The world’s focus on the pharmaceutical industry’s ability to discover and develop new therapies has reached unprecedented levels, due to the COVID-19 pandemic. However, R&D is only the first half of the equation. Manufacturing will provide the critical follow-through necessary to deliver life-saving therapies that treat and prevent the novel coronavirus.
Never has it been more important to ensure that the pharmaceutical productivity chain is functioning as optimally as possible. Now, manufacturers have a proven and valuable new tool to achieve this goal. For the first time ever, artificial intelligence (AI) algorithms have been qualified, both for pharmaceutical as well as medical device manufacturing. This milestone is a welcomed advantage for a life sciences industry focused on the critical, pending need to accelerate safe and effective production of COVID-19 therapies, while also maintaining essential manufacturing operations of drugs and devices for other disease states.
Successful qualification of an AI algorithm for the highly regulated pharmaceutical manufacturing industry would be a milestone at any time, pandemic or not. Manufacturing decision making that is informed by qualified AI algorithms can potentially lower costs, increase profitability, and optimize manufacturing processes and the scaling of operations across production. But, until now, a standard procedure to ensure such algorithm governance did not exist.
Data Science and AI in Pharma
Uptake and utilization of AI by life science companies is gaining traction in research and clinical trial processes, but other arms of the industry have traditionally been slow adopters of this valuable resource. Recognizing the inherent value and potential benefits of AI, global regulatory bodies have recently started examining its use for pharmaceutical environment processes. Last year, the FDA published a discussion paper to request feedback in regard to AI and Machine Learning (ML) usage for medical devices, and the European Medicines Agency (EMA) created a special task force to gather opinions from main pharmaceutical stakeholders regarding big data and AI-related technologies and tools.
As specifically related to pharmaceutical manufacturing, applying data science as the foundation for decision making to leverage productivity and quality was considered as early as 2009. The International Conference on Harmonisation’s (ICH) second review of the Pharmaceutical Development guideline, known as Q8, addressed the issue. Coupled with ICH’s2005 Quality Risk Management guideline (Q9), industry thinking started to shift away from the traditional assumption of manufacturing process invariability. [i][ii]
By recognizing and factoringin the variations and deviations inherent in drug production, the door opened for consideration of novel new technologies, such as AI, to inform and optimize the design, implementation and control of pharmaceutical manufacturing. However, it has taken over a decade for the industry to begin thinking seriously about adopting and leveraging AI to reduce manufacturing risk and optimize outcomes.
Leveraging AI in the 2020s
AI algorithms are extremely valuable to the highly regulated life sciences and healthcare industries because they can be used to make critical decisions during drug and medical device manufacturing processes. But because AI algorithms can be used in this capacity, they must be qualified to ensure that they enable established manufacturing goals. This pressing and unresolved manufacturing challenge for GxP environments is now resolved.
The newly published AI Algorithm Qualification study, recently accepted by the PDA Journal of Pharmaceutical Sciences and Technology for publication, demonstrated conclusively that AI algorithms can be qualified for pharmaceutical product and medical device productivity chains. The study used a Design of Experiments (DoE) approach, data generated from custom-designed equipment, and a GxP-compliant AI software-as-a-service (SaaS) platform to agnostically qualify the Isolation Forest algorithm. The resulting qualification guidelines can be abstracted to other AI algorithms, enabling pharmaceutical and medical device manufacturers to leverage these valuable tools in relation to productivity and quality decisions.
Leveraging AI as a multivariate tool in the highly complex and stringently regulated pharmaceutical and medical device industries is not only good practice, it is almost a foregone necessity. The COVID-19 pandemic has exposed the grim need for accelerating the development and delivery of effective medications to ensure world health. However, the safety, quality and efficacy (SQE) of all pharmaceuticals, as well as medical devices, must always be ensured, regardless of the diseases they combat or the immediacy with which they are needed. Doing so requires that all tools employed in SQE processes have been appropriately vetted and proven.
The statistical methods based on multivariate analysis (MVA) that are used during drug manufacturing provide critical information for decision making. MVA is used, for instance, to optimize the dry granulation phase in tablet manufacturing or for the identification of critical factors in granules particle size distribution within a fluid bed dryer. Nowadays, advanced analytics, such as AI, must be considered as tools linked to any process to acquire knowledge.
When analytics are being used as an instrument in manufacturing operations, they must be held to standards correspondingly stringent to those by which equipment is evaluated. The validity of the elements, procedures and phases involved across the full drug manufacturing continuum must follow the same criteria, independent of their nature. The qualification of AI algorithms fulfils this requirement, enabling them to be applied by AI models as the foundation for decision making to improve pharmaceutical and medical device manufacturing productivity. It is a potential game-changer on a global scale.
The qualification of AI algorithms now enables the risk-aware industries of pharmaceutical and medical device manufacturing to confidently adopt and implement this valuable, recently emerged technology. Leveraging AI as a multivariate tool can help companies operating in these highly regulated industries to optimize production processes, as well as predict and fix manufacturing anomalies. As a result, drugs and devices that are safer, more efficient and effective, and of higher quality can be made accessible to patients and healthcare providers battling COVID-19 and other diseases.
[i] ICH, editor (2005). Quality Risk Management Q9, volume Step 4. ICH Harmonised Tripartite Guideline, International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use.
[ii] Herwig, C., Wölbeling, C., and Zimmer, T. (2017). A holistic approach to production control: From industry 4.0 to pharma 4.0.
