Generative AI adaptive narratives to enhance nursing diagnostic reasoning: a classroom innovation
- Open Access
- 31.01.2026
- Research
Erschienen in:
Abstract
Introduction
Standardized nursing languages—NANDA-I nursing diagnoses, Nursing Interventions Classification (NIC), and Nursing Outcomes Classification (NOC)—are essential foundations for diagnostic reasoning, coherent care planning, and evidence-based nursing practice [1‐3]. Accurate diagnostic reasoning is central to nursing care, as imprecise diagnoses can lead to inappropriate interventions and suboptimal outcomes. However, undergraduate nursing students frequently struggle to interpret real patient cues, differentiate between closely related diagnoses, and select NIC–NOC combinations that are both coherent and contextually appropriate. Previous research indicates that learners often perceive standardized nursing languages as abstract systems disconnected from authentic patient experiences, hindering diagnostic accuracy and weakening taxonomic coherence [4, 5].
Narrative pedagogy has long been recognised as a powerful approach to address these challenges. Patient narratives expose students to the emotional, social, and experiential dimensions of illness, enabling engagement with subtle cues, contextual determinants of health, and the complexity of clinical encounters [6, 7]. Narrative learning has been associated with enhanced critical thinking, empathy, and clinical judgment [8, 9]. Nevertheless, narratives traditionally used in nursing education remain fundamentally static, offering a single version of a patient’s experience that cannot adapt to different learning needs or levels of diagnostic complexity. This limits opportunities for deliberate practice and reduces students’ exposure to variability in emotional tone, health literacy, readiness for change, and environmental barriers.
Anzeige
Across other health professions, artificial intelligence (AI) has increasingly been used to enhance diagnostic training through adaptive simulations and pattern-based signals. For example, AI-assisted tools in medical and veterinary education can identify diagnostic patterns that learners compare with clinical assessments, strengthening diagnostic competence through repetition and variation. This mirrors a core educational challenge in nursing, where learners must repeatedly interpret variable cues to refine diagnostic reasoning. Despite this parallel, the use of generative AI to support diagnostic training in nursing education remains largely unexplored.
Generative AI offers a timely and pedagogically promising solution. AI models can expand authentic clinical narratives into realistic, coherent, and context-sensitive variations while preserving fidelity to the original patient voice [10, 11]. In nursing education, AI has been applied to simulation, personalised feedback, and decision-support activities [12, 13], but no published studies have used generative AI to create adaptive narrative simulations specifically designed to support NANDA–NIC–NOC reasoning. This represents a significant gap, as taxonomic reasoning requires not only structured cognitive processing but also sensitivity to shifting contextual and emotional cues—precisely the type of variability that generative AI can model.
Transforming real patient narratives into adaptive, multi-layered simulations may provide valuable opportunities for students to practise diagnostic reasoning, identify defining characteristics, justify related factors, and align NIC interventions with measurable NOC outcomes. By generating controlled variations in emotional tone, health literacy, motivation, and environmental determinants of health, generative AI supports individualized, repeatable, and ethically safe practice—key elements in the development of diagnostic competence [14, 15].
This study addresses these pedagogical needs by developing and pilot-testing an instructional design that uses generative AI–enhanced narrative simulation to strengthen NANDA–NIC–NOC reasoning in undergraduate nursing students. Building on authentic coronary patient narratives from a previously published qualitative study [16], generative AI was used to produce adaptive narrative variations that were integrated into a structured learning activity.
Anzeige
The aim of this study was twofold:
(1)
To develop an instructional design integrating generative AI–enhanced narrative simulation into the teaching of standardized nursing language, and
(2)
To pilot-test its effects on students’ diagnostic accuracy, NIC–NOC coherence, and perceived authenticity of learning.
Methods
Study design
A mixed-methods pilot study with a convergent design was conducted, combining quantitative performance measures with qualitative reflective data. This approach was selected to capture both measurable learning outcomes and students’ experiences of the instructional innovation. Mixed-methods designs are particularly suitable for evaluating educational interventions involving emerging technologies, as they allow integration of outcome data with learners’ perceptions and reasoning processes [15].
