Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Generative Artificial Intelligence in Healthcare: Applications, Implementation Challenges, and Future Directions

Summary

This review gives a comprehensive and well-structured overview of generative AI in healthcare, covering its applications across clinical documentation, diagnostics, patient communication, drug discovery, medical education, mental health, and imaging. The authors also address current implementation challenges, ethical concerns, and regulatory consideration. The inclusion of up-to-date references (2022–2025) along with evidence from real-world pilots and clinical studies serve as an useful reference for clinicians, researchers, and healthcare policymakers, adding to the strength of the paper. 

General 

• The article is impressively exhaustive in its coverage. 
• It beautifully highlights GenAI's potential across multiple domains of healthcare. The structure is clear and logical– starting with foundational concepts and moving into application-specific details. 
• Inclusion of limitations, critical analysis of current studies and risks could further strengthen the soundness of the paper.

Article 

1. A short section on how references were gathered (e.g., keywords searched, databases used) would add transparency.
2. Figures and tables are well-integrated. Table 1 and Table 3 could benefit from improved formatting and column headings.

Review

1. The paper includes recent advancements along with an exhaustive array of clinical applications supported by current evidence.
2. The references are recent, appropriate, and diverse– good job!
3. The manuscript could be strengthened by including a critical discussion on the limitations and risks of using GenAI in the real-world.

Specific

• Lines 73–75: Update Figure 1 citation in correct numerical order.
  
• Section 3.1–3.3: Summarize each subsection to reinforce key takeaways.
  
• Line 273–278: Add a reference for the MySurgeryRisk system if available.

Author Response

Reviewers’ Comments

 

Reviewer 1

1. A short section on how references were gathered (e.g., keywords searched, databases used) would add transparency.

Response:

We appreciate the reviewer’s valuable suggestion regarding addition of methodology for literature survey as this would enhance and the transparency of the manuscript.

In response to this comment, we have incorporated Methodology as a separate section which describes how references were gathered on page no.3.

Modifications done are highlighted on page no. 3 in the manuscript

2. Figures and tables are well-integrated. Table 1 and Table 3 could benefit from improved formatting and column headings.

Response:

We appreciate the reviewer’s insightful suggestion regarding the improvement of table 1 and table 3 formatting and column heading.

To address this, we have carefully gone through the manuscript to ensure best suitable column heading and formatting.

Modifications done are highlighted in table 1and table 3 in the manuscript

Specific

1) Lines 73–75: Update Figure 1 citation in correct numerical order.

Response:

We appreciate the reviewer’s insightful suggestion regarding the citation of the figure 1 in the text.

To address this, we have added the appropriate citation in the text.

Modifications done are highlighted in text on page no 3.

 

2) Section 3.1–3.3: Summarize each subsection to reinforce key takeaways

Response:

We appreciate the reviewer’s valuable suggestion to summarize subsection 3.1-3.3 which are now renumbered as section 4.1-4.3. This would help to provide key takeaways to the readers.

To address this, we have added the appropriate summary of the above said sections i.e. 3.1-3.3

Modifications done are highlighted in text on page no.9, 12, 15.

3) Line 273–278: Add a reference for the MySurgeryRisk system if available.

Response:

We appreciate the reviewer’s insightful suggestion to add the suitable reference to Line 273–278.

To address this, we have added the appropriate reference in the text.

Modifications done are highlighted in the text on page no.10 and reference section.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The review article offers a comprehensive and timely synthesis of the current landscape, challenges, and future directions of artificial intelligence (AI) in medicine, with particular attention to generative AI tools and their broad-ranging applications. The article covers important topics such as, generative ai tools and applications, clinical practice and research integration, clinical decision support and education, challenges, and ethical considerations.

Overall, the review succeeds in synthesizing a complex and rapidly evolving field. Its structured approach makes, it a valuable resource for clinicians, researchers, and policymakers seeking to understand and navigate the transformative impact of AI. 
The article is very well presented and should be published in the current form.

