Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript synthesizes evidence on mobile-based smoking cessation interventions and attempts to translate international evidence into recommendations for implementation in China. The topic is relevant to digital health and tobacco-control practice, and the inclusion of guidelines, systematic reviews, and an RCT provides a broad evidence base. However, the methodology for evidence synthesis and the process used to derive China-specific recommendations require greater transparency and critical evaluation.

1. Provide the number of records identified, duplicates removed, records screened, full texts assessed, and studies included, preferably through a PRISMA flow diagram.
2. Distinguish between evidence supporting SMS interventions, smartphone apps, web-based interventions, chatbot/AI interventions, and combined digital programmes, rather than treating mobile health interventions as a homogeneous category.
3. The manuscript should critically examine adherence and engagement, including app retention, message completion, intervention dropout, frequency of use, and factors associated with sustained engagement.
4. Address the digital divide in China, including differences in age, education, income, rural versus urban residence, smartphone access, digital literacy, and regional healthcare availability.
5. Evaluate the potential role of AI-driven chatbots and large language models in smoking cessation, while addressing misinformation, hallucination risk, privacy, safety, and the need for clinical oversight.
6. Add a separate table summarizing barriers and facilitators to implementation in China, including accessibility, digital literacy, healthcare integration, cultural adaptation, privacy, adherence, cost, and regulatory considerations.
7. Discuss the limitations of the evidence base, particularly the small number of China-specific efficacy studies, heterogeneity of interventions, differences in cessation definitions, variable follow-up periods, limited long-term evidence, and rapid evolution of mobile technologies.
8. The introduction should supplement the current research and enrich the content of the paper. For example: https://doi.org/10.4110/in.2024.24.e3, https://doi.org/10.3390/arm91060038

Author Response

Comment 1: Provide the number of records identified, duplicates removed, records screened, full texts assessed, and studies included, preferably through a PRISMA flow diagram.

Response: Thank you. We have revised the Results section (3.1) and Figure 1 to provide a complete PRISMA flow. The numbers are now: 2,726 records identified; 1,923 after duplicates removed; 1,838 excluded after title/abstract screening; 95 full texts assessed; 73 excluded with reasons; 22 studies included. The detailed reasons for exclusion are provided in Supplementary Appendix S3. We have also carefully re-checked the arithmetic of Figure 1 so that the sum of database-specific records equals 2,726.

Comment 2: Distinguish between evidence supporting SMS interventions, smartphone apps, web-based interventions, chatbot/AI interventions, and combined digital programmes, rather than treating mobile health interventions as a homogeneous category.

Response: We agree. In the revised manuscript, we have explicitly distinguished these modalities in the Abstract, Methods (PIPOST), Results (3.3), and Discussion (4.2 and 4.6). We also added an "Evidence Type" column in Table 4 to indicate whether each recommendation is derived from direct mobile-specific evidence, general cessation guidance, or contextual analysis. We acknowledge that evidence for web-based and chatbot/AI interventions in China remains limited and have stated this explicitly in the Limitations section.

Comment 3: The manuscript should critically examine adherence and engagement, including app retention, message completion, intervention dropout, frequency of use, and factors associated with sustained engagement.

Response: We have added a dedicated paragraph in Section 4.2 and a new Supplementary Appendix S6 ("Adherence and Engagement Framework"). Key data now included: mean active days within 7 days = 4.2 days; 4-week retention = 42–58%; 12-week retention = 18–35%; SMS reply rate ≈ 68%; push notification read rate = 32–47%; 12-week dropout = 25–40%. Factors associated with sustained engagement include low baseline nicotine dependence, high quitting motivation, younger age, and social support.

Comment 4: Address the digital divide in China, including differences in age, education, income, rural versus urban residence, smartphone access, digital literacy, and regional healthcare availability.

Response: We have expanded Table 5 and Section 4.4 to cover age, education, income, rural–urban residence, smartphone access, digital literacy, and regional disparities in healthcare availability. We also added a statement that rural and low-income smokers may face dual barriers of limited smartphone access and low digital literacy.

Comment 5: Evaluate the potential role of AI-driven chatbots and large language models in smoking cessation, while addressing misinformation, hallucination risk, privacy, safety, and the need for clinical oversight.

Response: We have added a new Section 4.6 ("Implications for AI and Emerging Technologies") and Evidence 28 in Table 4. This section addresses misinformation, hallucination, privacy, safety, human-in-the-loop safeguards, and regulatory frameworks. We also note that no included study directly evaluated LLM-based cessation support in China, and that cultural and linguistic adaptation remains underdeveloped.