Consistent with the exploratory nature of a pilot study, the primary aims were to assess feasibility, acceptability, and preliminary educational effects rather than to test fully powered hypotheses. Quantitative analyses were therefore used to estimate effect sizes and inform the design of future, adequately powered studies.
Methodological orientation
This study is grounded in a pragmatic educational research paradigm, focusing on the practical evaluation of a teaching strategy within a real classroom context. Pragmatism prioritizes the usefulness of an intervention and its capacity to address an identified educational problem, rather than adherence to a single epistemological tradition.
From a pedagogical perspective, the instructional design draws on narrative pedagogy and experiential learning principles, emphasizing contextual interpretation, reflective reasoning, and iterative decision-making. The qualitative component was therefore used to explore how students experienced and interpreted the adaptive narratives, complementing the quantitative assessment of learning outcomes.
Participants and educational context
Participants were 46 s-year undergraduate nursing students enrolled in the optional course Therapeutic Education in Nursing, offered during the second semester of the Bachelor of Nursing programme at the University of Santiago de Compostela. The course focuses on patient education, health literacy, behavioural counselling and the application of standardized nursing language (NANDA-I, NIC, NOC) to support patient-centred teaching–learning processes. This pilot study was designed to evaluate feasibility, acceptability, and preliminary educational effects of the instructional model; therefore, no formal a priori sample size estimation was required. Methodological guidance for pilot studies indicates that groups of 20–30 participants are adequate to explore preliminary effect sizes, refine procedures, and determine feasibility for larger trials. Our total sample of 46 students (23 per group) lies within these recommended ranges and is consistent with similar pilot evaluations in nursing education.
The intervention was delivered as part of a scheduled 90-minute seminar within the course’s practical component. Although students had been introduced to basic NANDA-I concepts in their first-year Fundamentals of Nursing course, none had previously engaged in structured taxonomic reasoning activities using real or simulated patient narratives.
Anzeige
Students attended one of two parallel seminar sessions and were randomly assigned to:
-
Control group (n = 23): static authentic patient narrative.
-
AI-enhanced group (n = 23): adaptive narrative simulations expanded from the discourse of 12 coronary patients in the original qualitative study.
Both sessions were facilitated by the same instructor to ensure consistency. Participation formed part of routine learning activities and was not tied to formal grading.
Procedures and overall study flow
The study followed a two-phase sequential structure. After recruitment and random allocation, each group participated in a 90-minute seminar session in which they completed a structured taxonomic reasoning activity based on either static or AI-enhanced narrative materials. All students received identical learning objectives, instructions and access to NANDA-I, NIC and NOC reference manuals.
Anzeige
Immediately after the activity, students completed the quantitative measures (diagnostic accuracy, NIC–NOC coherence and self-efficacy). They also submitted a short written reflection describing their learning experience.
AI-enhanced narrative materials were developed prior to the pilot and are described in Phase 1. Implementation of the seminar and evaluation procedures are detailed in Phase 2.
Phase 1: development of the generative AI–enhanced narrative simulation
Source narratives
Authentic coronary patient narratives were extracted from a previously published qualitative study examining patients’ perceptions of nutritional risk during cardiac rehabilitation [16]. Only anonymised, publicly available excerpts were used. These narratives describe challenges related to health literacy, emotional responses, environmental barriers, and readiness for lifestyle change. All source material was reviewed to ensure suitability for educational use and ethical integrity.
Artificial intelligence: model, prompting, and narrative transformation pipeline
AI model specification
A large generative artificial intelligence language model based on the GPT-4 architecture (OpenAI), accessed between March and May 2024 through a secure institutional interface, was used to generate adaptive narrative variations. The model was employed exclusively as a text-generation tool to expand and vary existing anonymised patient narratives.
