Optimizing and Accelerating Medical Device Regulatory Approvals with AI and RWiD

Artificial intelligence (AI) and real-world imaging data (RWiD) are fundamentally reshaping and accelerating regulatory approval for medical devices. By delivering more effective, accurate, and transparent regulatory efficiency to a historically manual, slow-motion approach, AI and RWiD are fundamentally changing the regulatory landscape. AI gives the ability to automate functions, reliability, predict, and approve compliance in near real-time, ultimately shortening approval time and error rate. RWiD integrates all of this rich, real-world data into the regulatory review process, amplifying confidence in the safety and effectiveness of devices, while allowing for expedited access to the marketplace. Ultimately, the administrative functions of device development, or regulatory submissions, are transformed by the convergence of AI and RWiD. Market access, innovation, and patient outcomes are all measurably enhanced through monitoring of device performance, clinical utility, and automatic registration of all events regulated by the FDA.

With new advancements in AI, the U.S. Food and Drug Administration (FDA) is adapting its pre-market approval processes to include this technological advancement and be able to pilot new regulations for medical devices. The FDA has released a number of guidance documents to ease the approval of AI-enabled medical devices due to the rapid implementation in the healthcare space.

In January 2021, the FDA published the “Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan,” which stands as the first federal framework of its kind outlining the regulatory path for AI/ML-based SaMDs. Then, in October of 2021, the FDA published a joint paper outlining guiding principles to support the development of safe and effective medical devices that incorporate AI/ML. In June 2024, the FDA released additional guidance on “Transparency for Machine Learning-Enabled Medical Devices (MLMDs): Guiding Principles,” which was intended to provide additional guidance on transparency for MLMDs. Importantly, the FDA's draft guidance from 2025 on "Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations" describes the approach of considering risk management across the total product lifecycle (TPLC) of the device.

Although the FDA is currently developing full regulations for AI/ML, significant principles are being advanced. One area of importance is ensuring that AI technologies provide real value to patients, not just hype. These regulations take into account the role that AI plays in "supporting" safe and effective patient care and reaffirm the value of the "human-in-the-loop" and human oversight with AI in workflow.

The Role of Real-World Imaging Data

In recent years, the medical device approval regulatory landscape has adapted to include advanced digital health solutions and the rapid advancements in AI, particularly in imaging diagnostics. A fundamental driver of this transformation is the expanding use of RWiD, which include:

‍

Accelerated Evidence Generation

Regulatory bodies are increasingly open to utilizing real-world data (RWD), particularly imaging data, to substantiate the safety, efficacy, and labeling of medical devices. Real-world imaging data, derived from actual patient care (sources include clinical registries, etc.), can not only accelerate evidence generation but also decrease the time and cost associated with traditional clinical trial methods.

‍

Impact on Approval Timelines

Between January 2020 and July 2024, 117 medical devices had utilized RWD in their FDA submissions, with 63% of these supporting their approval with the evidence. The 510(k) and PMA were the most frequent regulatory submissions that utilized the RWD, while the RWiD regulatory pathway was most often used in cardiovascular and radiology AI application disclosures, where study registry-based evidence provided feasibility for strong real-world clinical data. 

‍

Impact on Generalizability of AI Models

Kaushal and colleagues published a paper in JAMA in 2020, which showed that 71% of patient cohorts in the US that were used for training of deep learning models in medicine came from only 3 states: California, Massachusetts, and New York. One can imagine that the diversity from those populations was limited and therefore the generalizability of the trained AI models was limited as well. With Segmed, diverse RWiD can be provided from multiple states in the US and internationally, resulting in high quality training and validation data. This may result in improved generalizability of AI models. 

‍

How AI Enhances and Optimizes the Regulatory Process

Automated Regulatory Tech

Regulatory authorities are now developing AI tools for their use to centralize screening data, identify risk signals, and assess submissions, enhancing the quality of supervision and scope while allowing quicker reviews. Examples include open-source parameterized compliance-checking frameworks and software that reads source code for defects.

‍

Adaptive and Iterative Change Management

The FDA has introduced "Predetermined Change Control Plans" (PCCPs) in their new guidance. This allows manufacturers to provide evidence up front and anticipate new updates expected for the AI, which will streamline the ability of regulatory bodies to approve iterative changes without having to begin the regulatory process for each change.

‍

Real-World Use Cases of AI and RWiD Driving Faster FDA Approvals

AI for Radiology Triage

An AI-based computer-aided detection tool is helping humans to automatically interpret radiological images and identify fractures and other anomalies. Using large-scale real-world imaging datasets for the training and validation of their AI, developers could show regulators compelling data for clinical utility that allowed for quicker FDA clearance.

‍

AI-Powered Imaging Companion

An AI platform assists radiologists by performing quantitative and qualitative analysis of clinical images, such as chest CTs. The solution’s use of diverse, real-world imaging repositories demonstrated reliability and generalizability, streamlining the FDA’s 510(k) clearance process and reducing time-to-approval.

‍

Wearable Device with ECG and Arrhythmia Monitoring

Wearable health technology incorporating ECG and atrial fibrillation history analysis used large real-world monitoring datasets to show effectiveness. This evidence accelerated regulatory review, enabling rapid authorization of digital health features that support continuous monitoring outside clinical settings.

‍

AI for Early Sepsis Risk Prediction

AI-based predictive software for sepsis risk assessment within 24 hours integrates imaging data with real-world hospital EHRs. By validating performance across diverse hospital datasets, the solution gained FDA marketing authorization faster than traditional clinical trial approaches.

‍

Experience the Segmed difference today

Segmed has instant access to large libraries of high-quality, de-identified medical imaging data from over 2,000 healthcare locations around the world. The regulatory-grade datasets, multimodal and longitudinal images, are appropriate for pre-market submissions and post-market surveillance.

Segmed's platform links imaging data with EHRs, claims, and patient outcomes via advanced tokenization, offering a comprehensive, "360-degree" view of device safety and performance in real-world clinical settings. This integration is critical for convincing regulators using robust RWiD evidence. AI companies use Segmed to accelerate model training, validation, and performance testing, enabling faster, more accurate FDA approvals. Segmed’s high-resolution RWiD aids in ongoing monitoring, earlier detection of adverse events, comprehensive therapy assessment, and supports regulatory submissions for label changes or extension via real-world evidence.

‍
‍Connect with us today to explore how our diverse, tokenized imaging datasets can enhance AI model development in the medical device industry.