Science and engineering departments face high student attrition due to perceived difficulty of courses in these disciplines. To subdue student attrition, students need to be guided by individual tutors to help them learn, practice and test their understanding of concepts. However, due to the exorbitant cost and time involved, this is not practical. In this article, we argue that computer-based tutors authored by teachers can serve as a useful tool to assist student learning in challenging scientific concepts. About 8,000 fine-grained interactions with our tutor by 72 students in two countries--USA and Philippines--were analysed in the framework of learning curve theory to estimate prior knowledge, learning rate and residual error rate to gauge tutor efficiency. Computer-based tutors accelerate learning and such tutors are viable, effective, and facile options to improve student learning.