No model fine-tuning, retraining, or autonomous clinical decision-making was performed. The AI system did not generate nursing diagnoses, interventions, or outcomes, nor did it evaluate student performance. All diagnostic reasoning and NANDA–NIC–NOC mapping tasks were completed solely by students using standard reference materials.
The AI model was not trained or annotated using study-specific data, and no human labelling or algorithmic learning processes were involved in the generation of the narratives.
Prompt design and narrative generation process
To ensure transparency, reproducibility, and pedagogical alignment, a structured prompt-engineering approach was implemented. The AI model was provided with:
(a)
Authentic anonymised coronary patient narrative excerpts from the original qualitative study, and.
(b)
Explicit instructional prompts defining the scope and constraints of narrative generation.
Prompts were designed to guide the AI output while preventing the introduction of new clinical conditions or diagnostic conclusions. All parameters governing narrative variation were predefined by the research team to reflect realistic sources of diagnostic variability encountered in nursing practice.
Narrative transformation pipeline
Following published recommendations for responsible AI use in health professions education [11, 13], the narrative transformation pipeline consisted of four controlled steps:
1.
Baseline reproduction
Verbatim replication of the original narrative excerpt to ensure fidelity to the patient’s voice and content coherence.
2.
Contextual expansion
Enrichment of socio-emotional and environmental detail without altering the patient’s core meaning or clinical context [10].
3.
Adaptive variation
Generation of multiple versions of the same narrative through controlled manipulation of predefined parameters, including health literacy level, emotional tone (e.g., fear, frustration, ambivalence), readiness for change, environmental barriers (e.g., hospital vending machines, food insecurity), and behavioural consistency or ambivalence. These parameters were defined a priori by the research team to mirror real-world diagnostic variability.
4.
Pedagogical structuring
Formatting of each AI-generated narrative into a standardized educational case template, including cue-rich text, explicit and implicit data points, prompts for NANDA-I diagnosis formulation, and structured fields for NIC intervention selection and NOC outcome mapping.
This pipeline produced a set of adaptive narrative simulations that preserved the authenticity of the original patient narratives while introducing pedagogically relevant variability for practicing diagnostic reasoning. A conceptual diagram illustrating the narrative transformation pipeline is presented in Fig. 1.
Human validation and quality assurance
To ensure methodological rigor and ethical integrity, all AI-generated narratives underwent structured human review prior to their use in the educational intervention. This review was conducted by the author, a PhD-prepared nurse educator with expertise in standardized nursing languages and qualitative research.
Narratives were evaluated for:
(a)
Fidelity to the original patient meaning and voice,
(b)
Internal narrative coherence,
(c)
Clinical plausibility within the scope of nursing practice, and.
(d)
Alignment with the intended pedagogical objectives for NANDA–NIC–NOC reasoning.
AI-generated outputs that introduced clinical inconsistency, unintended bias, or content beyond the defined educational objectives were discarded or revised. No narrative was used without explicit human validation. Given the exploratory and pedagogical nature of this pilot study, formal inter-rater reliability testing or benchmarking against external diagnostic criteria was not conducted.
Fig. 1
Generative AI narrative transformation pipeline. Note: Figure 1 depicts the four-step procedural workflow used to generate adaptive narrative simulations from authentic anonymised patient excerpts. Arrows indicate the sequence of processing steps (not causal/statistical relationships among study variables)
An example of one original narrative and its corresponding AI-generated variations is shown in Table 1.
