Akademik Veri Yönetim Sistemi
Inorganic Chemistry Communications, cilt.182, 2025 (SCI-Expanded, Scopus)
In this study, machine learning (ML) was integrated with synthesis to develop and optimize novel oxovanadium(IV) thiosemicarbazone complexes with enhanced therapeutic potential. Four novel complexes (I–IV) were synthesized and their structures were determined using single-crystal X-ray diffraction, and spectroscopic techniques. Anticancer activity was evaluated in lung (A549, Calu1, H1650) and bone cancer (MG63, Saos2, SW1353) cell lines, with normal cells (Beas2B, HC). Antimicrobial efficacy was tested against nine bacterial strains including MRSA and VRE, plus C. albicans. Dose-response relationships and crystallographic feature importance were analyzed using four ML algorithms (ANN, Random Forest, SVM, XGBoost) with Leave-One-Out cross-validation. The complexes showed potent anticancer activity (GI50: 1.01–1.53 μg/mL), outperforming 5-fluorouracil, with lower toxicity in normal cells (GI50: 1.06–1.72 μg/mL). Compound I exhibited the widest therapeutic window and strongest inhibition of cancer cell migration. Moderate antibacterial activity against resistant strains (MIC: 32–64 μg/mL) was shown by complexes III and IV. An accuracy of R2 = 0.994 in predicting growth inhibition was achieved by ML analysis using artificial neural networks. Crystallographic parameters that predict biological activity were successfully identified through this ML-integrated approach, enabling rational design of oxovanadium(IV) complexes with selective anticancer activity and antimicrobial effects. Density Functional Theory (DFT/B3LYP) calculations with a 6-31G++(d,p) basis set were used to optimize molecular geometries and analyze electronic properties, including HOMO-LUMO energy levels. ADMET studies confirmed the drug-like potential of these complexes.