Peer-reviewed
publications.

Conductive polymers (CPs) are a class of functional polymers that can respond to a voltage stimulus. As a drug delivery system (DDS), the voltage applied to CPs can modulate drug release, offering unprecedented spatial and temporal control as we strive for precision and personalised medicines. However, this requires the drug to be charged in order to respond to a voltage stimulus, where unfortunately there are numerous drugs that are neutrally-charged. Herein, we propose an innovative solution that leverages charged polymeric particles as an intermediary vehicle. In this investigation, we electrospray paracetamol, a neutrally-charged drug, with poly(lactic-co-glycolic acid) (PLGA) to obtain drug-loaded microparticles with an overall negative charge of −53.01 ± 2.71 mV. The drug-loaded microparticles were then mixed with a CP and fabricated into films using a three dimensional (3D) printer. The films were subjected to voltages of 0, +1 and −1 V, where multivariate analysis of variance (MANOVA) revealed statistical significance in their drug release profiles. Applying −1 V (i.e., the same charge as the PLGA particle) resulted in a three-fold increase in drug release compared to 0 V (i.e., passive release), increasing from 9.82 ± 1.32 % to 29.03 ± 2.34 %. Furthermore, the platform was confirmed to respond and exhibit pulsatile release when switching the voltage “on-and-off”. Another benefit of leveraging PLGA discovered included sustained drug release, which is discussed. The findings suggest that charged polymeric particles can enable voltage-responsive release of neutrally-charged drugs from CPs, and signals a promising strategy for widening the clinical impact of CPs as DDSs.

Oral medications can be bioaccumulated or metabolised by gastrointestinal bacteria in a process collectively termed “drug depletion”. The precise biological mechanisms governing strain-specific depletion remain poorly understood, and systematic experimental classification of drug-strain interactions via in vitro studies is both costly and time-consuming. In this study, artificial intelligence (AI) methodologies combining machine learning (ML) and natural language processing (NLP) were applied to predict strain-specific drug depletion. The dataset comprised 16,802 drug-strain interaction pairs, with drugs represented by physicochemical descriptors and bacterial strains represented by whole-genome sequences. NLP techniques were used to transform genomic data into feature representations suitable for ML model training. The resulting models achieved strong predictive performance, with a balanced accuracy of 0.90 ± 0.02 and Matthews correlation coefficient of 0.54 ± 0.10. Feature importance analysis revealed that both drug properties and genomic features contributed to model predictions. Among the highest-ranking genomic features, BLASTX annotation identified several enzymes with known or plausible roles in drug metabolism. To further explore the mechanistic relevance of these features, two candidate enzymes were selected for molecular docking against drugs experimentally observed to be depleted. Glycosidase was found to possess binding energies of -8.69 and -7.88 kcal/mol for the two cardiac glycoside drugs digitoxin and digoxin, respectively; whereas acetyl-CoA carboxylase biotin carboxylase presented with binding energies for between -7.09 and -7.74 kcal/mol at one of its druggable sites. Collectively, these findings establish a proof-of-concept AI-driven framework that integrates predictive performance with mechanistic interpretability in the study of drug-microbiome interactions. The broader implications and limitations of applying AI in this context are also discussed. These preliminary findings offer a promising strategy for accelerating drug developments through using AI to rapidly highlight potential drug interactions.

Pharmaceutical 3D printing represents an advanced manufacturing technique enabling customized dosage forms. This research investigates large language models fine-tuned on over 1400 fused deposition modeling formulations to recommend suitable excipients based on active pharmaceutical ingredient doses and forecast filament mechanical properties. Llama2 was best suited for recommending excipients for FDM formulations.

Pharmaceutical 3D printing relies on digital workflows that introduce vulnerabilities like cyberattacks. Using denoising autoencoders across 1,623 formulations with 336 ingredients, researchers simulated data corruption (1%-50% deletion and noise). Results demonstrated R² scores from 0.989 at 1% corruption to 0.924 at 50%, outperforming traditional ML techniques and safeguarding formulation data integrity.

