Application of Machine Learning in the Development of Fourth Degree Quantitative Structure-Activity Relationship Model for Triclosan Analogs Tested against Plasmodium falciparum 3D7.

Despite new therapies against malaria, this disease remains one of the main causes of death affecting humanity. The phenomenon of resistance has caused concern as the drugs no longer have the same efficacy, forcing scientific research to develop new methods that prospect new molecules. With the advancement of artificial intelligence, it becomes possible to apply machine learning (ML) techniques in the discovery and evaluation of new molecules, by employing the quantitative structure-activity relationship (QSAR), a classic method that uses regressions to create a model that allows identifying and evaluating new drug candidates. This work combined QSAR with ML and developed a supervised model that modeled a fourth degree polynomial equation capable of identifying new drug candidates derived from the triclosan compound-a classic inhibitor of Plasmodium falciparum growth, the cause of severe malaria. The model produces an R (2) greater than 80% for training and concurrent testing, as well as a correlation index greater than 80% between the calculated and experimental pEC(50) (negative logarithm of half maximal effective concentration) data. In addition, a web software (PlasmoQSAR) was created that allows researchers to calculate the EC(50) (half maximal effective concentration) of new molecules using the developed analytical method.