At present, the research on the teacher-student interaction (TSI) mode in smart classroom (SC) was not in-depth enough, and a few studies mainly focuses on top-down theoretical deduction. This study adopted a bottom-up approach to study the TSI modes in SC and proposed specific implementation suggestions, which had important theoretical value and practical significance for promoting the development of students' higher-order thinking (HOT) and the digital transformation of education practice. The study used lag sequential analysis to encode and analyze 12 SC teaching videos that won first prize in Jilin Province, China. A total of 4617 sequence relationships and 48 significant activity sequences were generated. The analysis of these significant activity sequences showed that SC was divided into four types of TSI modes: discovery question, conceptual scheme, validation scheme, and creating products. There was a sequential relationship between the four types. The research proposed specific implementation suggestions from four aspects: interaction goals, interaction methods, interaction content, and interaction media. This will guide teachers in building high-quality SC and assist students in developing HOT.