Table 1
Authentic patient narrative and AI-generated adaptive variations used in the educational intervention
Narrative type | Narrative content | Pedagogical focus (NANDA–NIC–NOC) |
|---|---|---|
Real narrative (static) | “I can’t really read the small letters on food packaging…” | Identification of Deficient Knowledge (00126) or Ineffective Health Management (00078); NIC: Health Education (5510), Nutrition Counseling (5246); NOC: Knowledge: Diet (1813) |
AI variation 1: Low health literacy | “I try to read what the packages say, but for me it all looks like strange words…” | Reinforces Deficient Knowledge (00126); NIC: Teaching: Individual (5606); NOC: Knowledge: Diet (1813) |
AI variation 2: Heightened anxiety | “Since my heart problem, I’m obsessed with food…” | Differentiation between Anxiety (00146) and Ineffective Health Management; NIC: Anxiety Reduction (5820); NOC: Anxiety Level (1211) |
AI variation 3: Increased motivation | “After the cardiac rehab sessions, I really want to do things right…” | Readiness for Enhanced Health Management (00162); NIC: Motivational Interviewing (5248); NOC: Self-Management: Cardiac Disease (1617) |
Phase 2: pilot evaluation
Educational intervention
To clearly differentiate the instructional materials provided to each condition, Table 2 summarizes the educational components used in the control and AI-enhanced groups during the intervention. The intervention consisted of a 90-minute session integrated into regular coursework.
Table 2
Comparison of learning conditions between the control and AI-enhanced groups
Component | Control group | AI-enhanced group |
|---|---|---|
Narrative format | Single static authentic patient narrative | Multiple adaptive AI-generated variations |
Narrative variability | None | High (emotional tone, literacy, motivation, context) |
Cognitive demand | Stable cues | Variable cues requiring diagnostic revision |
Learning task | Single NANDA–NIC–NOC mapping | Repeated mapping across narrative variations |
Pedagogical focus | Baseline taxonomic reasoning | Adaptive and context-sensitive reasoning |
Control condition
Students analysed the single static narrative and completed a full NANDA–NIC–NOC mapping exercise.
AI-enhanced condition
Students worked with two to three adaptive variations derived from the same patient, exposing them to different cue patterns, emotional expressions and contextual determinants.
Required tasks for all students.
1.
Extract explicit and implicit cues.
2.
Select and justify a NANDA-I diagnosis.
3.
Choose coherent NIC interventions.
4.
Map appropriate NOC outcomes.
5.
Produce a brief patient-centred care plan.
Outcomes
1.
Diagnostic accuracy (NANDA-I)
Assessed using a rubric adapted from a previously validated diagnostic evaluation framework [5].
2.
NIC–NOC coherence
Assessed using a researcher-developed rubric.
3.
Self-Efficacy in standardized nursing language
Measured with a 7-item Likert scale.
4.
Qualitative reflections
Explored perceived authenticity, engagement and contextual understanding.
Data analysis
Quantitative analysis included descriptive statistics, independent-samples t-tests or Mann–Whitney U tests, and effect sizes (Cohen’s d). Statistical analyses were conducted using SPSS version 29.
Qualitative data were analysed using reflexive thematic analysis following the six-phase approach described by Braun and Clarke [17]. This method was selected for its flexibility and suitability for exploring experiential learning data in educational research. Analysis proceeded through familiarisation with the data, initial coding, theme development, review, definition, and reporting. Coding focused on students’ descriptions of diagnostic reasoning processes, interpretation of narrative cues, and perceived authenticity of the learning activity. Themes were developed inductively from the data while remaining theoretically informed by concepts of narrative pedagogy and clinical reasoning.
Anzeige
Rigor and trustworthiness
Several strategies were employed to enhance rigor and trustworthiness in the qualitative component of the study. Credibility was supported through close engagement with the data and the use of illustrative quotations to ground interpretations in participants’ accounts. Reflexivity was addressed through ongoing critical reflection by the researcher, who acted as both educator and analyst, acknowledging this dual role and its potential influence on interpretation.
Transparency was ensured by providing a clear description of the data collection, analysis procedures, and analytical framework. Given the exploratory and classroom-based nature of the study, member checking and inter-rater coding were not conducted; however, analytic decisions were systematically documented to support dependability.
Results
A total of 46 undergraduate nursing students participated in the study. Participants were randomly assigned to either the control group (n = 23), which worked with static authentic patient narratives, or the AI-enhanced group (n = 23), which received adaptive simulations generated from the narratives of 12 coronary patients originally interviewed in the qualitative study previously published on coronary patient narratives [16]. No significant baseline differences were found between groups in age, gender, or previous exposure to standardized nursing language.
