, including diagnostic imaging, clinical decision support, patient monitoring, and treatment planning.

According to a press release from the FDA in January 2025, "The FDA has authorized more than 1,000 AI-enabled devices through established premarket pathways" [1]. This rapid growth reflects the increasing recognition of AI's potential to enhance healthcare delivery and improve patient outcomes. In 2023 alone, the FDA approved a record 221 AI-powered medical devices, and in the first half of 2024, an additional 107 were approved [2].

However, this growth also brings significant challenges. AI-enabled medical devices must navigate complex regulatory frameworks, ensure data privacy and security, address ethical concerns, and maintain clinical effectiveness throughout their lifecycle. These challenges require specialized expertise in AI architecture, regulatory compliance, and healthcare integration.

The role of AI architecture is essential within medical device companies, responsible for defining and implementing AI architecture, integrating models with enterprise systems, and ensuring regulatory compliance. This report explores the key aspects of this role, the challenges it addresses, and the strategies for successful implementation of AI in medical devices.

2. Regulatory Landscape for AI in Medical Devices

2.1 FDA Regulatory Framework

The U.S. Food and Drug Administration (FDA) has been actively developing a regulatory framework to address the unique challenges posed by AI and machine learning in medical devices. The FDA's approach recognizes that AI-enabled medical devices differ from traditional medical devices in their ability to learn and evolve over time, presenting novel regulatory challenges.

In January 2021, the FDA published the "Artificial Intelligence and Machine Learning Software as a Medical Device Action Plan" (AI/ML SaMD Action Plan), followed by guidance documents addressing various aspects of AI regulation [3]. Most recently, on January 6, 2025, the FDA published the Draft Guidance: "Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations" [4]. This draft guidance proposes both lifecycle considerations and specific recommendations to support marketing submissions for AI-enabled medical devices, with a comment period open until April 7, 2025 [5].

The January 2025 guidance is intended to be "the first guidance to provide comprehensive recommendations for AI-enabled devices throughout the total product lifecycle, providing developers an accessible set of considerations that tie together design, development, maintenance and documentation recommendations to help ensure safety and effectiveness of AI-enabled devices" [1]. This guidance complements the previously issued final guidance on predetermined change control plans for AI-enabled devices, which provides recommendations on how to proactively plan for device updates once the product is on the market.

According to the Federal Register notice, the draft guidance "proposes recommendations for the design, development, and implementation of AI-enabled devices that sponsors may wish to consider using throughout the total product lifecycle (TPLC)" [6]. The FDA is specifically requesting public comment on several aspects, including:

• The draft guidance's alignment with the AI lifecycle
• The adequacy of the recommendations to address concerns raised by emerging technology such as generative AI
• The approach to performance monitoring
• The type of information about AI-enabled devices that should be conveyed to users and the most appropriate approach to deliver that information [1]

2.2 AI Lifecycle Management

The FDA's regulatory approach increasingly focuses on AI Lifecycle Management (AILC) as a framework for ensuring the safety and effectiveness of AI-enabled medical devices throughout their lifecycle. In July 2024, the FDA's Digital Health Center of Excellence (DHCoE) published a blog post outlining a lifecycle management approach for AI-enabled healthcare [7].

The AILC model extends traditional Software Development Lifecycles (SDLCs) to address the unique characteristics of AI, particularly its ability to learn and adapt in real-world settings. The FDA's model presents the AI lifecycle as a wheel with seven stages:

1. Planning and Design: Problem definition, data collection planning, algorithm selection, and model design
2. Data Collection and Management: Data suitability, quality assurance, privacy and security, governance
3. Model Building and Tuning: Model selection, hyperparameter tuning, feature engineering, validation
4. Verification and Validation: Evaluation metrics, data verification, validation strategies, documentation
5. Model Deployment: Scalability, integration, monitoring and logging, compliance
6. Operations and Monitoring: Real-time monitoring, alerting, performance optimization, drift detection
7. Real-world Performance Evaluation: KPI definition, production data collection, feedback collection, continuous improvement [8]

The FDA sees this model as "a playbook for helping to assess standards and a launchpad for other projects, such as the development of a checklist for the systematic creation and evaluation of AI" [7]. According to the DHCoE, "Standards play a role in the AILC by helping to ensure quality, facilitate interoperability, and promote ethical practices. They also help guide development, enhance transparency, support compliance, encourage innovation, and build trust" [9].

