Organic crystal structure prediction via coupled generative adversarial networks and graph convolutional networks.

Organic crystal structures exert a profound impact on the physicochemical properties and biological effects of organic compounds. Quantum mechanics (QM)-based crystal structure predictions (CSPs) have somewhat alleviated the dilemma that experimental crystal structure investigations struggle to conduct complete polymorphism studies, but the high computing cost poses a challenge to its widespread application. The present study aims to construct DeepCSP, a feasible pure machine learning framework for minute-scale rapid organic CSP. Initially, based on 177,746 data entries from the Cambridge Crystal Structure Database, a generative adversarial network was built to conditionally generate trial crystal structures under selected feature constraints for the given molecule. Simultaneously, a graph convolutional attention network was used to predict the density of stable crystal structures for the input molecule. Subsequently, the distances between the predicted density and the definition-based calculated density would be considered to be the crystal structure screening and ranking basis, and finally, the density-based crystal structure ranking would be output. Two such distinct algorithms, performing the generation and ranking functionalities, respectively, collectively constitute the DeepCSP, which has demonstrated compelling performance in marketed drug validations, achieving an accuracy rate exceeding 80% and a hit rate surpassing 85%. Inspiringly, the computing speed of the pure machine learning methodology demonstrates the potential of artificial intelligence in advancing CSP research.

Predicting liposome formulations by the integrated machine learning and molecular modeling approaches.

Liposome is one of the most widely used carriers for drug delivery because of the great biocompatibility and biodegradability. Due to the complex formulation components and preparation process, formulation screening mostly relies on trial-and-error process with low efficiency. Here liposome formulation prediction models have been built by machine learning (ML) approaches. The important parameters of liposomes, including size, polydispersity index (PDI), zeta potential and encapsulation, are predicted individually by optimal ML algorithm, while the formulation features are also ranked to provide important guidance for formulation design. The analysis of key parameter reveals that drug molecules with logS [-3, -6], molecular complexity [500, 1000] and XLogP3 (≥2) are priority for preparing liposome with higher encapsulation. In addition, naproxen (NAP) and palmatine HCl (PAL) represented the insoluble and water-soluble molecules are prepared as liposome formulations to validate prediction ability. The consistency between predicted and experimental value verifies the satisfied accuracy of ML models. As the drug properties are critical for liposome particles, the molecular interactions and dynamics of NAP and PAL liposome are further investigated by coarse-grained molecular dynamics simulations. The modeling structure reveals that NAP molecules could distribute into lipid layer, while most PAL molecules aggregate in the inner aqueous phase of liposome. The completely different physical state of NAP and PAL confirms the importance of drug properties for liposome formulations. In summary, the general prediction models are built to predict liposome formulations, and the impacts of key factors are analyzed by combing ML with molecular modeling. The availability and rationality of these intelligent prediction systems have been proved in this study, which could be applied for liposome formulation development in the future.

How can machine learning and multiscale modeling benefit ocular drug development?

The eyes possess sophisticated physiological structures, diverse disease targets, limited drug delivery space, distinctive barriers, and complicated biomechanical processes, requiring a more in-depth understanding of the interactions between drug delivery systems and biological systems for ocular formulation development. However, the tiny size of the eyes makes sampling difficult and invasive studies costly and ethically constrained. Developing ocular formulations following conventional trial-and-error formulation and manufacturing process screening procedures is inefficient. Along with the popularity of computational pharmaceutics, non-invasive in silico modeling & simulation offer new opportunities for the paradigm shift of ocular formulation development. The current work first systematically reviews the theoretical underpinnings, advanced applications, and unique advantages of data-driven machine learning and multiscale simulation approaches represented by molecular simulation, mathematical modeling, and pharmacokinetic (PK)/pharmacodynamic (PD) modeling for ocular drug development. Following this, a new computer-driven framework for rational pharmaceutical formulation design is proposed, inspired by the potential of in silico explorations in understanding drug delivery details and facilitating drug formulation design. Lastly, to promote the paradigm shift, integrated in silico methodologies were highlighted, and discussions on data challenges, model practicality, personalized modeling, regulatory science, interdisciplinary collaboration, and talent training were conducted in detail with a view to achieving more efficient objective-oriented pharmaceutical formulation design.

