Predicting Biodegradability using Thermodynamic Descriptors

The ability to predict the biodegradability of a given compound is crucial in understanding the fate of this compound in the environment, which can have tremendous public health implications. Traditionally, the biodegradability prediction relies on the regression analysis utilizing quantitative structural relationships. This paper presents a novel prediction method utilizing neural networks and other physical/chemical/thermodynamical data besides structures. This method has shown to be successful at predicting biodegradability with better accuracy than some current popular methods.