Application of machine learning to a material library for modeling of relationships between material properties and tablet properties

Ten Years of the Manufacturing Classification System: A review of literature applications and an extension of the framework to continuous manufacture

The MCS initiative was first introduced in 2013. Since then, two MCS papers have been published: the first proposing a structured approach to consider the impact of drug substance physical properties on manufacturability and the second outlining real world examples of MCS principles. By 2023, both publications had been extensively cited by over 240 publications. This article firstly reviews this citing work and consider how the MCS concepts have been received and are being applied. Secondly, we will extend the MCS framework to continuous manufacture.The review structure follows the flow of drug product development focussing first on optimisation of API properties. The exploitation of links between API particle properties and manufacturability using large datasets seems particularly promising. Subsequently, applications of the MCS for formulation design include a detailed look at the impact of percolation threshold, the role of excipients and how other classification systems can be of assistance. The final review section focusses on manufacturing process development, covering the impact of strain rate sensitivity and modelling applications.The second part of the paper focuses on continuous processing proposing a parallel MCS framework alongside the existing batch manufacturing guidance. Specifically, we propose that continuous direct compression can accommodate a wider range of API properties compared to its batch equivalent.

Sample Size Requirements of a Pharmaceutical Material Library: A Case in Predicting Direct Compression Tablet Tensile Strength by Latent Variable Modeling

The material library is an emerging, new data-driven approach for developing pharmaceutical process models. How many materials or samples should be involved in a particular application scenario is unclear, and the impact of sample size on process modeling is worth discussing. In this work, the direct compression process was taken as the research object, and the effects of different sample sizes of material libraries on partial least squares (PLS) modeling in the prediction of tablet tensile strength were investigated. A primary material library comprising 45 materials was built. Then, material subsets containing 5 × i (i = 1, 2, 3, …, 8) materials were sampled from the primary material library. Each subset underwent sampling 1000 times to analyze variations in model fitting performance. Both hierarchical sampling and random sampling were employed and compared, with hierarchical sampling implemented with the help of the tabletability classification index d. For each subset, modeling data were organized, incorporating 18 physical properties and tableting pressure as the independent variables and tablet tensile strength as the dependent variable. A series of chemometric indicators was used to assess model performance and find important materials for model training. It was found that the minimum R2 and RMSE values reached their maximum, and the corresponding values were kept almost unchanged when the sample sizes varied from 20 to 45. When the sample size was smaller than 15, the hierarchical sampling method was more reliable in avoiding low-quality few-shot PLS models than the random sampling method. Two important materials were identified as useful for building an initial material library. Overall, this work demonstrated that as the number of materials increased, the model's reliability improved. It also highlighted the potential for effective few-shot modeling on a small material library by controlling its information richness.

A tabletability change classification system in supporting the tablet formulation design via the roll compaction and dry granulation process

In this paper, the material library approach was used to uncover the pattern of tabletability change and related risk for tablet formulation design under the roll compaction and dry granulation (RCDG) process. 31 materials were fully characterized using 18 physical parameters and 9 compression behavior classification system (CBCS) parameters. Then, each material was dry granulated and sieved into small granules (125-250 μm) and large granules (630-850 μm), respectively. The compression behavior of granules was characterized by the CBCS descriptors, and were compared with that of ungranulated powders. The relative change of tabletability (CoTr) index was used to establish the tabletability change classification system (TCCS), and all materials were classified into three types, i.e. loss of tabletability (LoT, Type I), unchanged tabletability (Type II) and increase of tabletability (Type III). Results showed that approximately 65% of materials presented LoT, and as the granules size increased, 84% of the materials exhibited LoT. A risk decision tree was innovatively proposed by joint application of the CBCS tabletability categories and the TCCS tabletability change types. It was found that the LoT posed little risk to the tensile strength of the final tablet, when Category 1 or 2A materials, or Category 2B materials with Type II or Type III change of tabletability were used. Formulation risk happened to Category 2C or 3 materials, or Category 2B materials with Type I change of tabletability, particularly when high proportions of these materials were involved in tablet formulation. In addition, the risk assessment results were verified in the material property design space developed from a latent variable model in prediction of tablet tensile strength. Overall, results suggested that a combinational use of CBCS and TCCS could aid the decision making in selecting materials for tablet formulation design via RCDG.