Quantitative Analysis for Chinese and US-listed Pharmaceutical Companies by the LightGBM Algorithm.

AIM: This article aims to quantitatively analyze the growth trend of listed pharmaceutical companies in the US and China by a machine learning algorithm. BACKGROUND: In the last two decades, the global pharmaceutical industry has faced the dilemma of low research & development (R&D) success rate. The US is the world's largest pharmaceutical market, while China is the largest emerging market. OBJECTIVE: To collect data from the database and apply machine learning to build the model. METHODS: LightGBM algorithm was used to build the model and identify the factor important to the performance of pharmaceutical companies. RESULTS: The prediction accuracy for US companies was 80.3%, while it was 64.9% for Chinese companies. The feature importance shows that the net profit growth rate and debt liability ratio are significant in financial indicators. The results indicated that the US may continue to dominate the global pharmaceutical industry, while several Chinese pharmaceutical companies rose sharply after 2015 with the narrowing gap between the Chinese and US pharmaceutical industries. CONCLUSION: In summary, our research quantitatively analyzed the growth trend of listed pharmaceutical companies in the US and China by a machine learning algorithm, which provide a novel perspective for the global pharmaceutical industry. According to the R&D capability and profitability, 141 US-listed and 129 China-listed pharmaceutical companies were divided into four levels to evaluate the growth trend of pharmaceutical firms.

Machine learning in accelerating microsphere formulation development.

Microspheres have gained much attention from pharmaceutical and medical industry due to the excellent biodegradable and long controlled-release characteristics. However, the drug release behavior of microspheres is influenced by complicated formulation and manufacturing factors. The traditional formulation development of microspheres is intractable and inefficient by the experimentally trial-and-error methods. This research aims to build a prediction model to accelerate microspheres product development for small-molecule drugs by machine learning (ML) techniques. Two hundred eighty-six microsphere formulations with small-molecule drugs were collected from the publications and pharmaceutical company, including the dissolution temperature at both 37 ℃ and 45 ℃. After the comparison of fourteen ML approaches, the consensus model achieved accurate predictions for the validation set at 37 ℃ and 45 ℃ (R2 = 0.880 vs. R2 = 0.958), indicating the good performance to predict the in vitro drug release profiles at both 37 ℃ and 45 ℃. Meanwhile, the models revealed the feature importance of formulations, which offered meaningful insights to the microspheres development. Experiments of microsphere formulations further validated the accuracy of the consensus model. Furthermore, molecular dynamics (MD) simulation provided a microscopic view of the preparation process of microspheres. In conclusion, the prediction model of microsphere formulations for small-molecule drugs was successfully built with high accuracy, which is able to accelerate microspheres product development and promote the quality control of microspheres for the pharmaceutical industry.

Prediction of lipid nanoparticles for mRNA vaccines by the machine learning algorithm.

Lipid nanoparticle (LNP) is commonly used to deliver mRNA vaccines. Currently, LNP optimization primarily relies on screening ionizable lipids by traditional experiments which consumes intensive cost and time. Current study attempts to apply computational methods to accelerate the LNP development for mRNA vaccines. Firstly, 325 data samples of mRNA vaccine LNP formulations with IgG titer were collected. The machine learning algorithm, lightGBM, was used to build a prediction model with good performance (R 2 > 0.87). More importantly, the critical substructures of ionizable lipids in LNPs were identified by the algorithm, which well agreed with published results. The animal experimental results showed that LNP using DLin-MC3-DMA (MC3) as ionizable lipid with an N/P ratio at 6:1 induced higher efficiency in mice than LNP with SM-102, which was consistent with the model prediction. Molecular dynamic modeling further investigated the molecular mechanism of LNPs used in the experiment. The result showed that the lipid molecules aggregated to form LNPs, and mRNA molecules twined around the LNPs. In summary, the machine learning predictive model for LNP-based mRNA vaccines was first developed, validated by experiments, and further integrated with molecular modeling. The prediction model can be used for virtual screening of LNP formulations in the future.

