AI-Driven Optimisation of Biochemistry instruction: Integrating Computational Modelling, 3D Molecular Simulation and Intelligent Adaptive Learning Systems
Litsenziya: cc-by
Annotatsiya
Biochemistry schooling remains challenged by the phenomenon of high conceptual solidity, poor retention, and low scalability of personalization. In this study, an integrated method of teaching biochemistry developed with the use of artificial intelligence (AI) is investigated. The method consists of the use of an intelligent tutoring system (ITS), large language model (LLM) based on Digital Human, 3D simulation, and monitoring of students in danger based on AI. The method was tested through a 12-week course of biochemistry (n=240). The student performed with the use of a randomized control trial with the comparison of four forms of teaching biochemistry: traditional lecture, AI-supported training, 3D simulation only, and AI and simulation. The combined approach showed the best results in post-test compared to other approaches with results being 83.5% ± 7.6%. The increase in results compared to traditional teaching was 38.2%. The achievement of Bloom’s taxonomy was at 68% in comparison to 38% of controls. The dropout risk classifier based on the Transformer architecture has shown AUC = 0.957 with precision equal to 89.2% and F1 = 85.7%. A 3D knowledge gain surface was created to recognize possible combinations of engagement and simulation types that can lead to the best result, as well as Bayesian AI recommendation model. NASA-TLX analysis allowed identifying the decline of cognitive load comparing it to the traditional teaching method at the level of 29%.
Hali tarjima qilinmagan