Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Handling and storing chemicals in the industrial world require a conscientious hazards assessment and the implementation of appropriate measures to ensure safety for the workers, the company, the environment and the society. Reliable evaluations and predictions based solely on the molecular structure would represent a valuable tool in the preliminary hazard assessment process as it would reduce the necessary time and resources for extensive testing. This work presents how Quantitative Structure-Property Relationships (QSPR) of various hazardous chemical properties (i.e. thermal decomposition enthalpy or Minimal Ignition Energy) can be built to meet these needs. Results are to be illustrated with some examples. Two sets of chemicals were studied and the corresponding experimental results were correlated to 90 molecular descriptors. The models were built following a regression analysis. The best multi-linear regressions presenting 6 parameters or less with high determination and cross-validation coefficients are withheld as predictive models.
Victor Panaretos, Laya Ghodrati
Victor Panaretos, Laya Ghodrati