One of the issues in advanced oxidation process (AOP) implementation in drinking water treatment is the presence of natural organic matter (NOM) in the water resources. Although most of the NOM is removed during the pre-sedimentation and coagulation-flocculation process in conventional drinking water treatment, some of the NOM can still be present in the later stage of treatment where AOP is implemented. Therefore, the assessment of NOM transformation in AOP treatment is needed to clarify its potential as a toxic substance after the treatment. In this study, the persistent NOM compounds in the treatment were analyzed as the first step of NOM assessment in AOP. The observed NOM was a Suwanee River NOM treated under UV radiation in a D<sub>2</sub>O solvent. The NOM compositions before and after photolysis treatment were obtained using FT-ICR mass spectrometry as the dataset for this study. A binary classification task was assigned to the XGBoost model in R to predict the persistent compounds. The optimized model showed the area under the curve is 0.93 with sensitivity and specificity of 0.94 and 0.78 respectively, indicating an accurate performance of the developed model. The variable importance plot of the model revealed that molecule size is the most important descriptor in the prediction followed by the oxygen-carbon ratio and double band equivalent of oxygen.

Predicting the persistent natural organic matter from photolysis treatment using machine learning

Where thousands of chemistry professionals meet to share ideas and advance scientific and technical knowledge. Spring 2023 abstracts, posters, and presentations will launch at the end of March.

ACS Spring 2023

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.