2.3 Predetermined Change Control Plans

One crucial component of the FDA's approach to AI/ML-enabled medical devices is the concept of Predetermined Change Control Plans (PCCPs). PCCPs provide a framework for making iterative improvements to AI algorithms while maintaining regulatory compliance.

In December 2024, the FDA finalized recommendations for streamlining the approval process for medical devices that use artificial intelligence. According to the American Hospital Association, "The guidance recommends information to include in a predetermined change control plan as part of a marketing submission for a medical device using AI. The PCCP should include a description of the device's planned modifications; methods to develop, validate and implement the modifications; and an assessment of the modification's impacts" [10].

The Veranex guide on navigating FDA's latest guidance (January 2025) explains that PCCPs are designed to support iterative improvements in software and simplify regulatory processes. According to the guide, a PCCP must include:

1. Description of Planned Modifications: Details of the changes anticipated for the AI model
2. Performance Evaluation Protocols: Predefined acceptance criteria and methods for verifying and validating modifications
3. Impact Assessment: Analysis of potential benefits and risks associated with implementing post-approval modifications and plans to mitigate risks [11]

This approach allows manufacturers to update their AI models within predefined parameters without requiring new regulatory submissions for each update, addressing one of the key challenges of regulating adaptive AI systems.

3. Industry Leaders in AI-Enabled Medical Devices

3.1 Major Players and Their Approaches

Several large medical technology companies have established themselves as leaders in the development and deployment of AI-enabled medical devices:

1. GE Healthcare: According to Radiology Business, GE Healthcare led the field with 72 AI-enabled devices cleared or authorized by the FDA as of May 2024 [12]. GE Healthcare's approach to AI falls into three categories:

   • Products that make imaging more efficient (like Air Recon DL, which enhances image quality and reduces MRI scan time)
   • AI that aggregates data from various sources to support clinical decisions
   • Enterprise-level systems for planning [13]

2. Siemens Healthineers: Radiology Business reported that Siemens Healthineers had 64 AI-enabled devices on the FDA's list as of May 2024 [12]. Siemens Healthineers focuses on integrating AI into imaging machines and developing diagnostic algorithms. Examples include features to optimize patient positioning for MRI scans and algorithms to diagnose conditions like coronary artery calcification [13].

3. Canon: With 28 FDA-authorized AI-enabled devices according to Radiology Business [12], Canon is a significant player in the AI medical device market, particularly in radiology applications.

4. Philips: With 27 FDA-authorized AI-enabled devices [12], Philips has developed AI solutions focused on patient care, including mobile connectivity and visualization tools for medical outcomes.

5. Aidoc: This Tel Aviv-based company specializes in computer-aided triage systems and had 23 FDA-authorized AI-enabled devices as of May 2024 [12].

Market research shows that Siemens Healthineers held the highest market share in North American Diagnostic Imaging at 43.09% as of 2023 [14]. According to another market report, "In 2022, a major share of the Multimodal imaging market was held by GE healthcare (US), Koninklijke Philips N.V. (Netherlands), Canon Medical Systems Corporation (Japan), Siemens AG (Germany)" [15].

3.2 Market Trends and Applications

The application of AI in medical devices spans multiple specialties, though radiology remains the most prevalent. According to Goodwin Law, "From the 1990s to mid-2024, radiology devices accounted for about 76% of all AI medical device approvals, far more than the runner-up — cardiovascular — which captured 10% of approvals" [2].