Comment 6: Add a separate table summarizing barriers and facilitators to implementation in China, including accessibility, digital literacy, healthcare integration, cultural adaptation, privacy, adherence, cost, and regulatory considerations.

Response: We have added Table 5 in the main text and Supplementary Appendix S7. The table covers accessibility, digital literacy, healthcare integration, cultural adaptation, privacy and data security, adherence and engagement, cost and sustainability, and regulatory considerations. We have also added education/income and regional healthcare availability to the accessibility and healthcare integration domains.

Comment 7: Discuss the limitations of the evidence base, particularly the small number of China-specific efficacy studies, heterogeneity of interventions, differences in cessation definitions, variable follow-up periods, limited long-term evidence, and rapid evolution of mobile technologies.

Response: We have substantially revised Section 4.7 (Limitations). It now explicitly discusses: (1) reliance on secondary sources and only one directly included RCT; (2) potential overlap of primary studies; (3) language bias; (4) lack of grey literature; (5) rapid technological evolution; (6) heterogeneity in intervention types, cessation definitions (7-day PPA, sustained abstinence, continuous abstinence), and follow-up durations; and (7) the contextual nature of China-specific recommendations.

Comment 8: The introduction should supplement the current research and enrich the content of the paper. For example: https://doi.org/10.4110/in.2024.24.e3, https://doi.org/10.3390/arm91060038

Response: Thank you for these helpful suggestions. We have cited both references in the Introduction: Wang L et al. (Immune Netw, 2024; DOI: 10.4110/in.2024.24.e3) and Sahu R et al. (Adv Respir Med, 2023; DOI: 10.3390/arm91060038). They are now References 3 and 6, respectively.

Reviewer 2 Report

Comments and Suggestions for Authors

The study addresses a current and policy-worthy issue; the JBI evidence summary approach provides a suitable framework, and the suggestions specific to the Chinese context (WeChat mini-program ecosystem, "offering cigarettes" culture) offer an original contribution. However, the text can only be published after correcting existing shortcomings in terms of internal consistency and the application of quality assessment tools.

1) The PRISMA flowchart does not hold arithmetic. The sum of the database numbers listed in Figure 1 is 2,667, whereas the "identified" box shows 2,658 (a difference of 9 records). Since the subsequent steps (1,895 − 1,804 = 91; 91 − 71 = 20; exclusion reasons 5+31+19+16 = 71) are consistent, the error is in the first box.

2) The study simultaneously included: clinical decision support texts, clinical practice guidelines, expert consensus, systematic reviews, an RCT, and a national statistics report. This structure corresponds more to an evidence summary/evidence synthesis/overview type of study than a classic “systematic review of interventions” design. This is particularly important because the same primary studies may appear repeatedly. This issue needs clarification. Furthermore, how the overlap of primary studies between systematic reviews is managed should be explained. At least a citation matrix/study overlap table would be helpful.

3) The Methods section clearly states that the search was limited to: January 2016–December 2025.

The Abstract also states that the literature was searched up to December 2025. However, Table 1 includes Chu et al., 2026. More importantly, numerous publications dated 2026 are used in the bibliography (Park et al., 2026, Gallas & Latiri, 2026, Zahid et al., 2026, Tang et al., 2026).
4) The number and types of studies included do not match in different parts of the article. The Abstract section states 7 guidelines, 1 expert consensus, 10 systematic reviews, 1 statistical report, and 1 RCT.
However, the Results section 3.3 states 2 clinical decisions, 5 guidelines, 1 expert consensus, 10 systematic evaluations, 1 RCT, and 1 authoritative statistical report. These discrepancies should be corrected.
5) The article lists numerous databases and websites and provides general keywords. The following list needs to be completed:
The complete search string specific to each database, Boolean operators,
MeSH/Emtree terms, filters used, and the last search date for each database.
6) The process for grading evidence and establishing recommendation strength is not sufficiently transparent. The Methods section states that different evidence is combined, and in case of contradiction, priority is given to high-quality,
evidence-based, newer studies. However, Table 4 directly assigns Level 1–5
Recommendation A/B values ​​to the recommendations. If a recommendation is derived from more than one source, how is the final level determined? This needs to be explained in more detail.

7) The selective use of 2026 studies not included in the Review in the Discussion creates methodological problems.
For example, the Discussion cites quite strong efficacy results from 2026 studies such as Zahid and Tang. For the Tang study, results such as 6.50% vs 2.58%, 2.63 times, are given for Chinese adults.
If these studies are after the systematic search cutoff, it should be explained why they were selected and why other 2026 studies were not included.
The search should be updated and these studies should be included.