Author Response

 

Reviewers’ Comments

 

1) The review article offers a comprehensive and timely synthesis of the current landscape, challenges, and future directions of artificial intelligence (AI) in medicine, with particular attention to generative AI tools and their broad-ranging applications. The article covers important topics such as, generative AI tools and applications, clinical practice and research integration, clinical decision support and education, challenges, and ethical considerations.

Overall, the review succeeds in synthesizing a complex and rapidly evolving field. Its structured approach makes, it a valuable resource for clinicians, researchers, and policymakers seeking to understand and navigate the transformative impact of AI. 
The article is very well presented and should be published in the current form.

Response:

We sincerely thank the reviewer for their thoughtful and encouraging feedback. We are pleased to know that the manuscript was found to be comprehensive, well-structured and relevant to the ongoing discourse in healthcare AI. We especially appreciate your recognition of the article's synthesis of current developments, challenges, and ethical considerations surrounding generative AI.

Your positive remarks regarding its value to clinicians, researchers, and policymakers reinforce our goal of providing a practical and informative resource. We are grateful for your recommendation for publication and have made every effort to ensure the manuscript meets the highest academic and ethical standards.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

This study has a focus on the use of GenAI in healthcare and discusses the applications, implication challenges, and future directions. The summarization is good, however this is more like a report instead of a scientific paper. To improve the quality of this study, the authors need to clarify the methodology used in this research. For example, is this a literature review? If so, how did the authors conduct the literature review? 

Also, there is unresolved comment in the submitted manuscript. Please double-check the manuscript before re-submission. 
There is also a figure in the Abstract, please check if it fits the journal requirement. 

I hope these suggestions are helpful for the improvement of this manuscript. 

Author Response

 

Reviewers’ Comments

 

 

1) To improve the quality of this study, the authors need to clarify the methodology used in this research. For example, is this a literature review? If so, how did the authors conduct the literature review? 

Response:

We appreciate the reviewer’s valuable suggestion regarding addition of methodology for literature survey as this would enhance and the transparency of the manuscript.

In response to this comment, we have incorporated Methodology as a separate section which describes how references were gathered on page no.3.

Modifications done are highlighted on page no. 3 in the manuscript

 

2) There is unresolved comment in the submitted manuscript. Please double-check the manuscript before re-submission. 

Response:

We appreciate the reviewer drawing attention to the unresolved comment. We have carefully reviewed the entire manuscript and have now addressed and removed all tracked changes, editor comments to ensure the document is clean and ready for resubmission.

3) There is also a figure in the Abstract, please check if it fits the journal requirement. 

Response:

We appreciate your attention to journal formatting standards. After reviewing the journal’s submission guidelines, we have confirmed that Figure 1 in the abstract complies with the manuscript requirements. It is designed to provide a concise visual overview and enhance reader engagement. Nevertheless, we remain open to relocating or modifying the figure based on the Editorial Office’s final formatting guidance.

Reviewer 4 Report

Comments and Suggestions for Authors

This paper provides a comprehensive review of generative artificial intelligence (AI) applications in healthcare, covering clinical documentation, patient communication, diagnostics, medical imaging, drug discovery, and medical education. It highlights the transformative potential of generative AI in improving efficiency, reducing administrative burdens, and enhancing patient engagement while acknowledging challenges such as accuracy, bias, privacy, and ethical concerns. The review synthesizes recent advancements, clinical studies, and regulatory considerations, offering a balanced perspective on the current state and future directions of generative AI in healthcare. The paper is well-structured, with clear examples and references to recent research, making it a valuable resource for clinicians, researchers, and policymakers.