The U.S. medical imaging device market is projected to reach $10.9 billion by 2030, led by GE Healthcare, Siemens Healthineers, and Philips, according to a report by iData Research [16]. The report notes that "The field of imaging technology is continually progressing, with exciting developments like clearer images, quicker scans and the use of artificial intelligence (AI) and machine learning" [16].

Examples of AI-enabled devices include:

• AI-Rad Companion by Siemens Healthineers: Provides quantitative and qualitative measurements of clinical images and analysis of clinical data
• LumineticsCore by Digital Diagnostics: Automatically detects diabetic retinopathy by analyzing images
• Atrial fibrillation history feature by Apple: Uses Apple Watch data to show users how often they had signs of a common heart arrhythmia [13]

Looking ahead, industry experts see potential for AI expansion into areas like robotics, where AI could support pre-surgical planning, augmented reality applications during procedures, and post-surgical analytics [13].

4. AI Architecture for Medical Devices

4.1 Core Components of AI Architecture

Effective AI architecture for medical devices typically includes several key components that must be carefully designed and integrated to ensure safety, effectiveness, and regulatory compliance:

1. Data Acquisition and Management: Systems for collecting, storing, and managing healthcare data from various sources, including electronic health records, medical imaging, and patient monitoring devices. The FDA's AILC model emphasizes the importance of data suitability, quality, privacy, security, and governance at this stage [8].

2. Model Development and Selection: Frameworks and methodologies for developing, selecting, and training AI models for specific medical applications. According to the FDA's AILC model, this includes model selection, hyperparameter tuning, feature engineering, cross-validation, ensemble methods, and model interpretability [8].

3. Verification and Validation: Systems for thoroughly testing and validating AI models to ensure they meet safety, effectiveness, and performance requirements. The FDA emphasizes the importance of evaluation metrics, data verification, deployment testing, validation strategies, model comparison, error analysis, and robust documentation [8].

4. Deployment Infrastructure: Technical infrastructure for deploying trained models in clinical settings, including cloud or edge computing resources, APIs, and integration with existing healthcare IT systems.

5. Monitoring and Feedback: Tools for continuous monitoring of AI model performance, collecting feedback, and identifying potential issues or areas for improvement. The FDA's AILC model highlights the importance of real-time monitoring, alerting and notifications, logging and auditing, and drift detection in the operations phase [8].

6. Security and Privacy: Robust security measures to protect sensitive healthcare data and ensure compliance with privacy regulations such as HIPAA. This includes encryption, access controls, and secure data transmission.

7. Explainability and Transparency: Systems for making AI decisions interpretable and transparent to healthcare providers and regulators, addressing the "black box" problem of complex AI algorithms.

4.2 MLOps for Medical Devices

Machine Learning Operations (MLOps) is increasingly important for managing the complexity of AI-enabled medical devices. MLOps extends DevOps principles to machine learning, focusing on automating and operationalizing the end-to-end ML lifecycle.

Key aspects of MLOps for medical devices include:

1. Model Versioning and Tracking: Systems for tracking different versions of AI models, including their training data, hyperparameters, and performance metrics.

2. Automated Testing and Validation: Automated processes for testing and validating AI models against predefined requirements and performance metrics.

3. Continuous Integration and Deployment: Tools and processes for integrating model updates into the broader device software and deploying them safely.

4. Performance Monitoring: Systems for continuous monitoring of AI model performance in real-world settings, identifying potential drift or degradation.

5. Documentation and Traceability: Comprehensive documentation of the entire AI lifecycle, ensuring traceability for regulatory compliance.

MLOps is especially critical for medical devices that implement Predetermined Change Control Plans (PCCPs), as it provides the infrastructure and processes needed to manage the continuous learning and adaptation of AI models within regulatory boundaries.