Medicines remain ineffective for over 50% of patients due to conventional mass production with fixed dosages. This study introduces a deep learning model to predict whether drug-loaded formulations can be manufactured using selective laser sintering, achieving 90% accuracy in printability prediction. The interpretable, uncertainty-optimized system represents the first in the field to accomplish this for pharmaceutical 3D printing.

This perspective examines how AI is transforming the scientific research process, from hypothesis generation to experimental design, with specific focus on pharmaceutical applications and the implications for drug discovery pipelines.

M3DISEEN, a web-based pharmaceutical software, accelerates FDM 3D printing using AI. From 614 drug-loaded formulations with 145 excipients, AI models predicted printability (76% accuracy) and filament characteristics (67% accuracy). Processing temperatures were predicted with mean absolute errors of 8.9°C (HME) and 8.3°C (FDM) by solely inputting pharmaceutical excipient trade names.

3D Printing could transform pharmaceutical manufacturing but requires eco-friendly approaches. Using Design of Experiments and ML across 136 runs, researchers identified that reducing build plate temperature and trial-and-error substantially decreases CO2 emissions. Only the ML pipeline could accurately predict emissions, demonstrating its value for sustainable Industry 5.0 manufacturing.

ML models using 170 formulations from 78 materials incorporated FT-IR, XRPD and DSC data. Formulation composition alone achieved F1 score of 81.9%, individual analytical techniques achieved 80.1-84.2%, while a consensus model combining all three methods achieved 88.9%. ML predictions benefit from multi-modal data combining numeric, spectral, thermogram and diffraction data.

We developed ML models that accurately predict nanoparticle size distribution from electrospray process parameters, enabling precise control over drug delivery particle characteristics without extensive trial-and-error experimentation.

The gut microbiome is critical to health, with dysbiosis promoting disease. Active ML predicted how excipients influence intestinal growth of Lactobacillus paracasei. Beginning with only six labeled interactions, the model predicted 111 additional excipient outcomes using uncertainty sampling, achieving 67.70% certainty. This inaugural application of active ML in microbiome science optimizes probiotic delivery.

Food-mediated changes to drug absorption can have significant implications for safety and efficacy. Analyzing 311 drugs with 20+ physicochemical properties, ML achieved 70-80% accuracy distinguishing food effects. Random forest performed best, with calculated dose number as the most important feature. This tool can screen compounds, reduce animal testing, and expedite oral drug development.

Electrohydrodynamic processes are promising healthcare fabrication technologies with FDA-approved products. EHD can rapidly manufacture nano-sized materials but requires significant resources. This review demonstrates how ML can enhance EHD workflows and includes an ML pipeline introduction to encourage adoption, with potential to accelerate research discoveries and enable workflow automation.

Orodispersible films offer significant clinical potential but traditional manufacturing limits personalization. Using direct ink writing to create ODFs under 100μm thick, ML classified films by active ingredient using NIR spectroscopy with 100% accuracy. Partial least squares verified dosages with R² exceeding 0.96 across paracetamol, caffeine, and theophylline formulations.

Over 150 drugs are susceptible to metabolism by gastrointestinal microorganisms. Using literature mining and unsupervised learning on 455 drug-microbiota interactions, researchers developed 11 supervised models. The tuned extremely randomised trees classifier achieved AUROC 75.1%, weighted recall 79.2%, balanced accuracy 69.0% on 91 validation drugs, enabling efficient drug screening for microbiota-mediated depletion.

Three-dimensional printing is transforming pharmaceutical development. Analyzing 968 formulations from 114 articles, ML models achieved up to 93% accuracy for hot melt extrusion processes. An artificial neural network predicted drug release times with mean error of approximately 24 minutes, demonstrating ML effectively models the 3D printing workflow for drug delivery systems.

Rheology is indispensable for formulation development. Using viscosity measurements, researchers developed models predicting printability for FDM 3D printed tablets. Testing polycaprolactone with ciprofloxacin and PEG at 130-170°C, they identified an operating viscosity window of 100-1000 Pa·s. ML models forecast dissolution with f2 similarity score of 90.9, enabling high-throughput formulation screening.