May 8, 2024

expert reaction to study using AI to predict potential chemicals formed when heating flavoured vape liquids

A study published in Scientific Reports looks at AI and predicting chemicals formed in vapes. 

Prof Jacob George, Professor of Cardiovascular Medicine and Therapeutics, University of Dundee, said:

“There are close to 40,000 different flavours in the market worldwide today and making sense of their effects will require a combination of techniques including automated mapping algorithms and creation of neural networks such as this. While this study cannot give us definitive answers of the risks of flavoured vapes on human health, this study may be a helpful early step to identifying signals that could then lead to further, more in depth research into heat-induced breakdown of chemicals used in flavourings. This study has combined artificial intelligence with previously known published information to predict that when heating a combination of chemicals in flavoured vapes there might result in a harmful toxicant being produced, and these predictions can then be tested with further studies. There is, as yet, very little good-quality evidence of either safety or harm of these flavourings and so I welcome novel strategies as employed by these researchers.”

‘Forecasting vaping health risks through neural network model prediction of flavour pyrolysis reactions’ by Akihiro Kishimoto et al. was published in Scientific Reports at 10:00am UK Time on Wednesday 8 May 2024. 

DOI: 10.1038/s41598-024-59619-x

Declared interests

Prof Jacob George: I have no COI with regards to the contents of this study