Revolutionizing Drug Formulation Development: The Increasing Impact of Machine Learning

Abstract

Over the past few years, the adoption of machine learning (ML) techniques has rapidly expanded across many fields of research including formulation science. At the same time, the use of lipid nanoparticles to enable the successful delivery of mRNA vaccines in the recent COVID-19 pandemic demonstrated the impact of formulation science. Yet, the design of advanced pharmaceutical formulations is non-trivial and primarily relies on costly and time-consuming wet-lab experimentation. In 2021, our group published a review article focused on the use of ML as a means to accelerate drug formulation development. Since then, the field has witnessed significant growth and progress, reflected by an increasing number of studies published in this area. This updated review summarizes the current state of ML directed drug formulation development, introduces advanced ML techniques that have been implemented in formulation design and shares the progress on making self-driving laboratories a reality. Furthermore, this review highlights several future applications of ML yet to be fully exploited to advance drug formulation research and development.

Introduction

In recent years, there has been significant interest in the application of machine learning (ML), which is a branch of artificial intelligence, to accelerate development in the pharmaceutical sciences [1], [2]. ML involves the development of algorithms to analyze data and identify statistical patterns or relationships. These relationships can be leveraged to make predictions about new data, derive insights for decision making, and discover underlying structures or characteristics in complex datasets. The process of learning from the data involves using optimization techniques to adjust the algorithm’s parameters and improve its accuracy. Although both ML and statistics center around data analysis, they differ in their primary objective. Statistics traditionally focuses on hypothesis testing, confidence intervals, and model interpretation, while ML emphasizes predictive modeling, optimization, and pattern recognition [3]. Thus, ML has a wide range of applications, including image and speech recognition, natural language processing, and predictive analytics [4], [5]. The broad applicability of these tools has given rise to a surge in the adoption of applied ML methods across many industrial sectors [6], [7], [8] including drug formulation design and development [1], [2].

Pharmaceutical formulation plays a critical role in the development of safe, effective, and stable medicines. Through the optimization of drug formulations, pharmaceutical scientists can confer safety and/or efficacy improvements to therapeutic agents. These improvements can be the difference between clinical success and failure. For instance, it has recently been estimated that up to 90% of new therapeutics fail in clinical trials (from 2010 to 2017) due to poor efficacy, unacceptable toxicity, and/or poor drug-like properties [9]. However, optimized drug formulations can improve efficacy, reduce toxicity, and improve druggability. There are many examples of pharmaceutical formulations that have been designed to overcome such challenges. For example, Vyxeos® is a liposomal formulation, encapsulating a synergistic drug combination of daunorubicin and cytarabine, that results in an improvement in clinical efficacy relative to the free drug combination [10].

Optimizing drug formulations is essential for the development of safe and effective medicines, as it can significantly impact clinical success. However, the design and development of advanced pharmaceutical products is a complex process that requires significant time, resources, and expertise. This complexity arises from numerous factors, including the need to consider various parameters related to the drug, excipients, and manufacturing conditions within a high-dimensional design space. As a result, experimental evaluation of all parameter combinations is prohibitive. ML has the potential to enable pharmaceutical scientists to map the relationship between the composition and performance of advanced drug formulations to enhance a priori formulation design. Ultimately, ML tools may help navigate high-dimensional design spaces in search of drug formulations with targeted properties in a time and resource efficient manner.

In 2021, our group published a review article that examined efforts to integrate ML into drug formulation development [1]. Specifically, the article introduced fundamental ML principles (e.g., different ML models and the concept of cross-validation) to readers by summarizing the applications of ML in pharmaceutical research dating back to the 1990s. A Web of Science^TM search shows the number of research articles that include drug formulation development and aspects of ML has continued to grow (Figure 1). Of course, in recent years the term “machine learning” has gained prominence in research terminology, replacing specific modeling techniques such as linear regression and principal component analysis. This shift reflects the growing recognition of the broader field of ML as a powerful tool for data analysis and prediction. For instance, linear regression and principal component analysis are now commonly thought of as introductory ML techniques for supervised and unsupervised learning, respectively. While linear regression and principal component analysis are fundamental ML techniques, it is important to note that ML encompasses a vast array of methods and algorithms beyond these introductory techniques. More advanced approaches, such as random forest, support vector machine, neural network, kernel ridge regression, and deep learning, offer more complex and sophisticated modeling capabilities, each suited for different types of data and problem domains. It is these more advanced approaches that are typically incorporated into more recent studies to design advanced drug delivery systems. As shown in Figure 1, there has been a surge in the use of ML, with 50% of all papers from the past two decades published over the past two years.

The current article builds on the previous article, with a summary of the latest studies that employ more advanced ML techniques to guide and accelerate development of a broader spectrum of drug formulations. In addition, this review outlines exciting future directions for the field. For a high-level introduction to ML methods and a stepwise summary of deploying ML pipelines in formulation development, we recommend the previous article [1]. For those with a more focused interest in a specific delivery strategy or dosage form, we recommend publications introducing ML in the context of solid dosage forms [11], hot-melt extrusion [12], nanomedicines [13], and 3D printing [14].