Issue link: https://www.e-digitaleditions.com/i/1509086
36 Pharmaceutical Technology ® Trends in Formulation 2023 eBook PharmTech.com CPHI BarCelona Coverage use real-world data for analysis, basically opening the door for all industries—pharma included—to benefit from its analytical capabilities. While global funding from venture investors has plunged compared to 2022, dropping nearly 50% to about US$21 billion in 2023 (3), AI remains one of the bright spots for venture capital firms, attracting nearly US$3 billion in investments. Models such as the open-source text-generating GPT and text-to-image have made it possible for ven- tures large and small to jump on the generative AI train, while open efforts have made available models that previously would have been sequestered by large commercial labs. For pharma, this will translate to faster ramp up times to gain competency. FDA is embracing AI FDA has been advocating for industry to pursue ad- vanced manufacturing solutions for decades. The es- tablishment of the Emerging Technology Team in 2014 provided a vehicle for industry to discuss technical and regulatory issues relating to different stages of drug de- velopment with the objective of defining a path forward (4). Under the umbrella of advanced manufacturing, FDA's Center for Drug Evaluation and Research (CDER) established the Framework for Regulatory Advanced Manufacturing Evaluation (FRAME) initiative to sup- port the adoption of advanced manufacturing tech- nologies that could bring benefits to patients (5). The technology priorities identified include end-to-end continuous manufacturing, distributed manufacturing (DM), point-of-care manufacturing, and AI. In May 2023, FDA issued two white papers to spur the conversation around AI and the subset of AI machine learning (ML) (6). The agency highlighted that focused adoption of AI is happening with more than 100 drug and biologic ap- plications in 2021 containing components of AI as part of their submission. However, to spur greater adoption, the papers bring to light the risks associated with using AI, including model biases used to train ML algorithms, and emphasize the need to address inaccuracies and completeness of the underlying model data. In addi- tion, the papers outline the role of performance moni- toring over the models to ensure they remain reliable, relevant, and consistent over time. Potential applications for AI in drug development CDER identif ied the following four areas in their white paper (6,7) to industry where AI could have an immediate impact: • Process design and scale-up: AI models such as machine learning—generated using process development data—could be leveraged to more quickly identify optimal processing parameters or scale-up processes, reducing development time and waste. • Advanced process control (APC): APC allows dy- namic control of the manufacturing process to achieve a desired output. AI methods can also be used to develop process controls that can predict the progression of a process by using AI in com- bination with real-time sensor data. • Process monitoring and fault detection: AI methods can be used to monitor equipment and detect changes from normal performance that trigger maintenance activities, reducing process downtime. • Trend monitoring: AI can be used to examine consumer complaints and deviation reports con- taining large volumes of text to identify cluster problem areas and prioritize areas for contin- ual improvement. This offers the advantage of identifying trends in manufacturing-related de- viations to support a more comprehensive root cause identification. For most pharma and biologic innovators and man- ufacturers, the benefits of applying AI are still elusive. AI in pharma and biotech While industr y has historically been cautious in adopting technology, there is targeted integration of AI. Figure 1 captures primary areas where the phar- maceutical industry is exploring how AI can drive business performance: Drug discovery and development. Few enhance- ments have had as profound impact on the cost of drug development as improving the clinical success rate of molecules in development. Today nine out of 10 mole- cules never make it to market with nearly 58% failing after a successful Phase II clinical study, representing the majority of the cost associated with bringing a drug to market. There are multiple areas where we have seen AI provide value in reducing program risk, reducing drug development cost and time to market. Drug screening. Finding a novel molecule with the right balance of on-target affinity and desired physico- chemical properties while considering key factors such as toxicity and bioactivity is the primary challenge during drug screening. AI allows a drug sponsor to in- crease the number of rationally designed compounds assessed, improving the chances for identifying a prom- ising molecule. The field of digital chemistry utilizes ML and in-silico modeling to screen billions of molecular structures and fragments considering the pharmaco- kinetic criteria relating to Adsorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) criteria typically derived from Phase I studies. Despite its advantages, AI faces significant data challenges relating to the scale, growth, diversity, and uncertainty of the data. The data sets available for drug development in pharmaceutical companies can in- volve millions of compounds, and traditional ML tools