Adel Alhowyan

Predicting drug solubility in supercritical CO₂ is challenging due to limited data, and strong inter-compound variability conditions that hinder generalization to unseen drugs. Although conventional machine learning models often achieve high accuracy, they rely on spurious correlations and lack interpretability, limiting practical use. This study proposes a unified framework combining symbolic regression with domain generalization techniques, including Invariant Risk Minimization (IRM) and Domain-Adversarial Neural Networks (DANN), to enable equation discovery. The framework enables the discovery of invariant, physically consistent equations that generalize beyond the training distribution. Symbolic regression extracts closed-form relationships, while IRM and DANN enforce invariance across domains within a multi-objective optimization setting. The optimized models achieved validation R ² values of 0.907, 0.984, and 0.965 for symbolic regression, DANN, and GPR, respectively. Although the highest predictive accuracy was obtained by DANN and GPR, symbolic regression remained competitive under domain-wise evaluation while uniquely providing explicit closed-form relationships between solubility and the governing physicochemical variables. The results therefore demonstrate that predictive performance, interpretability, equation complexity, and physical consistency should be considered jointly when developing transferable models for supercritical CO₂ solubility.

Accurate prediction of pharmaceutical solubility in supercritical CO₂ (SC-CO₂) systems is critical for green drug formulation and process intensification, yet existing machine learning studies largely prioritize predictive accuracy while overlooking mechanistic interpretability and causal understanding. This study proposes an integrated interpretable artificial intelligence framework for modeling drug solubility in SC-CO₂ by combining high-performance gradient boosting algorithms, explainable machine learning, Bayesian optimization, and causal inference. A curated dataset comprising 1,618 pharmaceutical compounds characterized by molecular weight, melting point, temperature, and pressure was employed. Model interpretability was investigated through SHAP analysis and partial dependence plots to quantify feature contributions and nonlinear response patterns. To move beyond correlation-based interpretation, a domain-constrained causal graph together with input-perturbation (what-if) sensitivity analysis was used to compare predictive importance against hypothesized direct causal contribution, within the limits of an observational, non-experimentally-validated causal structure. Explainability analysis identified pressure, and molecular weight, -specific properties as dominant predictive drivers, while causal analysis revealed melting point as a hidden but structurally influential determinant despite its lower statistical importance. Counterfactual simulations further exposed nonlinear and context-dependent solubility responses under ± 20% perturbations of key variables, highlighting asymmetric system sensitivities. Additionally, interaction analysis uncovered strong thermodynamic coupling effects, particularly between pressure and temperature, governing solubility dynamics.

Traditional QSAR toxicity models are, in general, assessed with random train-test splits where structural overlaps between train and test compounds are allowed. This usually results in an inflated predictive performance. Therefore, this paper looks at the problem of toxicity prediction under the structural distribution shift and checks if representation-level invariance can contribute to better cross-family generalization. A set of 1792 structurally diverse organic molecules for which toxicity data (e.g., Tetrahymena pyriformis pIGC₅₀) were determined experimentally was modeled with physicochemical descriptors. In order to depict the realistic scenarios of model use, the leave-one-cluster-out (LOCO) protocol was applied to enforce strict structural separation of training and test domains. Baseline neural models lost a lot of their prediction accuracy under LOCO versus random splits, thus exposing a very large generalization gap. On the other hand, invariant learning methods such as invariant risk minimization, contrastive alignment, and domain-adversarial training managed not only to reduce the cross-domain error but also to make residual distributions more stable. The embedding of latent space further demonstrated that invariance helps to get rid of cluster-specific signals while keeping toxicity-relevant gradients intact. From a practical perspective, these results suggest that robust in silico predictive toxicology, further under structural distribution shift, can be achieved through domain-aware validation and invariant representation learning.

We have examined the efficiency of drug delivery for targeted therapy by theoretical models. Machine learning strategy was tested to analyze the drug delivery of nanomedicines to the desired sites for efficient treatment. The inputs to the models are properties of nanoparticles, tumor model, cancer type, administration dose of drug, while the outputs are delivery efficiency of drug in various organs. The ML models employed include Elastic Net Regression (EN), K-Nearest Neighbors (KNN), Polynomial Regression (PR), and an ensemble method using Boosting to leverage the strengths of these individual models. The dataset includes both categorical and numerical variables related to nanoparticle properties and their distribution across organs like the tumor, spleen, heart, liver, lung, and kidney. Hyperparameter optimization was enhanced using the Adaptive Tree-Structured Parzen Estimator (ATPE), which improves upon traditional Bayesian optimization methods. The results demonstrated that the AdaBoost KNN model consistently outperformed other models in predicting nanoparticle biodistribution, with high R² values and low RMSE across multiple organ targets, thus proving effective in capturing complex input-output relationships. This approach offers a robust framework for advancing predictive modeling in biomedical applications involving nanoparticles.