4.3 Cloud vs. Edge Computing

AI-enabled medical devices often leverage both cloud and edge computing architectures, each with distinct advantages and limitations:

Cloud Computing:

• Provides virtually unlimited computational resources for training complex AI models
• Enables centralized data storage and management
• Facilitates continuous learning and improvement based on aggregated data
• Requires robust internet connectivity and raises privacy concerns when processing sensitive healthcare data

Edge Computing:

• Processes data locally on the device itself
• Reduces latency for time-critical applications
• Enhances privacy by keeping sensitive data local
• Reduces dependence on internet connectivity
• Limited by the computational capabilities of the device

Many modern AI-enabled medical devices use hybrid approaches, leveraging cloud resources for training and edge computing for inference, balancing performance, privacy, and reliability requirements.

Microsoft's Azure AI Foundry, for example, provides foundation models for healthcare that facilitate AI-powered analysis of various medical data types, including medical imaging for radiology, pathology, and ophthalmology [17]. These cloud-based resources can be combined with edge computing capabilities to create effective AI architectures for medical devices.

5. Data Privacy and Security Considerations

5.1 HIPAA Compliance and AI

The Health Insurance Portability and Accountability Act (HIPAA) establishes critical requirements for protecting patient health information in the United States. AI-enabled medical devices must navigate these requirements carefully, particularly when processing Protected Health Information (PHI).

According to a 2023 article from Accountable HQ, "With AI's increasing role in healthcare, attention to HIPAA compliance becomes essential... As AI applications often involve handling sensitive health data, they must adhere to these regulations. However, implementing AI while also ensuring HIPAA compliance can be challenging" [18].

Key HIPAA considerations for AI in medical devices include:

1. Data De-identification: Proper de-identification of PHI used for training AI models, following HIPAA's de-identification standards. According to Tebra, "Before feeding medical images into the AI system, the hospital deploys a rigorous process to de-identify the data. This involves removing all direct and indirect identifiers to ensure that individual patients could not be linked to the images" [19].

2. Business Associate Agreements: Establishing appropriate agreements when sharing PHI with third-party AI developers or service providers. According to MobiDev, "If you want to use a third-party AI platform or APIs, you must find one that provides a BAA, as required by HIPAA's regulations" [20].

3. Access Controls: Implementing robust access controls to ensure that PHI is only accessible to authorized individuals.

4. Audit Trails: Maintaining comprehensive audit trails of all access to and processing of PHI.

5. Data Minimization: Using only the minimum necessary PHI for the intended purpose.

The dynamic and evolving nature of AI technology presents unique challenges for HIPAA compliance. Accountable HQ notes that "AI applications require vast amounts of data for training, which may include sensitive health information. Ensuring this data is adequately de-identified to protect patient privacy, while still useful for AI, is a complex task" [18].

5.2 Data Governance Frameworks

Robust data governance frameworks are essential for managing the complex privacy and security requirements of AI-enabled medical devices. These frameworks typically include:

1. Data Classification: Categorizing data based on sensitivity and regulatory requirements.

2. Data Quality Management: Ensuring the accuracy, completeness, and reliability of data used for AI.

3. Data Lifecycle Management: Managing data from collection through processing, storage, and eventual deletion.

4. Risk Assessment: Regularly assessing and mitigating risks associated with data processing.

5. Security Controls: Implementing appropriate technical and organizational measures to protect data.

6. Training and Awareness: Ensuring that all personnel are aware of and follow data protection requirements.

A key challenge with AI systems is the potential for re-identification of supposedly de-identified data. Reuters notes, "Even if the data is initially de-identified in compliance with the applicable standard, AI products present some unique challenges to the de-identification process. As an AI product develops and expands, often new data elements are added to the AI system or the amount of data in a particular element is increased creating a potential for privacy issues" [21].

6. Ethical AI Governance

6.1 Ethical Frameworks for AI in Healthcare

The use of AI in medical devices raises significant ethical considerations, including fairness, transparency, accountability, and patient autonomy. Several ethical frameworks have been developed to guide the responsible deployment of AI in healthcare.

