Let’s take a step back. What is Artificial Intelligence, or AI as it has come to be known? If you ask Bing, Microsoft’s AI-powered search engine, it will come up with the following definition:

"AI is the intelligence exhibited by machines or software, distinct from human or animal intelligence. It involves creating systems that can mimic cognitive processes like reasoning and learning."

What is really exciting and concerning at the same time for AI is its ability to “mimic cognitive processes” and actually learn and improve over time. Machine learning (ML) is an application of AI, where computer models are trained on large sets of data to improve their performance over time, without any explicit human input.

Having a tool with infinite learning potential, with the ability to perform calculations in the range of quadrillions per second and not suffering from fatigue or errors, is an extremely valuable asset to any industry should, however, be harnessed with caution, and it is imperative we know its limitations. For instance, an artificial intelligence system is only as smart as the data it is fed; knowing the shortcomings of this data and understanding how an AI system processes it can help assess the robustness of AI-processing outcomes.

The question is not whether AI will revolutionize the world as we know it, because it already has – the real question is how and whether the net result will be positive or negative, and that depends on how we handle this great potential. In this article, we take a look at the impact AI already has on the pharmaceutical industry, how it could influence its future, and how the regulatory framework around it develops.

AI’s Potential

AI is already being used from the early stages of drug discovery to improve success rates, in clinical trials’ recruitment, and in finding new biomarkers. Biomarkers are biological indicators that can be used to diagnose diseases or monitor the progression of a disease. AI algorithms can analyze huge amounts of data from various sources, such as electronic health records, medical images, and genomic data, to identify new biomarkers.

Furthermore, researchers can use AI to analyze vast amounts of data and identify patterns that humans may not be able to detect. Based on this, new drug targets can be identified, and the efficacy of new drugs predicted, all in-silico. Many big pharma companies have established collaborations with AI-specializing companies and have started using ML to screen huge libraries of chemical compounds and connect their structural characteristics with their potential biological activity. By predicting those compounds’ pharmacokinetic behavior in terms of absorption, distribution, metabolism, and excretion (ADME), as well as their potential toxicity, they identify candidate molecules for further development. All this can lead to the development of more effective, safe medicines more quickly while minimizing studies on animals.

An important subset of the drug discovery process is the repurposing of existing compounds for new indications, an area where AI can significantly contribute by predicting efficacy against specific targets, possible side effects, and drug-drug interactions. Existing medicines already have established safety and pharmacokinetic properties, so using these compounds to treat new diseases can save significant development time and cost.

AI can also be used in the next step of development: clinical trials. It can screen patient data, identify the most suitable patients for a trial, and provide an indication of the likelihood of success.

AI and ML can additionally be used for the development and optimization of pharmaceutical manufacturing. For instance, AI systems can be fed data from trials during development, and based on that, manufacturing process parameters can be predicted. Quality by Design (QbD) has been used in the pharmaceutical industry for over 20 years. In QbD, carefully selected trial schemes during development establish a “design space,” within which material attributes and process parameters can be varied while still achieving the same Critical Quality Attributes (CQAs) for the finished dosage form (FDF).

Used to describe the measurement of critical quality and performance attributes of materials and processes during processing. The more times the manufacturing process runs and the results are evaluated against the targets set, the system can suggest or even make adjustments to optimize the process and finished dosage form characteristics.

Pharmaceuticals’ packaging is already partially automated, using machine-based visual inspection and quality control tests. These processes have recently improved even further with AI, which can identify defects early in the packaging process to reduce waste and increase efficiency. Additionally, AI can use sales and orders data to predict which products will be stored the longest and allocate storage slots accordingly, further improving operational efficiency.

How are regulators dealing with AI?

The US Food and Drug Administration (FDA) has recognized the growing use of AI/ML across the drug development life cycle and in multiple therapeutic areas. The FDA has observed a significant increase in the number of drug and biologics application submissions using AI/ML components over the past few years, with more than 100 submissions reported in 2021. In response, the agency has accelerated efforts to create a regulatory ecosystem that balances innovation with public health.

As part of this effort, the FDA’s Center for Drug Evaluation and Research (CDER), in collaboration with the Center for Biologics Evaluation and Research (CBER) and the Center for Devices and Radiological Health (CDRH), issued an initial discussion paper to engage stakeholders and explore considerations for AI/ML use in drug and biologics development. The FDA has also published a discussion paper on aspects of pharmaceutical manufacturing where AI/ML can be applied, highlighting regulatory considerations such as how adjustments made by AI/ML systems in the manufacturing process should be handled in terms of lifecycle management and variations.

The European Medicines Agency (EMA) has published a reflection paper on AI use in the lifecycle of medicines, emphasizing a human-centric approach for AI/ML development and deployment and addressing ethical considerations.

The World Health Organization (WHO) has also recognized AI’s potential in healthcare, publishing a report on its use. The report highlights AI’s ability to improve healthcare outcomes and reduce costs, while also noting challenges such as data privacy, security, and the need for ethical and regulatory frameworks.

What should we look out for?

As in many industries, the use of AI in the pharmaceutical industry also raises concerns. Given that the industry uses patient data and its output is a product aimed at improving health, the risks of misuse and abuse of AI are immense. Regulatory frameworks for AI/ML are only beginning to take shape, but it is important that, alongside discussions of possibilities, discussions on regulation have also started.

The International Coalition of Medicines Regulatory Authorities (ICMRA) has published a report identifying key issues linked to the regulation of future therapies using AI. For example, sponsors, developers, and pharmaceutical companies should establish enhanced governance structures to oversee algorithms and AI deployments that are closely linked to the benefit/risk profile of a medicinal product. This underscores the importance of carefully controlling the data fed into AI systems, understanding how the systems process it, and being able to evaluate the robustness of their outcomes, ensuring that AI decisions prioritize medicine safety and effectiveness.

The ICMRA report also highlights areas where guidelines need to be developed, implicitly identifying new fields to consider when using AI/ML systems, such as data provenance, reliability, transparency, and understandability. The “transparency and explainability” of AI algorithms is essential to ensure that decisions made by AI are understandable and auditable.

Artificial Intelligence Revolution in the Pharmaceutical Industry

AI has the potential to revolutionize the pharmaceutical industry by improving every stage, from drug discovery, clinical trial recruitment, and identifying new biomarkers to manufacturing process development and optimization. However, the use of AI also raises concerns about patient safety, privacy, and regulatory oversight.

While regulatory frameworks for AI in the pharmaceutical industry are still evolving, the identification of key focus areas is an important step. AI is a tool that can help us make significant advances in developing safe and effective medicines, but it must be managed carefully, with a clear understanding of its limitations at every step. Achieving this requires close collaboration between regulators and the industry.

Dr. Charikleia Souli (MSc., PhD)
New Products Development Senior Manager, Research & Development Department

Constantinos Kritiotis
R&D Manager