In silico formulation prediction of drug/cyclodextrin/polymer ternary complexes by machine learning and molecular modeling techniques.

Ternary cyclodextrin (CD) complexes (drug/CD/polymer) can effectively improve the solubility of water-insoluble drugs with large size than binary CD formulations. However, ternary formulations are screened by a trial-and-error approach, which is laborious and material-wasting. Current research aims to develop a prediction model for ternary CD formulations by combined machine learning and molecular modeling. 596 ternary formulations data were collected to build a prediction model by machine learning. The random forest model achieved good performance with R2 = 0.887 in ST prediction and R2 = 0.815 in ST/SB prediction. Two ternary formulations (Hydrocortisone/ß-CD/HPMC and dovitinib/γ-CD/CMC) were used to validate the prediction model. Molecular modeling results showed that HPMC not only warped around hydrocortisone but also prevented CD molecules from self-aggregation to increase solubility. In conclusion, a prediction model for the ternary CD formulations was successfully developed, which will significantly accelerate the formulation screening process to benefit the formulation development of water-insoluble drugs.

Prediction of small-molecule compound solubility in organic solvents by machine learning algorithms.

Rapid solvent selection is of great significance in chemistry. However, solubility prediction remains a crucial challenge. This study aimed to develop machine learning models that can accurately predict compound solubility in organic solvents. A dataset containing 5081 experimental temperature and solubility data of compounds in organic solvents was extracted and standardized. Molecular fingerprints were selected to characterize structural features. lightGBM was compared with deep learning and traditional machine learning (PLS, Ridge regression, kNN, DT, ET, RF, SVM) to develop models for predicting solubility in organic solvents at different temperatures. Compared to other models, lightGBM exhibited significantly better overall generalization (logS ± 0.20). For unseen solutes, our model gave a prediction accuracy (logS ± 0.59) close to the expected noise level of experimental solubility data. lightGBM revealed the physicochemical relationship between solubility and structural features. Our method enables rapid solvent screening in chemistry and may be applied to solubility prediction in other solvents.

Computational pharmaceutics - A new paradigm of drug delivery.

In recent decades pharmaceutics and drug delivery have become increasingly critical in the pharmaceutical industry due to longer time, higher cost, and less productivity of new molecular entities (NMEs). However, current formulation development still relies on traditional trial-and-error experiments, which are time-consuming, costly, and unpredictable. With the exponential growth of computing capability and algorithms, in recent ten years, a new discipline named "computational pharmaceutics" integrates with big data, artificial intelligence, and multi-scale modeling techniques into pharmaceutics, which offered great potential to shift the paradigm of drug delivery. Computational pharmaceutics can provide multi-scale lenses to pharmaceutical scientists, revealing physical, chemical, mathematical, and data-driven details ranging across pre-formulation studies, formulation screening, in vivo prediction in the human body, and precision medicine in the clinic. The present paper provides a comprehensive and detailed review in all areas of computational pharmaceutics and "Pharma 4.0", including artificial intelligence and machine learning algorithms, molecular modeling, mathematical modeling, process simulation, and physiologically based pharmacokinetic (PBPK) modeling. We not only summarized the theories and progress of these technologies but also discussed the regulatory requirements, current challenges, and future perspectives in the area, such as talent training and a culture change in the future pharmaceutical industry.

An integrated computational methodology with data-driven machine learning, molecular modeling and PBPK modeling to accelerate solid dispersion formulation design.