Holistic AI, a company focused on AI governance, explains that "effective AI governance is essential to address these biases, which includes implementing robust policies and regulations, establishing ethical frameworks, and creating accountability mechanisms" [22]. They outline three key components of AI governance:

1. Policies and Regulations: "Effective development and deployment of AI systems require robust policies and regulations. These policies must target specific areas sensitive to bias, such as data collection, algorithm design, and decision-making processes" [22].

2. Ethical Frameworks: "Ethical frameworks establish guidelines for the responsible use of AI, highlighting principles such as fairness, accountability, and transparency. These frameworks assist developers and organizations in making ethical decisions throughout the AI lifecycle, from conception to deployment and beyond" [22].

3. Accountability Mechanisms: "Establishing clear accountability mechanisms is important for holding developers, organizations, and stakeholders responsible for the outcomes of AI systems. Examples of this are regular audits, impact assessments, and oversight committees that review AI systems for potential biases" [22].

6.2 Bias and Fairness in Medical AI

AI systems trained on historical healthcare data may inherit and amplify existing biases, potentially leading to disparate impacts on different patient populations. Addressing bias and ensuring fairness in medical AI requires:

1. Diverse and Representative Training Data: Ensuring that AI systems are trained on data that represents the diversity of patient populations.

2. Bias Detection and Mitigation: Regularly testing AI systems for bias and implementing measures to mitigate identified biases. Holistic AI recommends "techniques like bias detection and correction" and emphasizes that "algorithm development guidelines should ensure that AI systems are designed with fairness and equity in mind" [22].

3. Demographic Performance Analysis: Evaluating AI performance across different demographic groups to identify potential disparities.

4. Continuous Monitoring: Monitoring AI performance in real-world settings to detect and address emerging biases. According to Holistic AI, "Organizations need to commit to ongoing monitoring and updating of their AI systems and governance frameworks to address new challenges and incorporate fresh insights" [22].

The FDA's January 2025 guidance specifically addresses bias in AI-enabled medical devices. According to Orrick, the FDA recommends that companies "Ensure representativeness in data collection in developing, testing and monitoring the device throughout the product lifecycle, as well as in the evaluation of performance, across all relevant demographic groups of intended use (e.g., race, ethnicity, sex, and age)" [23].

6.3 Implementation Strategies

Successful implementation of ethical AI governance in medical devices requires several key strategies:

1. Stakeholder Involvement: Holistic AI emphasizes that "involving a diverse range of stakeholders for effective AI governance" is important. "Stakeholders include not only AI developers and data scientists but also ethicists, sociologists, legal experts, and representatives from affected communities" [22].

2. Transparency and Explainability: "Organizations must ensure that their AI systems are transparent and explainable, allowing users to understand how decisions are made and what factors influence those decisions" [22]. This is particularly important in healthcare, where understanding the rationale behind AI recommendations is crucial for clinical decision-making.

3. Regular Audits and Assessments: "To uncover and address biases in AI systems, it is necessary to implement regular audits and assessments. This process should include both internal reviews and evaluations by independent third-party auditors to ensure objectivity" [22].

4. Continuous Improvement: "AI governance requires a continuous and evolving approach. Organizations need to commit to ongoing monitoring and updating of their AI systems and governance frameworks to address new challenges and incorporate fresh insights" [22].

By implementing these strategies, medical device developers can ensure that their AI-enabled devices are not only safe and effective but also fair, transparent, and aligned with ethical principles.

7. The Role of AI Architects in Medical Devices

7.1 Key Responsibilities

AI architects in the medical device industry play a crucial role in designing, implementing, and maintaining AI systems that meet both technical requirements and regulatory standards. Their responsibilities typically include:

1. AI Strategy Development: Defining the overall AI strategy for medical devices, aligned with business objectives and regulatory requirements.

2. Architecture Design: Designing scalable and compliant AI architectures that address the unique challenges of healthcare applications.

3. Technical Leadership: Providing technical guidance and leadership for AI development teams, ensuring best practices are followed.

