Absorption Distribution Metabolism Excretion and Toxicity Property Prediction Utilizing a Pre-Trained Natural Language Processing Model and Its Applications in Early-Stage Drug Development.

Machine learning techniques are extensively employed in drug discovery, with a significant focus on developing QSAR models that interpret the structural information of potential drugs. In this study, the pre-trained natural language processing (NLP) model, ChemBERTa, was utilized in the drug discovery process. We proposed and evaluated four core model architectures as follows: deep neural network (DNN), encoder, concatenation (concat), and pipe. The DNN model processes physicochemical properties as input, while the encoder model leverages the simplified molecular input line entry system (SMILES) along with NLP techniques. The latter two models, concat and pipe, incorporate both SMILES and physicochemical properties, operating in parallel and with sequential manners, respectively. We collected 5238 entries from DrugBank, including their physicochemical properties and absorption, distribution, metabolism, excretion, and toxicity (ADMET) features. The models' performance was assessed by the area under the receiver operating characteristic curve (AUROC), with the DNN, encoder, concat, and pipe models achieved 62.4%, 76.0%, 74.9%, and 68.2%, respectively. In a separate test with 84 experimental microsomal stability datasets, the AUROC scores for external data were 78% for DNN, 44% for the encoder, and 50% for concat, indicating that the DNN model had superior predictive capabilities for new data. This suggests that models based on structural information may require further optimization or alternative tokenization strategies. The application of natural language processing techniques to pharmaceutical challenges has demonstrated promising results, highlighting the need for more extensive data to enhance model generalization.

Fine-tuning of BERT Model to Accurately Predict Drug-Target Interactions.

The identification of optimal drug candidates is very important in drug discovery. Researchers in biology and computational sciences have sought to use machine learning (ML) to efficiently predict drug-target interactions (DTIs). In recent years, according to the emerging usefulness of pretrained models in natural language process (NLPs), pretrained models are being developed for chemical compounds and target proteins. This study sought to improve DTI predictive models using a Bidirectional Encoder Representations from the Transformers (BERT)-pretrained model, ChemBERTa, for chemical compounds. Pretraining features the use of a simplified molecular-input line-entry system (SMILES). We also employ the pretrained ProBERT for target proteins (pretraining employed the amino acid sequences). The BIOSNAP, DAVIS, and BindingDB databases (DBs) were used (alone or together) for learning. The final model, taught by both ChemBERTa and ProtBert and the integrated DBs, afforded the best DTI predictive performance to date based on the receiver operating characteristic area under the curve (AUC) and precision-recall-AUC values compared with previous models. The performance of the final model was verified using a specific case study on 13 pairs of subtrates and the metabolic enzyme cytochrome P450 (CYP). The final model afforded excellent DTI prediction. As the real-world interactions between drugs and target proteins are expected to exhibit specific patterns, pretraining with ChemBERTa and ProtBert could teach such patterns. Learning the patterns of such interactions would enhance DTI accuracy if learning employs large, well-balanced datasets that cover all relationships between drugs and target proteins.

Dose Optimization of Vancomycin Using a Mechanism-based Exposure-Response Model in Pediatric Infectious Disease Patients.

PURPOSE: This study aimed to determine the appropriate vancomycin dosage, considering patient size and organ maturation, by simulating the bacterial count and biomarker level for drug administration in pediatric patients with gram-positive bacterial (GPB) infections. METHODS: Natural language processing for n-gram analysis was used to detect appropriate pharmacodynamic (PD) markers in infectious disease patients. In addition, a mechanism-based model was established to describe the systemic exposure and evaluate the PD marker simultaneously in pediatric patients. A simulation study was then conducted by using a mechanism-based model to evaluate the optimal dose of vancomycin in pediatric patients. FINDINGS: C-reactive protein (CRP) was selected as a PD marker from an analysis of ~270,000 abstracts in PubMed. In addition, clinical results, including the vancomycin plasma concentrations and CRP levels of pediatric patients (n = 93), were collected from electronic medical records. The vancomycin pharmacokinetic model with allometric scaling and a maturation function was built as a one-compartment model, with an additional compartment for bacteria. Both the effects of vancomycin plasma concentrations on the destruction of bacteria and those of bacteria on CRP production rates were represented by using a maximum achievable effect model (Emax model). Simulation for dose optimization was conducted not only by using the final model but also by exploring the possibility of therapeutic failure based on the MICs of vancomycin for GPB. Clinical cure was defined as when the CRP level fell below the upper limit of the normal range. Our dose optimization simulations suggested a vancomycin dosage of 10 mg/kg every 8 h as the optimal maintenance dose for pediatric patients with a postconceptual age <30 weeks and 10 mg/kg every 6 h for older children, aged up to 12 years. In addition, the MIC of 3 µg/mL was assessed as the upper concentration limit associated with successful vancomycin treatment of GPB infections. IMPLICATIONS: This study confirmed that the changes in bacterial counts and CRP levels were well described with mechanistic exposure-response modeling of vancomycin. This model can be used to determine optimal empiric doses of vancomycin and to improve therapeutic outcomes in pediatric patients with GPB.

Application of Size and Maturation Functions to Population Pharmacokinetic Modeling of Pediatric Patients.

Traditionally, dosage for pediatric patients has been optimized using simple weight-scaled methods, but these methods do not always meet the requirements of children. To overcome this discrepancy, population pharmacokinetic (PK) modeling of size and maturation functions has been proposed. The main objective of the present study was to evaluate a new modeling method for pediatric patients using clinical data from three different clinical studies. To develop the PK models, a nonlinear mixed effect modeling method was employed, and to explore PK differences in pediatric patients, size with allometric and maturation with Michaelis-Menten type functions were evaluated. Goodness of fit plots, visual predictive check and bootstrap were used for model evaluation. Single application of size scaling to PK parameters was statistically significant for the over one year old group. On the other hand, simultaneous use of size and maturation functions was statistically significant for infants younger than one year old. In conclusion, population PK modeling for pediatric patients was successfully performed using clinical data. Size and maturation functions were applied according to established criteria, and single use of size function was applicable for over one year ages, while size and maturation functions were more effective for PK analysis of neonates and infants.