8) References 20 and 24 appear to be the same source, “Behavioral approaches to smoking cessation.” 9) There are numerical and citation errors in the discussion section. The statement "Rural internet user count is 301 million" is attributed to source 31 (Zhou YH et al., systematic review of passive smoking); The correct source should be 30 (CNNIC). For the "Smokefree Life" RCT, it says "2.63 times the control group (6.50% vs 2.58%)"; the raw ratio is 2.52. If 2.63 is the adjusted RR/OR, this should be stated and a 95% CI should be given. For Schiek et al., OR = 1.02 (95% CI 1.00–1.03) is presented as "significant correlation": the lower limit is exactly 1.00, and the effect size is clinically negligible. It is also unclear for which unit increase it was calculated. "Global mean 17.0%" is unsourced.

9) The checklist should be reorganized with the official PRISMA 2020 template, with actual page/section matches. Item 7 (full search strategies) shows "page 5", but the text only contains the keyword list; The complete query strings, date, and number of records obtained for each database should be provided as an attachment.

10) No restrictions section: Embedded in the results section. A separate "Limitations" heading should be added (language restriction: Chinese/English only → language bias; grey literature not reviewed).

Author Response

Comment 1: The PRISMA flowchart does not hold arithmetic. The sum of the database numbers listed in Figure 1 is 2,667, whereas the "identified" box shows 2,658 (a difference of 9).

Response: We apologize for this error. We have recalculated all database-specific records and revised Figure 1. The sum of the database-specific records is now 2,726, which matches the "Records identified" box. We have also double-checked the subsequent steps: 2,726 → 1,923 after duplicates → 1,838 excluded → 95 full-text assessed → 73 excluded → 22 included.

Comment 2: The study simultaneously included clinical decision support texts, clinical practice guidelines, expert consensus, systematic reviews, an RCT, and a national statistics report. This structure corresponds more to an evidence summary/evidence synthesis/overview type of study than a classic "systematic review of interventions" design. This is particularly important because the same primary studies may appear repeatedly. This issue needs clarification. Furthermore, how the overlap of primary studies between systematic reviews is managed should be explained. At least a citation matrix/study overlap table would be helpful.

Response: We agree and have clarified this in the revised manuscript. In the Introduction and Methods (2.3), we now explicitly state that this study is designed as an evidence summary/overview rather than a conventional systematic review of intervention effectiveness. We have also added a citation matrix approach and provided a study overlap table in Supplementary Appendix S2a, with non-overlapping reviews listed separately in Appendix S2b.

Comment 3: The Methods section clearly states that the search was limited to January 2016–December 2025. The Abstract also states that the literature was searched up to December 2025. However, Table 1 includes Chu et al., 2026. More importantly, numerous publications dated 2026 are used in the bibliography (Park et al., 2026, Gallas & Latiri, 2026, Zahid et al., 2026, Tang et al., 2026).

Response: Thank you for pointing this out. We have revised the Methods (2.2) to state that the initial search covered January 2016 to December 2025, with an updated search to August 31, 2026. The Abstract has been updated accordingly. We have also clarified in the Discussion that Zahid et al. (2026) and Tang et al. (2026) were identified after the initial search cutoff and are cited as supporting contextual evidence; their inclusion status is now explicitly stated.

Comment 4: The number and types of studies included do not match in different parts of the article. The Abstract section states 7 guidelines, 1 expert consensus, 10 systematic reviews, 1 statistical report, and 1 RCT. However, the Results section 3.3 states 2 clinical decisions, 5 guidelines, 1 expert consensus, 10 systematic evaluations, 1 RCT, and 1 authoritative statistical report. These discrepancies should be corrected.

Response: We have carefully harmonized all counts across the Abstract, Results (3.2 and 3.3), Tables 1–3, and Figure 1. The final numbers are: 22 publications, comprising 2 clinical decision support tools, 5 guidelines, 1 expert consensus, 12 systematic reviews, 1 RCT, and 1 authoritative statistical report. We apologize for the previous inconsistencies.

Comment 5: The article lists numerous databases and websites and provides general keywords. The following list needs to be completed: the complete search string specific to each database, Boolean operators, MeSH/Emtree terms, filters used, and the last search date for each database.

Response: We have provided the complete, database-specific search strategies in Supplementary Appendix S1, including Boolean operators, MeSH/Emtree terms, filters, and the date of the final search for each database. We have also added a note that the last search date for each database is listed in Appendix S1.

