With the rapid development of Information and Communication Technology (ICT), traditional college mathematics teaching is increasingly showing deficiencies in knowledge delivery, student engagement, and personalized support, making it difficult to adapt to the diverse and advanced learning needs of current college students. Based on an in-depth analysis of the current situation and main issues of mathematics teaching in colleges and universities, this paper proposes a teaching method based on an ICT platform. The platform adopts a three-tier architecture comprising foundation, data, and application layers, integrating functions such as curriculum knowledge graph construction, learning behavior data collection, student portrait generation, personalized lesson plan design, and intelligent matching of teaching resources via the Teacher-Machine-Student (TMS) integration. This system facilitates precise alignment of instructional content with student profiles, enabling personalized learning arrangements through diverse pathways and levels. Its implementation enhances teaching efficiency, refines pedagogical pathways, and drives the transformation of college mathematics education toward intelligent, digitally enhanced formats.