Critical Gaps

Medical device development is a complex and highly regulated process that can take 3–10 years and $31–94 M of investment to reach commercialization [Reference Espinoza, Shah, Nagendra, Bar-Cohen and Richmond2]. Regulatory preparedness plays a critical role in clinical trial activation as over half of early-stage medical device development activities are associated with regulatory processes [Reference Marešová, Klímová, Honegr, Kuča, Ibrahim and Selamat3]. Most universities have limited regulatory expertise, and university technology transfer offices are resource-limited in how far and to what degree they can assist innovators in keeping abreast of regulatory guidance and decision-making. Barriers repeatedly identified by academic investigators include 1) limited understanding of the regulatory requirements for design controls and market approval [Reference Rose, Everts, Heller, Burke, Hafer and Steele4]; 2) a lack of funding to cover the high cost of product design and development [Reference Linehan and Chaney5], and 3) the changing environment of information technology and security [Reference Williams and Woodward6].

The sheer number of relevant guidance documents, testing protocols, and submission checklists for translational researchers to follow present a significant barrier to catalyzing translational efficiency and commercialization. From the onset, researchers have to make critical decisions regarding their biomedical device(s), such as materials selection, benchtop testing, animal models, and clinical study protocols which, if incomplete or inadequately documented, can undermine an otherwise successful premarket application. The consequences of potential missteps are evident in the US Food and Drug Administration (FDA) Center for Devices and Radiological Health (CDRH) as one-third of applications are rejected before the review due to errors and refuse-to-accept (RTA) rates for applications are high. The 510(k) pathway, the most common pathway for medical technology, representing over 3000 applications in 2020, exhibited an average RTA rate of 32.43% [7]. Within certain CDRH Offices, the average RTA is as high as 46% with rejection rates over 60% [7]. Common reasons for an RTA include but are not limited to (1) inaccurate device description, (2) incorrect predicate identification for the determination of substantial equivalence, (3) incorrect indication for use or labeling, (4) lack of sterilization or reprocessing information, (5) lack of biocompatibility testing or incomplete materials identification, and (6) irrelevant or inadequate test data. Currently, resources within the academic research institutions are not positioned to address the scale and complexity of the rapidly evolving regulatory requirements. Studies on the general dissemination of regulatory and research information show that the standard approach to disseminating information, e.g., by publication, is ineffective. Information uptake does not occur spontaneously nor in a product-specific manner. When it does, it is often not standardized across diverse settings leading to ineffective implementation [Reference Brownson, Eyler, Harris, Moore and Tabak8]. Furthermore, there is a need to streamline the regulatory and FDA submission workflows to manage a widening range of medical technologies.

The FDA and the National Institutes of Health (NIH) have taken steps to help alleviate some of these challenges for medical device developers by creating or funding a number of programs at academic institutions (Table 1). A common feature of these programs is individualized support from advisory committees on medical, engineering, regulatory, and marketing processes. These committees comprise individuals with extensive experience in the MedTech industry and provide valuable feedback based on investigator progress reports [Reference Rose, Wills and Bourassa-Shaw9,Reference Wells, Rebuffoni and LeBien10]. However, these programs are difficult to scale due to their reliance on limited numbers of experts. Software-based solutions have been successfully deployed in many aspects of academic research infrastructure to streamline, standardize, and multiply research efforts. Applications like REDCap for electronic data capture, OnCore for Clinical trials management, and i2b2 for cohort discovery have been widely adopted by academic medical centers. Unfortunately, there are few software options that support the needs of medical device developers Tools like Greenlight Guru for quality management and Nyquist Data for predicate discovery were developed for industry and have limited applicability and uptake in an academic setting due to their pricing and the need for personalized “translation” of the regulatory terms and process beforehand.

Table 1. National and institutional programs with federal funding support that address medical device development

FDA, Food and Drug Administration; NIH, National Institutes of Health; NSF, National Science Foundation.

In response to the aforementioned challenges, we see a significant need to develop better artificial intelligence (AI)-driven software solutions that will enable academic medical centers to scale medical device management in a cost-effective, sustainable, and accessible way. AI-driven software can reduce reliance on regulatory experts for many early tasks, including product risk classification (Classes I–III); product code selection; identifying applicable federal regulations (e.g. special controls); suggesting predicate devices; recommending testing protocols; and surfacing relevant guidance documents. A regulatory support engine that uses optical character recognition can potentially extract relevant information using text mining and natural language processing from existing FDA, EUDAMED, clinical trials, regulatory guidance, 510(k) summaries, or PubMed databases, as well as from ISO-recognized standards.

However, software alone is rarely the solution. All technologies require the appropriate people and processes in place to generate value. For example, simply installing a REDCap instance will not solve an institution’s data collection difficulties. Project managers, support teams, training sessions, office hours, and more are necessary to meaningfully derive value from REDCap. Regulatory management software will be no different. Today, almost nothing is known about how best to deploy this software at academic medical centers. It will be critical to develop and evaluate programmatic solutions that address the implementation, evaluation, and maintenance of these tools.