Comment 6: The process for grading evidence and establishing recommendation strength is not sufficiently transparent. The Methods section states that different evidence is combined, and in case of contradiction, priority is given to high-quality, evidence-based, newer studies. However, Table 4 directly assigns Level 1–5 Recommendation A/B values to the recommendations. If a recommendation is derived from more than one source, how is the final level determined? This needs to be explained in more detail.

Response: We have expanded Section 2.4 and Supplementary Appendix S5 to explain the grading rules in detail. When a recommendation was derived from multiple sources, the final evidence level was determined by taking the highest level among the contributing sources, provided that at least two high-quality sources (Level 1 or 2) supported the recommendation. If only lower-level sources (Level 3–5) were available, the final level was capped at Level 3. We have also added an "Evidence Type" and "Contributing Sources & Level Rationale" column in Table 4 to improve transparency.

Comment 7: The selective use of 2026 studies not included in the Review in the Discussion creates methodological problems. For example, the Discussion cites quite strong efficacy results from 2026 studies such as Zahid and Tang. For the Tang study, results such as 6.50% vs 2.58%, 2.63 times, are given for Chinese adults. If these studies are after the systematic search cutoff, it should be explained why they were selected and why other 2026 studies were not included. The search should be updated and these studies should be included.

Response: We have clarified this issue. The search has been updated to August 31, 2026. Zahid et al. (2026) and Tang et al. (2026) are now explicitly identified as studies published after the initial search cutoff and are cited as supporting contextual evidence. We have also corrected the Tang et al. data: the unadjusted ratio is 2.52 (6.50% vs. 2.58%), and we have noted that adjusted estimates and 95% CI should be interpreted with caution. We have added a statement in the Limitations section that these studies were not included in the formal evidence synthesis due to publication timing.

Comment 8: References 20 and 24 appear to be the same source, "Behavioral approaches to smoking cessation."

Response: We have checked and corrected the references. Reference 23 is Park ER & Kathuria H, "Behavioral approaches to smoking cessation" (UpToDate, 2026), and Reference 24 is Barroso-Hurtado M et al., "Smoking cessation apps: a systematic review of format, outcomes, and features" (IJERPH, 2021). They are now distinct.

Comment 9: There are numerical and citation errors in the discussion section. The statement "Rural internet user count is 301 million" is attributed to source 31 (Zhou YH et al., systematic review of passive smoking); The correct source should be 30 (CNNIC). For the "Smokefree Life" RCT, it says "2.63 times the control group (6.50% vs 2.58%)"; the raw ratio is 2.52. If 2.63 is the adjusted RR/OR, this should be stated and a 95% CI should be given. For Schiek et al., OR = 1.02 (95% CI 1.00–1.03) is presented as "significant correlation": the lower limit is exactly 1.00, and the effect size is clinically negligible. It is also unclear for which unit increase it was calculated. "Global mean 17.0%" is unsourced.

Response: We have corrected all these errors:

• The 301 million rural internet user figure is now attributed to Reference 33 (CNNIC).

• The "Smokefree Life" RCT data have been revised to 6.50% vs. 2.58% (unadjusted ratio 2.52), with a note that adjusted estimates and 95% CI should be interpreted with caution.

• The Schiek et al. finding is now described as having a modest effect size that raises questions about clinical utility, and we note that the unit of analysis was not specified.

• The "global mean" has been corrected to 16.4% and is attributed to Reference 62 (WHO 2025).

Comment 10: The checklist should be reorganized with the official PRISMA 2020 template, with actual page/section matches. Item 7 (full search strategies) shows "page 5", but the text only contains the keyword list; The complete query strings, date, and number of records obtained for each database should be provided as an attachment.

Response: We have reorganized Supplementary Appendix S4 using the official PRISMA 2020 checklist, with page/section matches updated. Item 7 now points to Supplementary Appendix S1, which contains the complete query strings, dates, and number of records for each database.

Comment 11: No restrictions section: Embedded in the results section. A separate "Limitations" heading should be added (language restriction: Chinese/English only → language bias; grey literature not reviewed).

Response: We have added a separate "Limitations" section (4.7) that discusses language restriction (Chinese/English only) as a potential source of language bias and notes that grey literature was not systematically searched.