(1) The paper extensively discusses applications but lacks a critical analysis of the limitations of generative AI in real-world clinical settings. For instance, how do hallucination rates compare across different AI models, and what are the clinical consequences of these errors?
(2) While the review mentions regulatory frameworks (e.g., FDA, EMA), it does not delve into specific guidelines or standards for validating generative AI tools in healthcare. Could the authors provide more details on current regulatory pathways or gaps?
(3) The section on bias and health equity is brief. Given the importance of this issue, could the authors expand on strategies to mitigate bias in AI models, such as dataset diversification or algorithmic fairness techniques?
(4) The paper cites numerous studies but does not systematically compare the performance of generative AI with traditional methods or human experts in key areas like diagnostics or drug discovery. A meta-analysis or comparative table would strengthen this aspect.
(5) The discussion on future directions is speculative. Could the authors ground their predictions with concrete examples of ongoing trials or technologies nearing clinical implementation?
(6) Some references are outdated (e.g., 2022–2023). Given the rapid evolution of AI, could the authors include more recent studies (2024–2025) to ensure the review reflects the latest advancements?
(7) The figures (e.g., Figure 4) are descriptive but could benefit from annotations or captions explaining their relevance to the text. For example, how does Figure 4 illustrate the integration of AI in clinical decision support?
(8) The paper does not address cost-effectiveness or scalability of generative AI solutions. How do these technologies compare in terms of implementation costs and resource requirements across different healthcare systems?

Author Response

 

Reviewers’ Comments

 

 

1) The paper extensively discusses applications but lacks a critical analysis of the limitations of generative AI in real-world clinical settings. For instance, how do hallucination rates compare across different AI models, and what are the clinical consequences of these errors?

Response:

Thank you for this valuable comment. We agree that addressing the limitations of generative AI particularly in real-world clinical settings is essential for providing a balanced and comprehensive perspective.

In response, we have revised the manuscript to include a new subsection critically examining key limitations, with a specific focus on the issue of hallucinations. These points are now covered in the revised “Challenges and Limitations” section.

Modification done are highlighted on page no.35

2) While the review mentions regulatory frameworks (e.g., FDA, EMA), it does not delve into specific guidelines or standards for validating generative AI tools in healthcare. Could the authors provide more details on current regulatory pathways or gaps?

Response:

We appreciate the reviewer’s insightful comment highlighting the need to elaborate on regulatory pathways and standards for generative AI (GenAI) validation in healthcare.

In response, we have added a paragraph, discussing the current regulatory landscape, particularly focusing on agencies such as the U.S. Food and Drug Administration (FDA), the European Medicines Agency (EMA), and relevant global efforts. This addition strengthens the manuscript’s discussion on implementation challenges by offering clarity on evolving compliance requirements and the pressing need for dedicated GenAI-specific regulatory frameworks..

Modification done are highlighted on page no.37

3) The section on bias and health equity is brief. Given the importance of this issue, could the authors expand on strategies to mitigate bias in AI models, such as dataset diversification or algorithmic fairness techniques?

Response:

We thank the reviewer for highlighting the need to further elaborate on bias mitigation strategies in generative AI models.

In response, we have significantly expanded the section on bias and health equity to provide a more comprehensive discussion. Specifically, we have added details on practical mitigation strategies, including dataset diversification, algorithmic fairness frameworks, synthetic data augmentation, bias audits, and post-deployment monitoring. These additions reflect current research and industry practices aimed at promoting equitable AI development and deployment.

Modification done are highlighted on page no.36

4) The paper cites numerous studies but does not systematically compare the performance of generative AI with traditional methods or human experts in key areas like diagnostics or drug discovery. A meta-analysis or comparative table would strengthen this aspect.

Response:

We thank the reviewer for this insightful comment. To enhance the manuscript’s analytical rigor, we have added a comparative paragraph that reflects on the performance of generative AI relative to traditional methods and human experts in key domains such as diagnostics and drug discovery.

Table no. 3 already includes benchmarking outcomes from recent studies where large language models (e.g., GPT-4) demonstrated diagnostic accuracy comparable to or exceeding that of senior clinicians in complex case scenarios.

A brief comparative overview has been added as a subsection in the clinical application to summarize these performance differentials, supported by select studies. We agree that future meta-analytical work across datasets and specialties would be valuable, and we have noted this as a recommended direction for further research.

Modification done are highlighted on page no.33

 

5) The discussion on future directions is speculative. Could the authors ground their predictions with concrete examples of ongoing trials or technologies nearing clinical implementation?