4. Regulatory Compliance: Ensuring AI systems meet FDA and other regulatory requirements, including documentation for premarket submissions.

5. Cross-functional Collaboration: Working with clinical, regulatory, quality, and software teams to integrate AI capabilities into medical devices.

6. Lifecycle Management: Implementing AI lifecycle management practices to ensure continuous monitoring, improvement, and compliance.

The FDA's evolving regulatory framework has increased the complexity of these responsibilities, requiring AI architects to stay current with the latest guidance and best practices.

7.2 Skills and Qualifications

Effective AI architects in the medical device industry need a unique combination of skills and qualifications:

1. Technical Expertise: Strong knowledge of AI/ML frameworks, algorithms, and best practices for healthcare applications.

2. Regulatory Knowledge: Understanding of FDA and other regulatory requirements for AI-enabled medical devices.

3. Healthcare Domain Knowledge: Familiarity with clinical workflows, healthcare data, and medical terminology.

4. Data Management: Expertise in managing and securing healthcare data in compliance with privacy regulations.

5. Communication Skills: Ability to communicate complex technical concepts to diverse stakeholders, including clinicians, executives, and regulators.

6. Ethical Considerations: Awareness of ethical issues related to AI in healthcare and approaches to address them.

As the field continues to evolve, AI architects must commit to ongoing learning and professional development to stay at the forefront of both technological advancements and regulatory changes.

8. Future Trends and Challenges

8.1 Emerging Technologies

Several emerging technologies are likely to shape the future of AI in medical devices:

1. Federated Learning: Enables AI models to learn from decentralized data without transferring sensitive information, addressing privacy concerns.

2. Multimodal AI: Integrates multiple data types (imaging, clinical records, genomics) for more comprehensive analysis. According to MedTech Dive, Siemens Healthineers is particularly interested in multimodal AI, "which brings together different pieces of data, such as imaging, laboratory results and a patient's medical history" [13].

3. Foundation Models: Large, versatile models capable of addressing diverse healthcare applications, such as those being developed by Microsoft's Azure AI Foundry [17].

4. AI at the Edge: Increasingly powerful edge computing capabilities enabling more sophisticated AI directly on medical devices.

8.2 Recent Challenges

Despite the promise of AI in medical devices, the industry faces some recent challenges that may impact future developments:

1. Regulatory Evolution: The regulatory landscape continues to evolve, with new guidance documents being issued regularly. The FDA's January 2025 draft guidance is open for comments until April 7, 2025, suggesting that regulatory requirements may continue to be refined [1].

2. Political Changes: According to STAT News, "Layoffs at the FDA appear to have hit the AI and digital health staff particularly hard at a time when federal regulators are scrambling for ways to keep tabs on hospitals and insurers that are embracing the technologies" [24]. This development, reported in February 2025, could impact the FDA's ability to review and approve AI-enabled medical devices efficiently.

3. Data Privacy Concerns: As AI becomes more integrated into healthcare, concerns about data privacy and security continue to grow. The challenge of maintaining HIPAA compliance while leveraging AI capabilities remains significant.

8.3 Addressing Skills Gaps

The rapidly evolving field of AI in healthcare faces significant skills gaps, requiring:

1. Specialized Training: Development of training programs for AI in healthcare applications.

2. Interdisciplinary Education: Programs that bridge clinical, technical, and regulatory domains.

3. Knowledge Sharing: Platforms for sharing knowledge and best practices across the industry.

4. Recruitment and Retention: Strategies for attracting and retaining talent in this competitive field.

Addressing these skills gaps will be essential for realizing the full potential of AI in medical devices.

9. Conclusion

AI-enabled medical devices offer tremendous potential to transform healthcare delivery and improve patient outcomes. However, realizing this potential requires navigating complex regulatory, technical, ethical, and organizational challenges.

