Researchers at Shiraz University and Islamic Azad University reported on July 31, 2026, that TiO2/ZnO nanocomposites can degrade metronidazole in water and that machine learning models can predict the process. The study synthesized the TiO2/ZnO nanocomposite by the sol-gel method and characterized it with EDX, TEM, FTIR, SEM, and XRD. The target pollutant was metronidazole, an antibiotic whose inappropriate disposal in aquatic and soil environments can contribute to bacterial resistance and threaten humans and other organisms.

The team used response surface methodology with central composite design to test and optimize four variables: pH, irradiation time, metronidazole concentration, and catalyst dose. Under the most advantageous conditions, the synthesized nanocomposite degraded metronidazole by 94.92%. The researchers also examined recyclability, degradation mechanism, and the effect of light source intensity on the photocatalytic process.

Machine learning and deep learning models were trained to predict photocatalytic degradation efficiency, guided by response surface methodology. The evaluated models were support vector regression, artificial neural network, fully connected neural network, and random forest. Support vector regression produced the strongest reported results, with R = 0.9495, R2 = 0.8502, MSE = 7.2560, and RMSE = 2.6937.

Feature importance analysis found that reaction time and pH were the most influential parameters, followed by metronidazole concentration. Catalyst dose had minimal impact in the analysis. The authors said kernel-based support vector regression was effective for modeling complex photocatalytic systems with limited data, and they declared no competing interests. The paper was received on 06 May 2026, accepted on 27 July 2026, and published on 31 July 2026 in Scientific Reports.

Researchers at Shiraz University and Islamic Azad University reported on July 31, 2026, that TiO2/ZnO nanocomposites can degrade metronidazole in water and that machine learning models can predict the process. The study synthesized the TiO2/ZnO nanocomposite by the sol-gel method and characterized it with EDX, TEM, FTIR, SEM, and XRD. The target pollutant was metronidazole, an antibiotic whose inappropriate disposal in aquatic and soil environments can contribute to bacterial resistance and threaten humans and other organisms.

The team used response surface methodology with central composite design to test and optimize four variables: pH, irradiation time, metronidazole concentration, and catalyst dose. Under the most advantageous conditions, the synthesized nanocomposite degraded metronidazole by 94.92%. The researchers also examined recyclability, degradation mechanism, and the effect of light source intensity on the photocatalytic process.

Machine learning and deep learning models were trained to predict photocatalytic degradation efficiency, guided by response surface methodology. The evaluated models were support vector regression, artificial neural network, fully connected neural network, and random forest. Support vector regression produced the strongest reported results, with R = 0.9495, R2 = 0.8502, MSE = 7.2560, and RMSE = 2.6937.

Feature importance analysis found that reaction time and pH were the most influential parameters, followed by metronidazole concentration. Catalyst dose had minimal impact in the analysis. The authors said kernel-based support vector regression was effective for modeling complex photocatalytic systems with limited data, and they declared no competing interests. The paper was received on 06 May 2026, accepted on 27 July 2026, and published on 31 July 2026 in Scientific Reports.