Response:

We sincerely thank the reviewer for this insightful suggestion. In response, we have revised the “Future Directions” section to include concrete examples of ongoing clinical trials, pilot programs, and near-term technologies currently under development or regulatory review. Inclusion of examples better contextualize the future landscape with evidence-based developments.

Modification done are highlighted on page no.39, 40

 

6) Some references are outdated (e.g., 2022–2023). Given the rapid evolution of AI, could the authors include more recent studies (2024–2025) to ensure the review reflects the latest advancements?

Response:

We thank the reviewer for highlighting the need for more up-to-date references to reflect the rapidly evolving landscape of generative AI in healthcare.

In response, we have already reviewed the recent literature and incorporated almost 80% references from 2024 and early 2025 publications throughout the manuscript. These additions include updates on clinical applications of GPT-4, multimodal LLMs in diagnostics, ongoing regulatory discussions, and emerging implementation studies. The references are from 2022–2023 are relevant and perquisite for the manuscript.

By integrating most of newer studies, we aim to ensure that the review captures the latest technological advances and policy developments, thereby enhancing its relevance and timeliness for the readership.

7) The figures (e.g., Figure 4) are descriptive but could benefit from annotations or captions explaining their relevance to the text. For example, how does Figure 4 illustrate the integration of AI in clinical decision support?

Response:

We appreciate the reviewer’s valuable suggestion regarding the need for clearer linkage between Figure 4 and the manuscript text.

To address this, we have revised the figure caption to explicitly describe how it demonstrates the integration of AI-based Clinical Decision Support Systems (AI CDSS) across multiple stages of clinical care including diagnosis, treatment planning, bedside monitoring, and emergency triage.

Modification done are highlighted on page no.13

 

8) The paper does not address cost-effectiveness or scalability of generative AI solutions. How do these technologies compare in terms of implementation costs and resource requirements across different healthcare systems?

Response:

We appreciate the reviewer’s valuable comment regarding the omission of cost-effectiveness and scalability considerations. In response, we have expanded the manuscript to include a dedicated discussion on these important aspects. The revised section now outlines the key drivers of implementation cost (e.g., infrastructure, model training, compliance, and workforce integration) and examines scalability in both high-resource and low-resource healthcare settings.

Modification done are highlighted on page no.38

 

Reviewer 5 Report

Comments and Suggestions for Authors

In this work, the authos explores the transformative potential of generative artificial intelligence (AI) in healthcare, focusing particularly on Large Language Models (LLMs) and image-generating systems. They illustrate key applications across clinical documentation, diagnostics, patient communication, drug discovery, and medical education. While early evidence suggests significant benefits—such as improved efficiency, enhanced diagnostic accuracy, and reduced administrative burden—the review emphasizes the need for rigorous validation, regulatory oversight, and ethical safeguards.

They also pose the challenges of integrating generative AI into real-world clinical settings, including risks of misinformation, bias, and patient privacy concerns and reflect on tits use for augmenting healthcare professionals capabilities, with the potential to drive a shift toward more proactive, personalized, and efficient care.

The paper is well-written and I want to suggest some points to be discussed.

• Generative AI is producing an extreme consumption of resources. Despite, its contribution to improve health population, it is evident that has a direct impact on other sustainable development goals. This has been already exposed respect to climate change. I suggest the authors to include a reflection in this line but in the context of health with suitable references.
• As it has been spoken of a co-scientist, what can be guidelines for using it as a co-doctor or assistant? This goes even beyond that enhance doctors capacilities.
• As we have said, it is well-known that LLM's permit to explore large amounts of scientific literature and can help to designed medical research. I suggest how to be considered for designing medical research agendas, as it has been recently reported concerning ERC and NSF revealing unbalanced domains.
• LLMs are a powerful for analyzing massive amounts of data and study trade-offs between potential treatments and seondary effects. I would suggest the authors to include some reflections in this line.
• What can explainability of machine learnign methods can provide in contrast with statistical methods?

