PREreview of How to Improve Mental Health in the Older Adults Through AI‐Enhanced Physical Activity: An Emerging Research Topic

Summary and Overall Impression

This is a timely and important piece as it asks a significant, relevant question: how can AI help older adults stay mentally healthy through physical activity? With populations aging fast and mental health needs growing, the topic couldn't be more needed. The authors clearly put in a lot of effort, pulling together nearly 2,000 papers and using detailed bibliometric tools to map out what's happening in the research world.

Despite the author's robust dataset and methods, the paper's delivery needs significant improvement. The paper would benefit more a more streamline structure that guides the reader through the logic of the research, clearer summaries at the end of each results section to explain why the findings matter, less repetition of terms and phrases, with more thoughtful reflection on their implications and stronger interpretation and framing to show how the findings connect to real-world needs and future opportunities.

Abstract

Major issue: The abstract hits the main ideas, but it's trying to do too much. It's packed with tool names and technical terms that may be impressive to bibliometric experts but aren't helpful for a broader audience, especially health or tech professionals who could actually apply these findings.

Suggestions:

• Begin with the problem: the growing mental health burden in aging populations.

• Briefly describe the approach and main findings in plain language.

• Highlight key outcomes (e.g., emerging intervention areas, top countries, key mechanisms like BDNF and resilience) without overloading on technical terms.

Introduction

Major Issue: The introduction is fragmented and hard to follow.

Suggestion: Walk the reader through three clear steps:

• Why older adults' mental health needs attention.

• Why physical activity is a solid solution.

• Why AI might make that solution more accessible, personalised, or sustainable.

Then clearly state no one's really connected all three before in a single review, and this paper does that.

Methods:

Major issue:

• The Methods section includes too much technical detail, like software versions, node sizes, and clustering parameters.

• The rationale for using tools like CiteSpace and VOSviewer is unclear. The authors don't explain why these tools were chosen or what patterns they aimed to uncover through keyword co-occurrence.

• The inclusion and exclusion criteria for selecting studies are not well-defined. It's unclear if there were quality checks, language filters, or manual screening steps.

Suggestion:

• Reduce the level of technical detail (e.g., software versions, node sizes, clustering parameters)

and relocate it to an appendix or supplementary section.

• Clarify the rationale for selecting bibliometric tools such as CiteSpace and VOSviewer, including the specific patterns or relationships the authors intended to identify.

• Emphasize and reference the visual workflow (Figure 1) earlier in the Methods section to help readers understand the analytical process.

• Specify the inclusion and exclusion criteria more clearly—such as quality thresholds, language limitations, and any manual screening procedures used, to strengthen transparency and trust in the dataset.

Results

Major issue: The visuals are strong, but the text is overly descriptive and lacks interpretation. It lists data without explaining its significance, making it hard to understand key takeaways.

Suggestion:

• After each sub-section (like institutions, countries, keywords), pause and summarise the meaning. Say something like: "This tells us that China is catching up in AI-driven elderly health research" or "The growing interest in cardiovascular outcomes suggests the field is expanding beyond just mental health."

• Add interpretation. For example, the preprint says, U.S. is leading. Is that due to more funding? A bigger aging population? Less stigma? These types of reflective questions make the data feel alive. 

• Also, try to avoid repeating the same phrases across sections. "BDNF," "resilience," and "cognitive health" pop up often with similar language. If something's important enough to mention multiple times, try offering a new angle or implication each time.

Discussion

Major issue: The discussion touches on important health outcomes, but the points are disjointed. It lacks a clear narrative connecting AI to mental, cognitive, cardiovascular, and bone health in a coherent way.

Suggestions:

• Connect the dots: How does better bone health affect mental health? Could increased mobility from AI-supported activity improve social participation, boosting mood?

• Also, show real-world roadblocks. Who's being left out of this tech-driven movement? What about older adults without smartphones, or with limited digital literacy? Those gaps need to be acknowledged here.

• A lot of the writing is optimistic, which is great, but some realism around access, ethics, and tech fatigue would make it feel more grounded.

Conclusion

Minor issue: The conclusion repeats a lot from earlier sections but doesn't quite land the plane. The reader should finish knowing what we should do with this. Right now, it's vague.

Suggestion:

• End with 2–3 concrete calls to action. For example:

• Fund more real-world trials of AI-powered exercise interventions.

• Build cross-disciplinary teams to design tech that works for older bodies and minds.

• Invest in digital literacy programs for older adults so they're not left behind.

• In the "future" section, instead of saying more research is needed, say of what kind? Are wearables the best bet? Virtual reality? Personalized coaching bots?

Final Notes:

The authors gathered an impressive dataset and built a map that could guide the future of AI and aging. But to get there, they must bring out the narrative more clearly. The findings are solid right now, but how they're delivered hides the full value. Focus on simplifying, connecting the dots, and being a little more reflective about what all this data really means for people, especially the older adults this research is intended to help. With those changes, this could go from a dense reference list to a foundational piece for the field.

Competing interests

The author declares that they have no competing interests.