Understanding Vaping Health Risks Through Neural Networks
In a recent study published in Scientific Reports, researchers have utilized a graph-convolutional neural network (NN) to predict and analyze the thermal decomposition products of e-liquid flavors to assess potential health risks. The rise in e-liquid consumption, particularly among the youth, has raised concerns about the long-term health impacts associated with the inhalation of chemically complex e-liquids.
Evaluating E-Liquid Flavor Chemicals
The research delves into 180 flavor chemicals commonly found in e-liquids, revealing a diverse array of functional groups that suggest varied pyrolysis reactions. The analysis considers molecular weight, polarity, and chemical properties, highlighting a moderate diversity within the chemical space for vaping flavors.
Workflow for E-Liquid Flavor Risk Assessment
A workflow was implemented to assess the risk of 180 e-liquid flavors, integrating NN predictions of pyrolysis reactions with experimental mass spectrometry (MS) data. The automated process correlated NN-predicted products with MS fragments to classify health risks using the Globally Harmonized System (GHS) and estimate reaction activation energies.
Graph-Convolutional NN Model for Pyrolysis Products Prediction
The research employed the Weisfeiler–Lehman neural network (W–L NN) model to predict pyrolysis products without specific training data, emphasizing the significance of bond changes in heat-driven pyrolysis reactions. The model’s implementation involved converting chemical structures into graph representations to predict bond-breaking changes.
Correlation with Experimental EI-MS Data
Experimental electron-impact mass spectrometry (EI-MS) data confirmed the accuracy of NN predictions, indicating a robust correlation between predicted pyrolysis products and actual fragmentation patterns seen in MS data.
Data Amalgamation and Health Risk Analysis
Combining NN predictions and EI-MS data identified a multitude of matched products across the e-liquid flavors analyzed, showcasing various health hazards such as acute toxins, health hazards, and irritants.
Prediction of Pyrolysis Activation Energies
A directed message-passing neural network (D-MPNN) was utilized to estimate activation energies for pyrolysis reactions, shedding light on potential health hazards under typical vaping conditions.
Conclusion
The study provides valuable insights into the complex chemistry of vaping products, offering a basis for future research and regulatory assessments. By utilizing advanced NN models and experimental data analysis, researchers can better understand and mitigate the potential health risks associated with vaping e-liquids.
IntelliPrompt curated this article: Read the full story at the original source by clicking here