From empirical exploration to data-driven innovation: The role of artificial intelligence in pharmaceutical taste masking

Abstract

The aversive taste of pharmaceutical formulations, especially bitterness, has been identified as a significant barrier to patient compliance. Taste masking technologies have emerged as significant approaches to improve the acceptability of orally administered drugs. These technologies have evolved from empirical methods to more systematic, precise, and intelligent solutions. Artificial intelligence (AI) advances have given rise to novel opportunities in taste prediction, excipient selection, and formulation optimization. This review critically examines the historical progression of taste masking technologies, elucidates the molecular and physiological basis of taste perception alongside interindividual variability, and compares conventional sensory and analytical techniques with emerging digital and biosensing platforms. Particular emphasis is placed on integrating AI into drug taste assessment, the development of masking strategies, and the control of manufacturing processes. Furthermore, the potential of AI-driven intelligent taste systems to enable precision masking, controlled release, and patient-specific palatability is discussed. This work outlines a conceptual and technological framework for next-generation taste optimization strategies, providing insights into developing smart, efficient, and personalized drug delivery systems.

Introduction

Taste is one of the five basic human senses, along with sight, touch, hearing, and smell [1], enabling the detection of chemical stimuli through specialized receptors in the oral cavity. It plays a critical role in food selection, nutrient intake, and drug acceptability [2]. Taste can be divided into five primary sensory experiences: sweet, sour, bitter, salty, and umami [3].

Taste directly impacts patient compliance in pharmacotherapy, as bitter-tasting pharmaceutical formulations may lead to medication refusal, compromising therapeutic efficacy. According to the World Health Organization (WHO) in 2022, approximately 30 % of pediatric patients refuse medications due to their bitter taste. More than 90 % of pediatricians report that taste and palatability are the biggest barriers to completing treatment [4]. Poor drug taste similarly reduces the acceptance of medications among the elderly and patients with swallowing difficulties, affecting therapeutic efficacy [5]. Thus, improving the palatability of drugs while ensuring their effectiveness is a key objective in developing new formulations.

The essence of taste masking technologies is to enhance the taste of drugs by modifying or masking their unpleasant tastes, making them more palatable to patients [6]. In recent years, with the advancement of precision medicine and personalized treatment approaches, optimizing the taste characteristics of medications for different patient groups has become a key strategy for enhancing drug compliance and improving the treatment experience. Taste masking technologies have evolved gradually from traditional empirical methods to more scientific, precise, and intelligent approaches.

Despite continuous advancements in taste masking strategies, from early use of sweeteners and coating techniques [7] to more sophisticated approaches like sweet cocrystal technology and 3D printing systems, the field still relies heavily on empirical methods. This experience-oriented approach results in a lack of uniform standards for taste identification and taste masking methods. Practical applications are susceptible to individual preferences, experimental conditions, and differences in evaluation subjects, resulting in poor reproducibility and consistency of results. More importantly, there is a fragmented structure across the stages of taste-related research, spanning taste identification, sensory evaluation, masking strategies, and outcome verification.

Consequently, artificial intelligence and other technological solutions are imperative to transition from empirical exploration to a more data-driven approach in taste masking technologies. With its powerful capabilities in data processing, pattern recognition, and optimization, artificial intelligence can be utilized for taste prediction, screening of taste masking substances, and process optimization. For instance, a graph neural network (GNN)-based model can accurately predict the bitterness of small-molecule compounds, with studies showing that this method can predict bitterness with over 90 % accuracy [8]. AI can also be used for the high-throughput screening of compounds with potential for taste masking [9], thereby improving the efficiency of formulation optimization and reducing experimental costs. Meanwhile, AI-driven dynamic process optimisation (e.g., digital twin systems) can monitor and optimise the production process in real time [10] by adjusting coating thickness [11] and controlling hot-melt extrusion parameters [12] to ensure the consisitency of taste masking. The incorporation of AI can potentially shift taste masking technologies from traditional empirical methods to modern formulation engineering and intelligent, data-driven evolution, thereby offering a new paradigm for personalized medicine and precise drug development.

Continue reading here

Chun Qiao, Jinru Hu, Yuxin Jin, Ruixiang Li, Fan Zhao, Yu Zhang, Xiaoyu Zhang, Ruofei Du, From empirical exploration to data-driven innovation: The role of artificial intelligence in pharmaceutical taste masking, European Journal of Pharmaceutics and Biopharmaceutics, Volume 218, 2026, 114934, ISSN 0939-6411, https://doi.org/10.1016/j.ejpb.2025.114934.

Read also our introduction article on Taste Masking here: