基于人工智能的《生物信息学》教学重构与达成度分析

AI-Based Teaching Reconstruction and Achievement Degree Analysis of “Bioinformatics”

摘要: 为破解传统《生物信息学》教学中“重软件操作、轻算法原理”以及“理论与实践应用脱节”的痛点，本文基于能力导向(OBE)教育理念，探索了人工智能(AI)赋能的《生物信息学》课程混合式教学重构，并在生物工程专业2023级两个班级(n = 83)的教学实践中进行了深入应用。课程建立了以理论基础、算法原理与综合应用为核心的三级课程目标，将大语言模型(LLM)助教及智能化教学工具深度融入课堂教学、R语言代码剖析、序列比对演练与实验数据流构建。基于课程考核数据的目标达成度定量分析表明：课程总达成度均值达0.782，其中以AI辅助代码理解与实践为主的课程目标2达成度最高(0.944)，而强调综合分析与复杂问题解决的课程目标3达成度为0.782，以期末闭卷考核为主的理论概念(课程目标1)达成度为0.748。数据揭示出学生在“工具流依赖”与“底层机制理解”之间的知识迁移壁垒。结合此教学痛点，本文提出了强化科研实战驱动、改“流程操作”为“问题驱动”、优化形成性动态评价等持续改进路径，为新工科背景下AI技术深度融合生物信息学及相关交叉学科教学提供了可借鉴的范式。

Abstract: To solve the pain points of traditional “Bioinformatics” teaching, such as “emphasizing software operation while ignoring algorithm principles” and “disconnection between theoretical knowledge and practical application”, this paper explores the artificial intelligence (AI)-enabled blended teaching reconstruction of the Bioinformatics course based on the Outcome-Based Education (OBE) concept. The reconstructed teaching model has been fully applied in the teaching practice of two classes (n = 83) of the 2023-grade Bioengineering major. This course constructs a three-level curriculum objective centered on theoretical foundation, algorithm principles and comprehensive application. Large Language Model (LLM) teaching assistants and intelligent teaching tools are deeply integrated into classroom teaching, R language code analysis, sequence alignment exercises and experimental data flow construction. The quantitative analysis of curriculum objective achievement degree based on course assessment data shows that the overall curriculum achievement degree reaches an average of 0.782. Among them, Curriculum Objective 2, focusing on AI-assisted code understanding and practical operation, achieves the highest achievement degree of 0.944; Curriculum Objective 3, highlighting comprehensive analysis and complex problem-solving ability, has an achievement degree of 0.782; Curriculum Objective 1, dominated by final closed-book examination for theoretical concepts, scores an achievement degree of 0.748. The data reveals the knowledge transfer barrier of students between “tool workflow dependence” and “underlying mechanism understanding”. In view of the above teaching problems, this paper proposes continuous improvement strategies, including strengthening scientific research practice-driven teaching, transforming “process-oriented operation training” into “problem-driven teaching”, and optimizing dynamic formative evaluation. It provides a referable paradigm for the in-depth integration of AI technology into the teaching of Bioinformatics and related interdisciplinary courses under the background of emerging engineering education.