Psychological care refers to the proactive psychological support provided by medical professionals within clinical settings (Cochrane, 2015), which is essential for optimal patient mental health, recovery and treatment continuity (Liao et al., 2020). However, nursing students often struggle to translate theoretical knowledge into clinical practice due to the subject’s inherent complexity and abstractness, which make concepts difficult to grasp and retain (De Vries & Timmins, 2012). Additionally, limited classroom time hinders the deep internalisation and transfer of knowledge (Carless-Kane & Nowell, 2023).

To address these challenges, massive open online courses (MOOCs) and small private online courses (SPOCs) using self-directed learning models have been integrated into nursing education (Zhu, 2019). Students use these platforms to engage with resources such as lecture videos and self-assessments, which are combined with flipped classrooms to create a blended teaching model (Shen & An, 2022). However, this model has inherent limitations: knowledge is often fragmented and lacks systematic connections, hindering students’ ability to construct a comprehensive framework (Ji & Zhong, 2023); the preset standardised resources are not easily adaptable to individual knowledge levels, thereby limiting personalised learning paths (Zeng et al., 2015); and delayed teacher feedback or difficulties in independently resolving problems reduce learning efficiency (Jiang & Liang, 2023; Mcandrew & Scanlon, 2013).

As a structured semantic network, the course knowledge graph visualises relationships between core concepts and organises teaching resources accordingly (Abu-Salih & Alotaibi, 2024), effectively aiding students’ understanding and retention of complex knowledge (Ramazanova, Sambetbayeva, Serikbayeva, Sadirmekova, & Yerimbetova, 2024). Studies indicate that knowledge graph-based personalised learning paths can enhance knowledge mastery and course completion rates (Ji et al., 2022, Qu et al., 2024, Yang et al., 2023). Concurrently, artificial intelligence (AI) is increasingly used in professional education. Intelligent question-answering systems based on professional knowledge outperform general models in terms of accuracy and effectiveness (Arun et al., 2024, Shi et al., 2023). Furthermore, AI assistants providing real-time feedback improve clinical skills and knowledge application more effectively than remote guidance (Fazlollahi et al., 2022). Building on these advances, Yang, Ma, and Yi (2025) integrated knowledge graphs, AI assistants, intelligent Q&A and teaching modules into an intelligent teaching system that generates personalised learning paths for differentiated instruction.

However, the mere integration of technical components does not guarantee pedagogical success. These developments collectively herald a new paradigm that transcends the traditional content-delivery platform: the Artificial Intelligence-Human Intelligence Collaboration (AI-HI Collaboration, aimed at fostering “co-intelligence”). This paradigm posits that the true potential of educational technology lies in achieving a purposeful and structured collaboration between human intelligence and artificial intelligence (Järvelä, Nguyen, & Hadwin, 2023). To actualize this concept, this study draws on the classic cognitive apprenticeship theory and develops the “structured cognitive apprenticeship model based on AI-HI collaboration of intelligent learning platform” (Brown, Collins, & Duguid, 1989). The theory emphasizes externalizing experts’ implicit thinking and developing learners’ higher-order cognitive skills through modeling, coaching and scaffolding (Cakmakci, Aydeniz, Brown, & Makokha, 2020). Building on this theory, the present study strategically employs two core modules—the knowledge graph and the problem graph—to construct dual-track cognitive scaffolding. Specifically, the knowledge graph translates abstract psychological concepts into a visualised, structured semantic network, facilitating students in building a systematic knowledge framework. Conversely, the problem graph deconstructs complex clinical scenarios into hierarchical reasoning steps, progressing from core issues to competency requirements and ultimately to quality-related questions. This structured approach provides a practical scaffold for training in clinical decision-making.

This approach upgrades the traditional master-apprentice relationship to a tripartite collaborative ecosystem designed and guided by teachers (leading HI), explored and constructed by students (central HI) and supported and empowered by the intelligent platform (empowering AI). This integrated method addresses key limitations of traditional online learning by employing structured networks to reduce fragmentation, offering instant AI feedback to minimise delay and enabling data-driven personalised learning.

Although integrating knowledge graphs, AI and teaching theory is promising, current research focuses mainly on technical feasibility and functional description (Li et al., 2024), lacking empirical evidence on the effectiveness and mechanisms of such integration within educational theory (Järvelä et al., 2023). Therefore, this study aims to evaluate the effectiveness of a structured Al-HI cognitive apprenticeship model in psychological nursing education and explore its underlying mechanisms. The study adopts an explanatory sequential mixed-methods design guided by two research questions: (1) whether the model outperforms a standard SPOC approach in improving nursing students’ knowledge and skills; and (2) how it enhances learning processes and bridges the theory–practice gap. A randomised controlled trial and qualitative analysis generate robust evidence and insights for designing digital learning environments.