Advancing algorithmic drug product development: Recommendations for machine learning approaches in drug formulation

Artificial intelligence is a rapidly expanding area of research, with the disruptive potential to transform traditional approaches in the pharmaceutical industry, from drug discovery and development to clinical practice. Machine learning, a subfield of artificial intelligence, has fundamentally transformed in silico modelling and has the capacity to streamline clinical translation. This paper reviews data-driven modelling methodologies with a focus on drug formulation development. Despite recent advances, there is limited modelling guidance specific to drug product development and a trend towards suboptimal modelling practices, resulting in models that may not give reliable predictions in practice. There is an overwhelming focus on benchtop experimental outcomes obtained for a specific modelling aim, leaving the capabilities of data scraping or the use of combined modelling approaches yet to be fully explored. Moreover, the preference for high accuracy can lead to a reliance on black box methods over interpretable models. This further limits the widespread adoption of machine learning as black boxes yield models that cannot be easily understood for the purposes of enhancing product performance. In this review, recommendations for conducting machine learning research for drug product development to ensure trustworthiness, transparency, and reliability of the models produced are presented. Finally, possible future directions on how research in this area might develop are discussed to aim for models that provide useful and robust guidance to formulators.

Leveraging a multivariate approach towards enhanced development of direct compression extended release tablets

Extended release formulations play a crucial role in the pharmaceutical industry by maintaining steady plasma levels, reducing side effects, and improving therapeutic efficiency and compliance. One commonly used method to develop extended release formulations is direct compression, which offers several advantages, such as simplicity, time savings, and cost-effectiveness. However, successful direct compression-based extended release formulations require careful assessment and an understanding of the excipients' attributes. The scope of this work is the characterization of the compaction behavior of some matrix-forming agents and diluents for the development of extended release tablets. Fifteen excipients commonly used in extended release formulations were evaluated for physical, compaction and tablet properties. Powder properties (e.g., particle size, flow properties, bulk density) were evaluated and linked to the tablet's mechanical properties in a fully integrated approach, and data were analyzed by constructing a principal component analysis (PCA). Significant variability was observed among the various excipients. The present work successfully demonstrates the applicability of PCA as an effective tool for comparative analysis, pattern and clustering recognition and correlations between excipients and their properties, facilitating the development and manufacturing of direct compressible extended release formulations.

Application of ensemble machine learning approach to assess the factors affecting size and polydispersity index of liposomal nanoparticles

Liposome nanoparticles have emerged as promising drug delivery systems due to their unique properties. Assessing particle size and polydispersity index (PDI) is critical for evaluating the quality of these liposomal nanoparticles. However, optimizing these parameters in a laboratory setting is both costly and time-consuming. This study aimed to apply a machine learning technique to assess the impact of specific factors, including sonication time, extrusion temperature, and compositions, on the size and PDI of liposomal nanoparticles. Liposomal solutions were prepared and subjected to sonication with varying values for these parameters. Two compositions: (A) HSPC:DPPG:Chol:DSPE-mPEG2000 at 55:5:35:5 molar ratio and (B) HSPC:Chol:DSPE-mPEG2000 at 55:40:5 molar ratio, were made using remote loading method. Ensemble learning (EL), a machine learning technique, was employed using the Least-squares boosting (LSBoost) algorithm to accurately model the data. The dataset was randomly split into training and testing sets, with 70% allocated for training. The LSBoost algorithm achieved mean absolute errors of 1.652 and 0.0105 for modeling the size and PDI, respectively. Under conditions where the temperature was set at approximately 60 °C, our EL model predicted a minimum particle size of 116.53 nm for composition (A) with a sonication time of approximately 30 min. Similarly, for composition (B), the model predicted a minimum particle size of 129.97 nm with sonication times of approximately 30 or 55 min. In most instances, a PDI of less than 0.2 was achieved. These results highlight the significant impact of optimizing independent factors on the characteristics of liposomal nanoparticles and demonstrate the potential of EL as a decision support system for identifying the best liposomal formulation. We recommend further studies to explore the effects of other independent factors, such as lipid composition and surfactants, on liposomal nanoparticle characteristics.

