Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. The manuscript lacks a research flowchart.
2. The sample size and distribution of the experimental design are limited, as the intervention group involved only a single site. This restricts the generalizability of the results, and the intervention time during a single course cycle may be insufficient to produce observable leaps in academic achievement.
3. Triple-difference (DDD) analysis results show that the App's overall pooled effect on ECG exam performance is nearly zero, with directional contradictions in performance across different exams. This indicates that the tool's direct contribution to improving student exam scores in the short term has not yet received statistical support.
4. The study points out a potential mismatch between the App's practice content and the exam questions. If the scaffolding provided by the App (such as waveform annotations) is not effectively transformed into students' independent deductive abilities in an unassisted exam environment, the effect of learning transfer will be limited.
5. The App was only introduced during a specific cardiovascular physiology module. The short exposure time and single curricular component may be insufficient for students to establish the deep cognitive changes required for processing complex ECGs, especially when facing atypical or complex clinical cases.  
6. It is suggested that the authors cite relevant latest research results (e.g., https://doi.org/10.3389/fpubh.2026.1747362) in the introduction or discussion sections to further strengthen the authority of the research background.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper mainly explores the "pitch encoding" model driven by generative artificial intelligence, aiming to rapidly develop customized educational applications. It uses the learning tool for interpreting electrocardiograms in medical education as a practical case to verify the feasibility of this development model and evaluate its application effect. This research topic closely aligns with the actual needs of developing artificial intelligence-assisted educational tools in medical education. The research design is complete and has a clear quasi-experimental framework. The discussion section deeply analyzes the value and limitations of pitch encoding. It has certain academic innovation and practical guidance significance for teacher-oriented educational technology innovation. Overall, this paper has a rigorous structure and solid research work. However, it still needs targeted revisions and improvements in terms of the comprehensiveness of the literature review, academic standardization, result discussion, and content completeness. Specific revision suggestions are as follows:

1. Generally well-written paper, to improve the quality of the paper, please check your sentences and/or English one more time.

2. Keywords should not include the word that existed in the title.

3. The introduction section briefly summarizes the application of generative artificial intelligence in the development of educational tools, but the coverage of the cutting-edge artificial intelligence modeling and algorithm-driven development methods is not comprehensive enough. It is suggested to supplement and integrate the latest intelligent algorithms and neural network modeling methods that have gained significant attention in the field of artificial intelligence applications and educational technology development in recent years, to expand the depth and cutting-edge nature of the literature review, and highlight the innovative aspects of the virtual coding paradigm in this article. The core methods and applicable scenarios referred to in this article are as follows:

*Artificial Intelligence-Enhanced Mathematical Derivation method (2026 Engineering Applications of Artificial Intelligence, 173,114464)

*Fractional sub-equation neural networks (fSENNs) method（2025 Chaoss 35:043110）

*Bilinear neural network method (2019 Nonlinear Dynamics 95:3041-3048)

*multi-modal neurosymbolic reasoning intelligent algorithm (2025 Chinese Physics Letters 42:100002)

*Bilinear residual network method (2022 Nonlinear Dynamics 108:521-531)

*Neural network-based symbolic calculation approach （2025 Chaos, Solitons and Fractals 194:116232）

4.The conclusion should include ideas for future work.

I recommend publishing the revised manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This is a timely and interesting paper. The strongest part is the practical demonstration that a single educator could rapidly build a curriculum-specific ECG learning app using a genAI-assisted workflow, with clearly described pedagogical features and strong student satisfaction scores. However, the educational effectiveness claim is much weaker than the framing sometimes suggests: the exam-level triple-difference results are mixed, and the pooled estimate is essentially null. The study is also limited by the use of one intervention site, aggregate rather than individual-level outcome data, no objective usage logs, and a modified survey instrument.

1. The paper would benefit from more methodological transparency. In particular, please clarify exactly how focal questions were selected across exams/cohorts, how the confidence intervals were computed, and what role GPT 5.4 played in the analysis pipeline so that the work is reproducible.
2. The evidence base is still limited. With only one treated site, aggregate site-level data, and no individual usage-outcome linkage, the causal claims should be softened further.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have revised the manuscript according to my suggestions.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have made efforts to address the previous comments. The revised manuscripts has improved. I would recommend acceptance.