Quantitative findings
Although both groups worked exclusively with written narratives, students in the AI-enhanced condition perceived the adaptive cases as more realistic and contextually rich. Quantitative results are summarised in Table 3. Across all outcome measures, the AI-enhanced group demonstrated significantly higher diagnostic accuracy, stronger NIC–NOC coherence and greater self-efficacy than the control group, with large effect sizes.
Table 3
Quantitative outcomes by study group
Outcome | Control group (n = 23) Mean (SD) | AI-enhanced group (n = 23) Mean (SD) | p-value | Effect size (d) |
|---|---|---|---|---|
Diagnostic accuracy (0–4) | 2.39 (0.62) | 3.17 (0.55) | < 0.001 | 1.34 |
NIC–NOC coherence (0–4) | 2.74 (0.69) | 3.51 (0.58) | < 0.001 | 1.25 |
Self-efficacy (1–5) | 3.08 (0.57) | 3.72 (0.49) | < 0.001 | 1.23 |
Qualitative findings
A reflexive thematic analysis of 46 reflective narratives generated three themes that help explain how the AI-enhanced narrative simulation supported learning. These themes summarise students’ perceived mechanisms of diagnostic reasoning development, including enhanced cue detection, iterative refinement and increased authenticity. Themes and illustrative quotations are presented in Table 4.
Students consistently highlighted that the adaptive narratives improved their ability to interpret fluctuating emotional tone, motivation and contextual barriers. They described revising their NANDA-I,–NIC–NOC decisions as conditions changed, mirroring real diagnostic practice. Perceived authenticity was enhanced by the grounding of all adaptive narratives in the voices of the original coronary patients.
Table 4
Themes identified through reflexive thematic analysis of student reflections
Theme | Description | Illustrative quotation |
|---|---|---|
Enhanced sensitivity to narrative cues | Improved detection of emotional, contextual, and behavioural cues | “Seeing how the same patient reacted differently made me pay closer attention to details.” |
Iterative refinement of reasoning | Revision of diagnostic and care-planning decisions across variations | “When the emotional tone changed, I had to rethink the diagnosis.” |
Perceived authenticity and engagement | Increased realism and engagement with learning materials | “It felt like meeting the same patient on different days.” |
Feasibility and acceptability
-
100% of participants completed the activity.
-
91% of students in the AI-enhanced group rated the experience as “useful” or “very useful.”
-
87% expressed interest in using additional AI-generated scenarios in future coursework.
-
No technical issues or ethical concerns were reported.
Discussion
This study provides preliminary evidence that generative AI–enhanced narrative simulation can strengthen undergraduate nursing students’ competence in using standardized nursing language. Students exposed to adaptive AI-generated cases demonstrated higher diagnostic accuracy, greater NIC–NOC coherence and increased self-efficacy than students working with a single static narrative. Qualitative findings revealed that the adaptive variations supported deeper engagement, improved cue sensitivity and encouraged iterative refinement of taxonomic reasoning—skills that are central to nursing diagnostic competence.
Integration with existing literature
The improved diagnostic accuracy observed in the AI-enhanced group aligns with research showing that narrative pedagogy helps learners interpret contextual cues and patient perspectives more effectively [6, 9]. However, traditional narratives are inherently static and cannot provide repeated, varied exposure to the complexity of real patient encounters. The adaptive AI-generated cases in this study overcame these limitations by maintaining fidelity to authentic patient voices while systematically varying emotional tone, health literacy, motivation and environmental conditions. This reflects the established value of scenario variation in simulation-based learning, which promotes deeper cognitive processing and reduces diagnostic anchoring [15].
The iterative refinement of NANDA–NIC–NOC mapping described by students further supports the relevance of deliberate practice theory [14]. Diagnostic reasoning requires repeated opportunities to test hypotheses, compare scenarios and revise decisions—yet generating such variability manually is time- and resource-intensive for educators. The AI narrative pipeline developed in this study addresses this pedagogical gap by enabling rapid production of multiple, realistic narrative variations that mirror the shifting cues seen in clinical encounters.