The FDA's evolving approach to regulating AI in medical devices, including its AI Lifecycle Management model and Predetermined Change Control Plans, provides a framework for addressing these challenges. Industry leaders like GE Healthcare, Siemens Healthineers, and Philips are driving innovation while working within these regulatory frameworks.

Effective AI architecture for medical devices must address the full lifecycle of AI systems, from data acquisition and model development to deployment, monitoring, and continuous improvement. This requires robust MLOps practices, careful consideration of cloud vs. edge computing approaches, and strong data governance frameworks.

HIPAA compliance and ethical considerations add additional layers of complexity, requiring AI architects to implement appropriate safeguards for patient privacy and measures to ensure fairness, transparency, and accountability.

As the field continues to evolve, ongoing collaboration among industry, regulators, healthcare providers, and patients will be essential to ensure that AI-enabled medical devices serve their ultimate purpose: improving the quality and accessibility of healthcare for all.

References

1. FDA. (2025, January). FDA Issues Comprehensive Draft Guidance for Developers of Artificial Intelligence-Enabled Medical Devices. Retrieved from https://www.fda.gov/news-events/press-announcements/fda-issues-comprehensive-draft-guidance-developers-artificial-intelligence-enabled-medical-devices

2. Goodwin Law. (2024, November 1). FDA Approvals Surge for AI-Enabled Medical Devices in 2024. Retrieved from https://www.goodwinlaw.com/en/insights/publications/2024/11/insights-technology-aiml-fda-approvals-of-ai-medical-devices

3. FDA. (2021, January). Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan. Retrieved from https://www.fda.gov/media/145022/download

4. FDA. (2025, January 6). Artificial Intelligence and Machine Learning in Software as a Medical Device. Retrieved from https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-and-machine-learning-software-medical-device

5. FDA. (2025, January 7). Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations. Retrieved from https://www.fda.gov/regulatory-information/search-fda-guidance-documents/artificial-intelligence-enabled-device-software-functions-lifecycle-management-and-marketing

6. Federal Register. (2025, January 7). Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations; Draft Guidance for Industry and Food and Drug Administration Staff; Availability. Retrieved from https://www.federalregister.gov/documents/2025/01/07/2024-31543/artificial-intelligence-enabled-device-software-functions-lifecycle-management-and-marketing

7. FDA. (2024, July 25). Blog: A Lifecycle Management Approach toward Delivering Safe, Effective AI-enabled Health Care. Retrieved from https://www.fda.gov/medical-devices/digital-health-center-excellence/blog-lifecycle-management-approach-toward-delivering-safe-effective-ai-enabled-health-care

8. Cosm. (2024). Navigating the AI Lifecycle Management Model: A Playbook for Medical Device Developers. Retrieved from https://www.cosmhq.com/resources-posts/navigating-the-ai-lifecycle-management-model-a-playbook-for-medical-device-developers

9. FDA. (2024, July 25). Blog: A Lifecycle Management Approach toward Delivering Safe, Effective AI-enabled Health Care. Retrieved from https://www.fda.gov/medical-devices/digital-health-center-excellence/blog-lifecycle-management-approach-toward-delivering-safe-effective-ai-enabled-health-care

10. American Hospital Association. (2024, December 5). FDA finalizes recommendations simplifying approval process for medical devices that use AI. Retrieved from https://www.aha.org/news/headline/2024-12-05-fda-finalizes-recommendations-simplifying-approval-process-medical-devices-use-ai

11. Veranex. (2025, January 21). Navigating FDA's Latest Guidance Updates for AI-Enabled Medical Devices: A Marketing Submission Guide. Retrieved from https://veranex.com/2025/01/21/navigating-fdas-latest-guidance-updates-for-ai-enabled-medical-devices-a-marketing-submission-guide/

12. Radiology Business. (2024, May 28). GE HealthCare leads with 72 AI-enabled devices; Siemens, Canon, Philips, Aidoc also make top 5. Retrieved from https://radiologybusiness.com/topics/artificial-intelligence/ge-healthcare-leads-72-ai-enabled-devices-siemens-canon-philips-aidoc-also-make-top-5