 

Author Response

 

Reviewers’ Comments

 

1) Generative AI is producing an extreme consumption of resources. Despite, its contribution to improve health population, it is evident that has a direct impact on other sustainable development goals. This has been already exposed respect to climate change. I suggest the authors to include a reflection in this line but in the context of health with suitable references.

Response:

We sincerely thank the reviewer for this insightful and timely comment. We agree that while generative AI holds significant potential to transform healthcare delivery and improve population health outcomes, it is essential to evaluate its broader implications particularly concerning sustainability and environmental impact.

In response, we have added a reflective subsection in the discussion highlighting the intersection of generative AI in healthcare and its alignment with Sustainable Development Goals (SDGs): Climate change.

Modifications done are highlighted on Page no. 36

2) As it has been spoken of a co-scientist, what can be guidelines for using it as a co-doctor or assistant? This goes even beyond that enhance doctors capabilities.

Response:

We thank the reviewer for this thought-provoking and forward-looking comment. The evolving role of generative AI in healthcare is beginning to transcend its earlier characterization as a "co-scientist" and is now venturing into the domain of a “co-doctor” or intelligent clinical assistant with potential to augment clinical decision-making, streamline patient communication, and support administrative duties.

In response, we have included a new subsection in the manuscript discussing the emerging paradigm of generative AI as a clinical copilot or augmented intelligence system that partners with physicians, rather than replaces them. We highlight proposed guidelines and ethical principles for such integration, which are crucial to ensure safe, effective, and equitable deployment. We have incorporated these perspectives in the Future directions and believe they add conceptual depth to the manuscript's future-oriented narrative.

Modifications done are highlighted on Page no. 37

3). It is well-known that LLM's permit to explore large amounts of scientific literature and can help to designed medical research. I suggest how to be considered for designing medical research agendas, as it has been recently reported concerning ERC and NSF revealing unbalanced domains.

Response:

We thank the reviewer for this important and forward-thinking observation. We fully agree that large language models (LLMs) have significant potential not only in analyzing existing medical literature but also in supporting the formulation of medical research agendas.

In response to the reviewer’s suggestion, we have added a paragraph in the revised manuscript discussing how LLMs could be harnessed for meta-research applications particularly in assisting the design of research strategies and balancing resource allocation across disciplines. This addition broadens the scope of our manuscript by highlighting the strategic and policy-level role of generative AI in healthcare research.

Modifications done are highlighted on Page no. 28

4). LLMs are a powerful for analyzing massive amounts of data and study trade-offs between potential treatments and secondary effects. I would suggest the authors to include some reflections in this line.

Response:

We sincerely thank the reviewer for highlighting this important dimension of LLM utility. We agree that one of the significant strengths of large language models lies in their capacity to analyze vast, heterogeneous datasets including clinical trial reports, real-world evidence, electronic health records (EHRs), and pharmacovigilance databases to uncover patterns and support nuanced decision-making in healthcare.

In response to the reviewer’s suggestion, we have incorporated a dedicated discussion on how LLMs can be employed to study trade-offs between therapeutic benefits and secondary or adverse effects.

Modifications done are highlighted on Page no. 18, 19

5).What can explainability of machine learning methods can provide in contrast with statistical methods?

Response:

We thank the reviewer for this excellent observation regarding the explainability of machine learning (ML) methods as compared to traditional statistical approaches.

In the revised manuscript, we have included a reflective comparison between these two paradigms.

Modifications done are highlighted on Page no. 33

 

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for this opportunity to review your manuscript again. 

However, I do not think the authors addressed my comments successfully. 

One very simple and obvious thing is the comment in the manuscript. The authors claimed that they reviewed the entire manuscript and addressed this issue. But it is still there. 

Another very critical issue is the methodology of the research. Although the authors added one paragraph, that is not sufficient as a section of methodology. The authors may want to check other literature review papers to see the details of this method. 

Reviewer 4 Report

Comments and Suggestions for Authors

I have no further comments.

Reviewer 5 Report

Comments and Suggestions for Authors

The paper can be accepted for publication.