Drugs in solid dispersion (SD) take advantage of fast and extended dissolution, thus attains a higher bioavailability than the crystal form. However, current development of SD relies on a random large-scale formulation screening method with low efficiency. Current research aims to integrate various computational tools, including machine learning (ML), molecular dynamic (MD) simulation and physiologically based pharmacokinetic (PBPK) modeling, to accelerate the development of SD formulations. Firstly, based on a dataset consisting of 674 dissolution profiles of SD, the random forest algorithm was used to construct a classification model to distinguish two types of dissolution profiles: "spring-and-parachute" and "maintain supersaturation", and a regression model to predict the time-dependent dissolution profiles. Both of the two prediction models showed good prediction performance. Moreover, feature importance was performed to help understand the key information that contributes to the model. After that, the vemurafenib (VEM) SD formulation in previous report was used as an example to validate the models. MD simulation was used to investigate the dissolution behavior of two SD formulations with two polymers (HPMCAS and Eudragit) at the molecular level. The results showed that the HPMCAS-based formulation resulted in faster dissolution than the Eudragit formulation, which agreed with the reported experimental results. Finally, a PBPK model was constructed to accurately predict the human pharmacokinetic profile of the VEM-HPMCAS SD formulation. In conclusion, combined computational tools have been developed to in silico predict formulation composition, in vitro release and in vivo absorption behavior of SD formulations. The integrated computational methodology will significantly facilitate pharmaceutical formulation development than the traditional trial-and-error approach in the laboratory.

Can machine learning predict drug nanocrystals?

Nanocrystals have exhibited great advantage for enhancing the dissolution rate of water insoluble drugs due to the reduced size to nanoscale. However, current pharmaceutical approaches for nanocrystals formulation development highly depend on the expert experience and trial-and-error attempts which remain time and resource consuming. In this research, we utilized machine learning techniques to predict the particle size and polydispersity index (PDI) of nanocrystals. Firstly, 910 nanocrystal size data and 341 PDI data by three preparation methods (ball wet milling (BWM) method, high-pressure homogenization (HPH) method and antisolvent precipitation (ASP) method) were collected for the construction of the prediction models. The results demonstrated that light gradient boosting machine (LightGBM) exhibited well performance for the nanocrystals size and PDI prediction with BWM and HPH methods, but relatively poor predictions for ASP method. The possible reasons for the poor prediction refer to low quality of data because of the poor reproducibility and instability of nanocrystals by ASP method, which also confirm that current commercialized products were mainly manufactured by BWM and HPH approaches. Notably, the contribution of the influence factors was ranked by the LightGBM, which demonstrated that milling time, cycle index and concentration of stabilizer are crucial factors for nanocrystals prepared by BWM, HPH and ASP, respectively. Furthermore, the model generalizations and prediction accuracies of LightGBM were confirmed experimentally by the newly prepared nanocrystals. In conclusion, the machine learning techniques can be successfully utilized for the predictions of nanocrystals prepared by BWM and HPH methods. Our research also reveals a new way for nanotechnology manufacture.

Predicting complexation performance between cyclodextrins and guest molecules by integrated machine learning and molecular modeling techniques.

Most pharmaceutical formulation developments are complex and ideal formulations are generally obtained after extensive experimentation. Machine learning is increasingly advancing many aspects in modern society and has achieved significant success in multiple subjects. Current research demonstrated that machine learning can be adopted to build up high-accurate predictive models in drugs/cyclodextrins (CDs) systems. Molecular descriptors of compounds and experimental conditions were employed as inputs, while complexation free energy as outputs. Results showed that the light gradient boosting machine provided significantly improved predictive performance over random forest and deep learning. The mean absolute error was 1.38 kJ/mol and squared correlation coefficient was 0.86. The evaluation of relative importance of molecular descriptors further demonstrated the key factors affecting molecular interactions in drugs/CD systems. In the specific ketoprofen-CD systems, machine learning model showed better predictive performance than molecular modeling calculation, while molecular simulation could provide structural, dynamic and energetic information. The integration of machine learning and molecular simulation could produce synergistic effect for interpreting and predicting pharmaceutical formulations. In conclusion, the developed predictive models were able to quickly and accurately predict the solubilizing capacity of CD systems. Current research has taken an important step toward the application of machine learning in pharmaceutical formulation design.