Using a Material Library to Understand the Change of Tabletability by High Shear Wet Granulation

Understanding the tabletability change of materials after granulation is critical for the formulation and process design in tablet development. In this paper, a material library consisting of 30 pharmaceutical materials was used to summarize the pattern of change of tabletability during high shear wet granulation and tableting (HSWGT). Each powdered material and the corresponding granules were characterized by 19 physical properties and nine compression behavior classification system (CBCS) parameters. Principal component analysis (PCA) was used to compare the physical properties and compression behaviors of ungranulated powders and granules. A new index, namely the relative change of tabletability (CoT_r), was proposed to quantify the tabletability change, and its advantages over the reworking potential were demonstrated. On the basis of CoT_r values, the tabletability change classification system (TCCS) was established. It was found that approximately 40% of materials in the material library presented a loss of tabletability (i.e., Type I), 50% of materials had nearly unchanged tabletability (i.e., Type II), and 10% of materials suffered from increased tabletability (i.e., Type III). With the help of tensile strength (TS) vs. compression pressure curves implemented on both powders and granules, a data fusion method and the PLS2 algorithm were further applied to identify the differences in material properties requirements for direct compression (DC) and HSWGT. Results indicated that increasing the plasticity or porosity of the starting materials was beneficial to acquiring high TS of tablets made by HSWGT. In conclusion, the presented TCCS provided a means for the initial risk assessment of materials in tablet formulation design and the data modeling method helped to predict the impact of formulation ingredients on the strength of compacts.

Machine learning guided microwave-assisted quantum dot synthesis and an indication of residual H2O2 in human teeth

The current preparation methods of carbon quantum dots (CDs) involve many reaction parameters, which leads to many possibilities in the synthesis processes and high uncertainty of the resultant production performance. Recently, machine learning (ML) methods have shown great potential in correlating the selected features in many applications, which can help understand the relevant structure-function relationships of CDs and discover better synthesis recipes as well. In this work, we employ the ML approach to guide the blue CD synthesis in microwave systems. After optimizing the synthesis parameters and conditions, the quantum yield (QY) increases to about 200% higher than the average value of the prepared samples without ML guidance. The obtained CDs are applied as fluorescent probes to monitor hydrogen peroxide (H₂O₂) in human teeth. The CD probe exhibits a linear relationship with the concentration of H₂O₂ ranging from 0 to 1.1 M with a lower detection limit of 0.12 M, which can effectively detect the residual H₂O₂ after bleaching teeth. This work shows that the adopted ML methods have considerable advantages in guiding the synthesis of high-quality CDs, which could accelerate the development of other novel functional materials in energy, biomedical, and environmental remediation applications.

A Multivariate Formulation and Process Development Platform for Direct Compression

The efficient development of robust tableting processes is challenging due to the lack of mechanistic understanding on the impact of raw material properties and process parameters on tablet quality. The experimental determination of the effect of process and formulation parameters on tablet properties and subsequent optimization is labor-intensive, expensive and time-consuming. The combined use of an extensive raw material property database, process simulation tools and multivariate modeling allows more efficient and more optimized development of the direct compression (DC) process. In this study, key material attributes and in-process mechanical properties with a potential effect on tablet processability and tablet properties were identified. In a first step, an extensive characterization of 55 raw materials (over 100 material descriptors) (Van Snick et al., 2018) and 26 formulation blends (31 material descriptors) (Dhondt et al., 2022) was performed. These blends were subsequently compacted on a compaction simulator under multiple process conditions through a design of experiments (DoE) approach. A T-shaped partial least squares (T-PLS) model was established which correlates tablet quality attributes with process settings, raw material properties and blend ratios. During future development of the DC formulation and process for a new active pharmaceutical ingredient (API), this model can then be used to provide a preliminary formulation and compaction process settings as starting point to be further optimized during development trials based on well-defined raw material characteristics and compaction tests. This study hence contributes to a better understanding on the impact of raw material properties and process settings on a DC process and final properties of the produced tablets; and provides a platform allowing a more efficient and more optimized development of a robust tableting process.