13. MedTech Dive. (2024, October 9). The number of AI medical devices has spiked in the past decade. Retrieved from https://www.medtechdive.com/news/fda-ai-medical-devices-growth/728975/

14. GlobalData. (2023). Market share of leading medical device players in Diagnostic Imaging (Geography: North America, 2023). Retrieved from https://www.globaldata.com/data-insights/hatcr/market-share-of-leading-medical-device-players-in-diagnostic-imaging--geography--north-america-502481/

15. MarketsandMarkets. (2023). GE healthcare (US) and Koninklijke Philips N.V. (Netherlands) are Leading Players in the Multimodal Imaging Market. Retrieved from https://www.marketsandmarkets.com/ResearchInsight/multimodal-imaging-market.asp

16. iData Research. (2023, October 16). GE Healthcare, Siemens Healthineers, and Philips Lead the US Medical Imaging Device Market to Reach $10.9B by 2030. Retrieved from https://fox2now.com/business/press-releases/ein-presswire/661742842/ge-healthcare-siemens-healthineers-and-philips-lead-the-us-medical-imaging-device-market-to-reach-10-9b-by-2030/

17. Microsoft. (2024). Foundation models for healthcare in Azure AI Foundry portal. Retrieved from https://learn.microsoft.com/en-us/azure/ai-studio/how-to/healthcare-ai/healthcare-ai-models

18. Accountable HQ. (2023, December 14). AI in Healthcare; What it means for HIPAA. Retrieved from https://www.accountablehq.com/post/ai-and-hipaa

19. Tebra. (2023, December 15). Healthcare AI and HIPAA privacy concerns: Everything you need to know. Retrieved from https://www.tebra.com/theintake/practice-operations/legal-and-compliance/privacy-concerns-with-ai-in-healthcare

20. MobiDev. (2024, November 28). How to Build HIPAA-Compliant AI Applications for Healthcare. Retrieved from https://mobidev.biz/blog/how-to-build-hipaa-compliant-ai-applications

21. Reuters. (2022, March 17). Data privacy and artificial intelligence in health care. Retrieved from https://www.reuters.com/legal/litigation/data-privacy-artificial-intelligence-health-care-2022-03-17/

22. Holistic AI. (2024, August 21). How to Mitigate Bias in AI Systems Through AI Governance. Retrieved from https://www.holisticai.com/blog/mitigate-bias-ai-systems-governance

23. Orrick. (2025, January). FDA Issues Draft Guidance on AI-Enabled Medical Devices. Retrieved from https://www.orrick.com/en/Insights/2025/01/FDA-Issues-Draft-Guidance-on-AI-Enabled-Medical-Devices

24. STAT News. (2025, February 17). Layoffs hit FDA's Center for Devices and Radiological Health. Retrieved from https://www.statnews.com/2025/02/16/fda-layoffs-center-for-devices-radiological-health-cdrh-artificial-intelligence-doge/

25. HIPAA Journal. (2024, May 2). Editorial: HIPAA, Healthcare Data, and Artificial Intelligence. Retrieved from https://www.hipaajournal.com/hipaa-healthcare-data-and-artificial-intelligence/

26. Davis Wright Tremaine. (2018, March). Privacy Please: HIPAA and Artificial Intelligence – Part I. Retrieved from https://www.dwt.com/blogs/privacy--security-law-blog/2018/03/privacy-please-hipaa-and-artificial-intelligence

27. Pragmatic Coders. (2024, November 6). Essential guide to 2025's HIPAA-compliant software development. Retrieved from https://www.pragmaticcoders.com/blog/essential-guide-to-2023s-hipaa-compliant-software-development

28. TechTarget. (2025, January). FDA issues AI-enabled device lifecycle management guidance. Retrieved from https://www.techtarget.com/HealthtechAnalytics/news/366617631/FDA-issues-AI-enabled-device-lifecycle-management-guidance