Predicting physical stability of solid dispersions by machine learning techniques.

Amorphous solid dispersion (SD) is an effective solubilization technique for water-insoluble drugs. However, physical stability issue of solid dispersions still heavily hindered the development of this technique. Traditional stability experiments need to be tested at least three to six months, which is time-consuming and unpredictable. In this research, a novel prediction model for physical stability of solid dispersion formulations was developed by machine learning techniques. 646 stability data points were collected and described by over 20 molecular descriptors. All data was classified into the training set (60%), validation set (20%), and testing set (20%) by the improved maximum dissimilarity algorithm (MD-FIS). Eight machine learning approaches were compared and random forest (RF) model achieved the best prediction accuracy (82.5%). Moreover, the RF models revealed the contribution of each input parameter, which provided us the theoretical guidance for solid dispersion formulations. Furthermore, the prediction model was confirmed by physical stability experiments of 17ß-estradiol (ED)-PVP solid dispersions and the molecular mechanism was investigated by molecular modeling technique. In conclusion, an intelligent model was developed for the prediction of physical stability of solid dispersions, which benefit the rational formulation design of this technique. The integrated experimental, theoretical, modeling and data-driven AI methodology is also able to be used for future formulation development of other dosage forms.

Deep learning for in vitro prediction of pharmaceutical formulations.

Current pharmaceutical formulation development still strongly relies on the traditional trial-and-error methods of pharmaceutical scientists. This approach is laborious, time-consuming and costly. Recently, deep learning has been widely applied in many challenging domains because of its important capability of automatic feature extraction. The aim of the present research is to apply deep learning methods to predict pharmaceutical formulations. In this paper, two types of dosage forms were chosen as model systems. Evaluation criteria suitable for pharmaceutics were applied to assess the performance of the models. Moreover, an automatic dataset selection algorithm was developed for selecting the representative data as validation and test datasets. Six machine learning methods were compared with deep learning. Results showed that the accuracies of both two deep neural networks were above 80% and higher than other machine learning models; the latter showed good prediction of pharmaceutical formulations. In summary, deep learning employing an automatic data splitting algorithm and the evaluation criteria suitable for pharmaceutical formulation data was developed for the prediction of pharmaceutical formulations for the first time. The cross-disciplinary integration of pharmaceutics and artificial intelligence may shift the paradigm of pharmaceutical research from experience-dependent studies to data-driven methodologies.

An Integrated Transfer Learning and Multitask Learning Approach for Pharmacokinetic Parameter Prediction.

BACKGROUND: Pharmacokinetic evaluation is one of the key processes in drug discovery and development. However, current absorption, distribution, metabolism, and excretion prediction models still have limited accuracy. AIM: This study aims to construct an integrated transfer learning and multitask learning approach for developing quantitative structure-activity relationship models to predict four human pharmacokinetic parameters. METHODS: A pharmacokinetic data set included 1104 U.S. FDA approved small molecule drugs. The data set included four human pharmacokinetic parameter subsets (oral bioavailability, plasma protein binding rate, apparent volume of distribution at steady-state, and elimination half-life). The pretrained model was trained on over 30 million bioactivity data entries. An integrated transfer learning and multitask learning approach was established to enhance the model generalization. RESULTS: The pharmacokinetic data set was split into three parts (60:20:20) for training, validation, and testing by the improved maximum dissimilarity algorithm with the representative initial set selection algorithm and the weighted distance function. The multitask learning techniques enhanced the model predictive ability. The integrated transfer learning and multitask learning model demonstrated the best accuracies, because deep neural networks have the general feature extraction ability; transfer learning and multitask learning improve the model generalization. CONCLUSIONS: The integrated transfer learning and multitask learning approach with the improved data set splitting algorithm was first introduced to predict the pharmacokinetic parameters. This method can be further employed in drug discovery and development.