The increased perceived authenticity also resonates with prior work demonstrating that emotional realism enhances empathy, engagement and clinical judgment [7, 8]. Students reported that the AI-enhanced cases felt like “meeting the same patient on different days,” reflecting the ecological validity of the model. Notably, the adaptive simulations were grounded in the real voices of twelve coronary patients who expressed fear, confusion, ambivalence and motivation surrounding dietary change—core elements that the AI preserved and expanded.
The pedagogical mechanisms observed in this study parallel trends across other health professions, where AI technologies are increasingly used to support diagnostic training. For example, AI-assisted auscultation devices in veterinary and medical education provide variable pattern-based cues that students compare with clinical assessments. Similarly, generative AI in this study did not “diagnose” but rather produced a controlled variety of narrative cues that students used to practice diagnostic reasoning. This alignment reinforces the legitimacy and educational value of integrating AI into nursing training. The absence of a formal sample size calculation is consistent with the goals and design of pilot studies, where the primary aim is to obtain feasibility insights and preliminary effect sizes to inform future, adequately powered research.
Mechanisms of learning
Three interacting learning mechanisms emerged from the qualitative analysis:
1.
Enhanced cue sensitivity
Students exposed to narrative variation paid closer attention to emotional tone, health literacy barriers and environmental constraints. This is essential for selecting accurate NANDA-I diagnoses, particularly when differentiating between closely related labels (e.g., Ineffective Health Management vs. Ineffective Health Maintenance).
2.
Iterative reasoning through variation
Changes across adaptations prompted students to revisit and justify their diagnostic and care-planning decisions. This process mirrors cognitive apprenticeship models that emphasize guided variation, repeated exposure and reflective decision-making.
3.
Authenticity-driven engagement
Students found the variations emotionally credible, realistic and relatable—a key driver of engagement. Engagement is a well-documented mediator of clinical reasoning, improving attention, retention and transferability of learning.
Together, these mechanisms suggest that AI-generated narrative variation supports diagnostic reasoning not by replacing human instruction but by enriching the learning environment with safe, controlled and ethically grounded complexity.
Educational significance
To our knowledge, this study represents one of the first applications of generative AI to expand authentic patient narratives into adaptive simulations designed to teach NANDA–NIC–NOC reasoning. The approach offers several pedagogical advantages:
-
Unlimited generation of realistic cases without requiring additional patient data.
-
Scalable variability allowing exposure to multiple scenarios within a single teaching session.
-
Personalization of learning, as instructors can tailor emotional tone, literacy level or environmental barriers.
-
Strengthening of diagnostic reasoning, an essential yet often underdeveloped competency in undergraduate education.
-
Ethically safe implementation, with all adaptive content grounded in anonymized, published narratives and manually verified.
Given the accelerating incorporation of AI into nursing curricula worldwide, this instructional design provides a structured and responsible model for leveraging generative technologies in ways that enhance, rather than replace, faculty expertise.
Limitations
Several limitations of this study should be acknowledged. First, as a pilot study conducted in a single academic institution with a modest sample size, the findings cannot be generalized to all undergraduate nursing programmes. The primary aim was to assess feasibility, acceptability, and preliminary educational effects rather than to provide definitive evidence of effectiveness. Future studies with larger, multi-site samples and a priori power calculations are needed to confirm and extend these findings.
Second, although the AI-generated narratives were grounded in authentic patient discourse and underwent structured human review, formal expert panel validation or inter-rater reliability procedures were not conducted. Future research could strengthen methodological robustness by incorporating multiple reviewers and formal agreement metrics to evaluate narrative fidelity and pedagogical alignment.
Third, outcome measures were collected immediately following the educational intervention. As such, the study does not provide information on long-term retention of diagnostic reasoning skills or transferability to clinical practice settings. Longitudinal designs examining sustained learning outcomes and performance during clinical placements would be valuable.
