FDA’s First-Ever Guidance on AI-Enabled Medical Devices throughout the Total Product Lifecycle: What Developers Need to Know.
With the significant transformations in the medical devices industry over the past few years, many speculate about its direction in 2025. The COVID-19 pandemic reshaped the sector dramatically, and while it has settled into a “new normal,” further changes are on the horizon, driven by advancements in artificial intelligence (AI) and machine learning (ML).
While the US Food & Drug Administration (FDA) received the first request for an AI-enabled device in 1995, the number of submissions has spiked in recent years, and it has now reached over 1,000 AI-enabled authorized devices.
The Integration of AI into medical devices has the potential to revolutionize healthcare by improving accuracy, efficiency, and patient outcomes. However, it also introduces challenges such as data privacy, bias mitigation, regulatory compliance, and the need for interdisciplinary collaboration between medical professionals, engineers, data scientists, and ethicists.
In response, last week, the FDA issued a draft guidance explaining what information developers should now include in premarket submissions for AI-enabled devices.
The document, titled “Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations”, provides guidance on the design, development, and maintenance of such devices to ensure they remain safe and effective. It addresses critical topics like transparency, bias mitigation, and postmarket monitoring requirements.
The FDA reviews medical devices through an appropriate premarket pathway, such as premarket clearance (510(k)), De Novo classification or premarket approval. The draft guidance also provides recommendations on designing and developing AI-enabled devices throughout the Total Product Life Cycle, (TPLC).
As stated by Troy Tazbaz, director of the Digital Health Center of Excellence within the FDA’s Center for Devices and Radiological Health, “it’s important to recognize that there are specific considerations unique to AI-enabled devices”. It adds “today’s draft guidance brings together relevant information for developers, shares learnings from authorized AI-enabled devices and provides a first point-of-reference for specific recommendations that apply to these devices”.
9 Key Considerations for the Design and Development of AI-enables Medical Devices.
- Device Description
- User Interface and Labeling
- Risk Assessment
- Data Management
- Model Description and Development
- Validation
- Device Performance Monitoring
- Cybersecurity
- Public Submission Summary
Let’s dive deep into each of these considerations:
1 . Device description
In particular, the draft guidance provides information that should be included in a marketing submission as part of the device description. This list includes:
- A statement that AI is used in the device
- Device inputs and outputs
- The role of AI in achieving the device’s intended use
- The intended users and their requisite expertise
- The use environment
- Installation and maintenance procedures and any calibration or configuration procedures.
2 . User Interface and Labeling
The draft guidance also explains that a device’s user interface should include labeling, addressing the following information:
- Statement that AI is used in the device
- Model inputs
- Model outputs
- Automation
- Model architecture
- Model development data
- Performance data
- Device performance metrics
- Performance monitoring
- All known limitations
- Installation and use
- Customization
- Metrics or visualization to provide context to the model output
- Patient/caregiver information
FDA recommends using a “model card” in the device labeling to communicate information about the AI-enabled device and even provides an example.
3 . Risk Assessment
As with other devices containing software, the draft guidance states that AI-enabled devices should include a risk management file, considering hazards across the Total Product Life Cycle (TPLC).
4 . Data Management
The draft guidance explains that, for an AI-enabled device, the model (i.e., the algorithm and the data used to train it) is part of the device’s mechanism of action. For this reason, the FDA recommends clearly describing the data management practices and characterization used in developing and validating the AI-enabled device. Data management is also important in identifying and mitigating bias through the inclusion of representative data in training and validation datasets. Manufacturers should provide evidence of sufficient segregation of training and validation datasets to address issues of bias and overfitting.
5 . Model Description and Development
The software description should include an explanation of each model used as part of the AI-enabled device, including a description of model inputs and outputs, model architecture, features, feature selection process and any loss function(s) used for model design and optimization, and model parameters.
Additionally, the software description should explain how the model was trained (e.g., optimization methods, training paradigms), metrics and results obtained for any tuning evaluation, pre-trained models, ensemble methods, how any thresholds were determined, and any calibration of the model output.
6 . Validation
For an AI-enabled device, the draft guidance explains that validation includes ensuring that the device will perform its intended use safely and effectively (including whether users consistently and correctly receive, understand, interpret, and apply information related to the AI-enabled device).
7 . Device Performance Monitoring
The draft guidance emphasizes the importance of independent datasets for performance validation. Also, a marketing submission must address the performance of AI-enabled devices in real-world environments and the risk that they will change or degrade over time. As stated in the FDA Digital Health and Artificial Intelligence Glossary,“medical AI-enabled products can experience data drift due to statistical differences between the data used for model development and data used in clinical operation due to variations between medical practices or context of use between training and clinical use, and changes in patient demographics, disease trends, and data collection methods over time”. FDA recommends that manufacturers proactively monitor, identify, and address device performance changes.
8 . Cybersecurity
Manufacturers of AI-enabled devices must include in their marketing submissions details on AI risks that can be impacted by cybersecurity threats. The draft guidance provides a list of examples of AI risks (i.e. vulnerabilities) that can be affected by cybersecurity threats, including data poisoning, model inversion/stealing, model evasion, data leakage, overfitting, model bias, and performance drift.
9 . Public Submission Summary
Device premarket submissions typically include a public submission summary. The draft guidance emphasizes the importance of transparency concerning these devices in the public summary. Specifically, the draft guidance states that the public summary should include:
- A statement that AI is used in the device
- An explanation of how AI is used as part of the device’s intended use
- A description of the class of model
- A description of the development and validation datasets
- A description of the statistical confidence level of predictions
- And a description of how the model will be updated and maintained over time.
AI in the development of drug and biological products
In addition to this document, on the same date, the FDA released another draft guidance document on artificial intelligence (AI): “Considerations for the Use of AI to Support Regulatory Decision-Making for Drug and Biological Products”. While the first offers guidance to support the safety and effectiveness of AI-enabled devices, the second provides recommendations on the use of AI for the development of drug and biological products.
Final Insights
The FDA has consistently promoted a Total Product Life Cycle approach to overseeing medical devices, including those enabled by artificial intelligence. Some recent efforts include developing guiding principles for good machine learning practice (GMLP) and transparency for machine learning-enabled devices to help promote safe, effective, and high-quality machine learning models and a public workshop on fostering a patient-centered approach to AI-enabled devices, with discussions of device transparency for users.
While still a draft, this guidance offers valuable insights into the FDA’s approach to AI-enabled devices. The agency accepts public comments on the draft until April 7, 2025, and has scheduled a public webinar on the draft guidance, on February 18, 2025.