This study employs density functional theory (DFT) and molecular dynamics (MD) simulations to investigate silicon carbide (SiC) nanocrystals as carriers for the anticancer drug Belzutifan. Among tested functional groups (-H, -OH, -NH₂, -COOH), carboxyl-functionalized SiC (SiC-COOH) exhibits superior drug loading capacity with an adsorption energy of -1.03 eV, representing a 25% improvement over conventional carbon-based carriers. The SiC-COOH system demonstrates exceptional stability with a formation energy of -5.42 eV/atom and a remarkably high dipole moment of 142.1 D, facilitating enhanced solubility. Electronic structure analysis reveals significant charge transfers (-0.25e) from Belzutifan to the nanocrystal, accompanied by a substantial reduction in the HOMO-LUMO gap from 5.427 eV (pristine SiC) to 3.41 eV (Belzutifan@SiC-COOH complex), indicating improved electronic coupling. MD simulations confirm the complex’s stability under physiological conditions, maintaining structural integrity with root-mean-square deviation (RMSD) values below 2.5 Å throughout the 5 ps simulation. Characteristic shifts in optical spectra (200–600 nm range) and IR vibrational modes (15–40 cm⁻¹) provide clear spectroscopic signatures of successful drug adsorption. The combination of strong binding (-1.03 eV adsorption energy), maintained biocompatibility, and tunable electronic properties positions functionalized SiC nanocrystals as a promising platform for targeted Belzutifan delivery, with potential applications in photo-triggered release systems.

Paracetamol (PCT) frequently contaminates natural water sources, posing potential risks to both human health and ecosystems. This study presents a computational investigation into the sensing capabilities of methylene-bridged [n]cycloparaphenylene ([n]MCPP, where n= 6, 8, and 10) nanorings for the detection of paracetamol using density functional theory (DFT) calculations. It was found that the stability of PCT@[n]MCPP complexes increases with the size of the [n]MCPP nanorings. The energy gap of the [n]MCPP nanorings is significantly influenced by the presence of the PCT drug, with the most substantial change (-44.79%) observed for the PCT@⁶MCPP complex. Additionally, non-covalent interaction (NCI) analysis reveals that van der Waals forces predominantly govern the interactions between paracetamol and the [n]MCPP nanorings. The calculated short recovery times and favorable sensor response factors of the PCT@[n]MCPP complexes at 298 K suggest that [n]MCPP nanorings can be used as promising materials for the detection of paracetamol. These findings underscore the potential of methylene-bridged [n]cycloparaphenylene nanorings as valuable candidates for identifying and eliminating PCT drug from the environment.

Lysosomes are essential intracellular organelles involved in plentiful cellular processes such as cell signaling, metabolism, growth, apoptosis, autophagy, protein processing, and maintaining cellular homeostasis. Their dysfunction is linked to various diseases, including lysosomal storage disorders, inflammation, cancer, cardiovascular diseases, neurodegenerative conditions, and aging. This review focuses on current and emerging therapies for lysosomal diseases (LDs), including small medicines, enzyme replacement therapy (ERT), gene therapy, transplantation, and lysosomal drug targeting (LDT). This study was conducted through databases like PubMed, Google Scholar, Science Direct, and other research engines. To treat LDs, medicines target the lysosomal membrane, acidification processes, cathepsins, calcium signaling, mTOR, and autophagy. Moreover, small-molecule therapies using chaperones, macro-therapies like ERT, gene therapy, and gene editing technologies are used as therapy for LDs. Additionally, endosymbiotic therapy, artificial lysosomes, and lysosomal transplantation are promising options for LD management. LDT enhances the therapeutic outcomes in LDs. Extracellular vesicles and mannose-6-phosphate-tagged nanocarriers display promising approaches for improving LDT. This study concluded that lysosomes play a crucial role in the pathophysiology of numerous diseases. Thus, restoring lysosomal function is essential for treating a wide range of conditions. Despite endosymbiotic therapy, artificial lysosomes, lysosomal transplantation, and LDT offering significant potential for LD control, there are ample challenges regarding safety and ethical implications.