Finally, while safeguards were implemented to ensure ethical and responsible use of generative AI, including controlled prompting and human validation, AI-generated content may still carry a risk of unintended bias or distortion. Continued refinement of ethical frameworks and institutional guidelines will be essential as generative AI tools are increasingly integrated into nursing education.
Future directions
Future research should:
-
Conduct larger, multi-site studies to evaluate scalability and comparative effectiveness.
-
Examine longer-term impacts on clinical placement performance, documentation accuracy and diagnostic consistency.
-
Validate the narrative transformation pipeline in other domains such as mental health, geriatrics or primary care.
-
Explore integration of generative AI into adaptive tutoring or competency-based learning systems.
-
Develop standardized frameworks to guide ethical and pedagogically sound use of AI-generated narratives in nursing education.
Conclusion
Generative AI–enhanced narrative simulation represents a novel and pedagogically robust strategy for strengthening undergraduate nursing students’ diagnostic reasoning using standardized nursing languages. By transforming authentic coronary patient narratives into adaptive, cue-rich scenarios, this instructional design directly addresses longstanding challenges in teaching NANDA–NIC–NOC, particularly the difficulty of translating theoretical taxonomies into meaningful clinical decisions. Findings from this pilot study suggest that generative AI can enhance diagnostic accuracy, improve the coherence of intervention–outcome mapping and foster greater learner engagement by simulating the emotional, contextual and behavioural variations observed in real patient encounters.
A key contribution of this work is demonstrating that generative AI can ethically and effectively expand narrative pedagogy while preserving the authenticity of patient voices. The AI pipeline does not replace or distort real narratives; instead, it amplifies them by generating controlled variations in emotional tone, health literacy, motivation and environmental barriers—dimensions that are essential for high-quality diagnostic reasoning but challenging to reproduce manually at scale. These adaptive narratives created a dynamic learning environment that encouraged students to test alternative diagnostic hypotheses, refine their clinical judgments and deepen their understanding of the relationships between NANDA diagnoses, NIC interventions and NOC outcomes.
Given the rapid emergence of AI tools in health professions education, this study offers a feasible, scalable and customizable method for integrating generative models into nursing curricula. The approach supports personalized learning, allows educators to modulate complexity and introduces realistic scenario variability without compromising patient confidentiality. Importantly, it positions AI as a complement—not a substitute—to human teaching, aligning with international calls for innovation, ethical integration and pedagogical relevance in nursing education.
Future research should examine long-term effects on clinical documentation, diagnostic consistency in practice settings and students’ performance during clinical placements. Multi-site studies will help determine generalizability, while refinement of the AI pipeline may facilitate applications in mental health, chronic illness management and interprofessional education.
Overall, this study provides early but compelling evidence that generative AI can meaningfully enrich the teaching of standardized nursing language. By expanding the possibilities of narrative pedagogy, generative AI supports the development of more reflective, context-aware and diagnostically competent future nurses prepared for evolving clinical environments.
Acknowledgements
The author would like to thank the undergraduate nursing students who participated in the educational activity and the University of Santiago de Compostela for providing the academic environment that made this project possible. Their engagement and openness greatly contributed to the development of this instructional innovation.
Declarations
Ethics approval and consent to participate
This study was conducted as part of a classroom-based educational innovation within the undergraduate nursing curriculum and did not involve patients, clinical interventions, or the collection of identifiable personal data. According to the regulations of the Comité de Ética de la Investigación de la Universidad de Santiago de Compostela, educational studies of this nature meet the criteria for exemption from formal ethical review; therefore, ethics approval was not required. The study adhered to the ethical principles of the Declaration of Helsinki. All participating students received written and verbal information about the activity, and participation was entirely voluntary and anonymous. Completion of the learning activity and associated questionnaires was considered informed consent, in accordance with institutional policy. The patient narratives used as source material were obtained from a previously published, fully anonymised qualitative study [16].
Consent for publication
Not applicable. The manuscript does not contain any identifiable images, personal information, or clinical details of individual participants.
Competing interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.