The introduction of artificial intelligence (AI) in the quality and regulatory functions is not a new concept, but the practical application of it is still in its early stages. Organizations are integrating generative, predictive, machine learning (ML), and other AI features into their practices to enhance quality and regulatory compliance. This blog post delves into the implementation and implications of applying generative AI to quality and regulatory.

What is Artificial Intelligence?

Simply put, artificial intelligence is an umbrella term encompassing a variety of machines capable of imitating human intelligence through cognitive function tasks.

Within the last four years, we have experienced a boom of AI models including chat forums, text-to-image, and text-to-video systems developed and marketed by large companies such as Google, Microsoft, OpenAI, and others. The ease and practicality of various types of AI is prevalent and appealing to numerous companies hoping to increase quality frameworks and regulatory compliance.

Application to Quality Control

Quality control has been a foundational feature of many operations since the emanation of manufacturing during the Industrial Revolution. With quality control comes monitoring and analyzing data. This can be a tedious and large task with considerable margin for human error, which may lead to corrective and preventative action, the addition of procedural steps, and additional laborious data analysis. Furthermore, a Quality Management System relies on effective information flow between the various departments of an organization. Because of the intricacies of the modern marketplace, effective communication between departments and systems has become more complicated and even more crucial. This is where AI comes into play.

Once a machine with AI capabilities has been fed the data and parameters, manual analysis and intervention become unnecessary. Through the creation of algorithms and workflows, AI can perform human cognitive functions seamlessly and with little error. The system can be capable of advanced problem solving, decision making, and analyzing ample amounts of data, generating nearly instantaneous results and responses. Quality Control is responsible for ensuring that products conform to specified requirements uniformly. Since AI has the ability to digest large volumes of information and compare numerous databases at once, it’s capable of increasing accuracy, scalability, and uniformity within the quality space, which in turn ensures product conformity and overall customer satisfaction.

Because of AI’s ability to effortlessly automatize complex quality control processes, its popularity in quality and regulatory settings is increasing rapidly. AI has the capability to increase speed and enhance efficiency, two desirable elements of organizations interested in quality control.

Application to Regulatory Compliance

Integrating AI into regulatory compliance systems creates similar benefits in both speed and efficiency. The key to applying AI to regulatory compliance is making the process systematic. Once pre-defined patterns and rules are introduced to a machine with the capability of computing algorithmic processes, collected results are swift and adequate, ensuring adherence to regulations and guidelines relevant to the business and its process.

Real World Application

It is easy enough to explain how AI works and its benefits to organizations, but is it actually applicable? In short, yes.

Roche, a biotech company and provider of in-vitro diagnostics (IVD), recently published on their website the use of machine learning artificial intelligence to discover new antibiotics. By harnessing the power of AI, Roche is digitizing drug discovery and development. Unlike traditional approaches, which start with known targets, AI will generate and analyze cellular and genetic data on a massive scale to yield wide-ranging maps of human biology. Barbara Lueckel, Head of Research Technologies, Roche Pharma Partnering, states the advancements being made at Roche leave her “optimistic that these technologies will significantly enhance [their] efforts to bring new medicines to patients as quickly and efficiently as possible.”

Quality assurance via AI is raising the bar for BMW Group quality standards. Through the implementation of optical quality control systems and visual and acoustic quality inspections with new AI sensor technology, BMW Group can analyze its production line in real-time to identify aberrations from standard quality. Peter Lehnert, Vice President New Technologies, Research and Innovations Digital Car states, “Artificial intelligence is an important part of the digital transformation at the BMW Group and thus one of the pioneers for inspiring and intelligent customer experiences.” Carsten Sapia, Vice President Strategy, Governance & IT Security states, “Responsible AI has a huge potential to create long-term benefits for all of our customers, employees, and other stakeholders. We are therefore committed to the advancement of AI driven by ethical principles that always put people first.”

AstraZeneca employs AI to enhance the precision of its quality control checks. By utilizing machine learning algorithms to assess drug images, they can detect imperfections that might be overlooked by human inspectors. This process guarantees that the drugs adhere to regulatory requirements. Moreover, the implementation of AI-driven tools has significantly reduced the time required for these inspections, allowing for a more efficient production workflow. The integration of AI technology within their design processes and environmental sustainability processes are also likely to improve customer experience and sustainability initiatives.

While AI is still at the forefront of research and is ever evolving, its early stages of development are being implemented in organizations’ systems today. The intent of AI is clear: quicker and easier. The ways in which this is applied to organizations’ systems have only just begun, so what can we expect?

Future Outlook

The possibilities for using AI are endless. We can expect to see advancements in data-based decision making, product discovery and development, data accuracy, and automation of routine tasks. This will free time for those responsible for more complex tasks and decision-making factors of their organization. This will also cause implementation of regulations from the FDA, which are already in the works.

The FDA has announced six new initiatives to address key aspects of AI/ML. The first initiative focuses on four projects the FDA will conduct to explore the use of simulations to supplement in silico data, or real medical patient datasets. The second initiative will utilize three programs aiming to analyze and minimize biases in AI training and testing methods to be used uniformly across various cohorts.

The third and fourth initiatives will jointly develop tools to choose appropriate metrics for evaluating AI-enable device performance, create methods to quantify uncertainty and develop statistical tools to design studies to continually evaluate the performance of evolving algorithms.

The fifth and sixth initiatives involve a total of 7 projects to improve regulatory assessments of new AI and ML-enabled devices with a focus on data harmonization and missing data. The programs will also develop tools for ongoing monitoring of the devices to address new types of data the devices did not encounter during the development process.

As AI revolutionizes quality and becomes a foundational block of organizations, we can expect additional industry changes.

Challenges and Considerations

The benefits of Artificial Intelligence and Machine Learning are accompanied by concerns that have been or should be considered when integrating AI into operations. Just like other applications, one must consider the security of AI platforms and the organization’s approach to customer privacy and cybersecurity. One must also consider the intended use of AI platforms, which is the highlight of current ethical topics associated with AI. The complexities of these processes in the development, deployment, use, and preservation of AI and ML technologies must be carefully managed prior to, and throughout the medical product life cycle.

The FDA reviews medical devices through 510k clearance, De Novo classification, or premarket approval (PMA). However, these approaches to ensuring the safety and efficacy of medical devices never had to consider the adaptation of AI and ML until now. While the FDA has proposed potential approaches to premarket reviews of AI, a regulatory framework to accommodate the adoption of AI and ML within the medical device space is still being developed. Some suggestions include developing a software precertification program and having algorithm change protocol (ACP) requirements.

Advances in AI in the medical device space must spawn new regulations and amendments or updates to current regulations. These changes are already well underway but continue to require ample consideration, advancements, and monitoring. Kent Walker, President of Global Affairs, Google & Alphabet, states, “AI is too important not to regulate, and too important not to regulate well.”

For more information on medical device that are already implementing AI/ML technologies, you can check out the FDA’s database for AI enabled medical device products: Artificial Intelligence and Machine Learning (AI/ML)-Enabled Medical Devices | FDA

Takeaway

Generative artificial intelligence is and will continue to reshape quality control and regulatory compliance in various industries.

Authors: Ashlee Bailey, Project Coordinator & Rachel Henderson, Project Coordinator