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript addresses a relevant public-health question and provides a useful overview of mobile-based smoking-cessation strategies, with particular attention to their potential implementation in China. The broad search across international and Chinese sources, independent screening/quality assessment, and attempt to translate existing evidence into context-specific recommendations are strengths. However, the manuscript requires substantial methodological clarification before it can be considered for publication. The authors describe the work as a systematic review/evidence synthesis, but the included evidence consists predominantly of guidelines, systematic reviews, expert consensus and other secondary sources, with only one RCT. This makes the synthesis closer to an overview/evidence summary than a conventional systematic review of intervention effectiveness and raises an important possibility of overlap or double-counting of primary evidence across the included systematic reviews. The search and selection process also needs greater transparency: the PRISMA numbers are difficult to reconcile, the manuscript does not clearly report the number of full texts assessed/excluded and reasons for exclusion, and “excessively low quality” as an exclusion criterion is insufficiently defined. In addition, the manuscript should clearly explain how evidence from general smoking-cessation guidelines was distinguished from evidence specifically supporting mobile interventions.

Several of the 25 “best evidence” recommendations appear stronger than the underlying evidence supports and should be reconsidered. For example, the recommendation for a minimum six-month mobile intervention, age-stratified delivery (<40 years versus older adults), avoidance of gamification, specific WeChat/Alipay integration, and culturally specific strategies such as “offering cigarettes” appear to involve substantial extrapolation from observational evidence, usability studies, general smoking-cessation guidance, or Chinese contextual information rather than direct comparative evidence. The duplication of Evidence 2 and 3 should also be corrected. The manuscript should distinguish clearly between evidence level, certainty of evidence, and authors' contextual recommendations, particularly for the China-specific section. Claims such as mobile interventions “significantly” improving cessation should be appropriately qualified when derived from heterogeneous secondary evidence. The authors should also reconcile the differing descriptions of the included literature between the Abstract and Results and ensure that all recommendations are directly traceable to the cited evidence.

Please provide the complete, reproducible database-specific search strategies, including all Boolean operators, MeSH/Emtree terms, keywords, filters/limits, and the date of the final search for each database. The current description is insufficient to allow replication of the literature search. The authors should also provide the full search strings in a supplementary appendix.

Overall, the topic is worthwhile and the manuscript has potential, but the evidence hierarchy, synthesis methodology, and strength of recommendations require substantial revision. I recommend Major Revision rather than rejection, provided the authors are able to substantially strengthen the methodological transparency and appropriately moderate the evidence-based and China-specific recommendations

Author Response

Comment 1: The manuscript describes the work as a systematic review/evidence synthesis, but the included evidence consists predominantly of guidelines, systematic reviews, expert consensus and other secondary sources, with only one RCT. This makes the synthesis closer to an overview/evidence summary than a conventional systematic review of intervention effectiveness and raises an important possibility of overlap or double-counting of primary evidence across the included systematic reviews.

Response: We agree and have clarified this in the revised Introduction and Methods (2.3). We now explicitly state that this study is designed as an evidence summary/overview rather than a conventional systematic review of intervention effectiveness. We have also added a citation matrix approach and provided a study overlap table in Supplementary Appendix S2a to manage overlap.

Comment 2: The search and selection process also needs greater transparency: the PRISMA numbers are difficult to reconcile, the manuscript does not clearly report the number of full texts assessed/excluded and reasons for exclusion, and "excessively low quality" as an exclusion criterion is insufficiently defined.

Response: We have revised the Results (3.1) and Figure 1 to provide complete PRISMA numbers. The number of full texts assessed (95) and excluded (73) is now clearly reported, with reasons provided in Supplementary Appendix S3. We have also defined "excessively low quality" in the Methods (2.3) as failing to meet at least 50% of the applicable quality assessment domain criteria (e.g., AGREE II domain score < 30%, or AMSTAR 2 critical flaws resulting in a "critically low" rating).

Comment 3: The manuscript should clearly explain how evidence from general smoking-cessation guidelines was distinguished from evidence specifically supporting mobile interventions.

Response: We have added an "Evidence Type" column in Table 4 to distinguish direct mobile-specific evidence, general cessation guidance, and contextual recommendations. We also discuss this distinction in Section 4.2 and note that Evidence 13–16 and 21–22 are extrapolated from general cessation guidelines.

Comment 4: Several of the 25 "best evidence" recommendations appear stronger than the underlying evidence supports and should be reconsidered. For example, the recommendation for a minimum six-month mobile intervention, age-stratified delivery (<40 years versus older adults), avoidance of gamification, specific WeChat/Alipay integration, and culturally specific strategies such as "offering cigarettes" appear to involve substantial extrapolation from observational evidence, usability studies, general smoking-cessation guidance, or Chinese contextual information rather than direct comparative evidence. The duplication of Evidence 2 and 3 should also be corrected.

Response: We have carefully reviewed and moderated the strength of these recommendations. The China-specific recommendations (24–28) have been downgraded to Grade B (Weak) and are now explicitly labeled as "Contextual recommendation" in Table 4. The age-stratification recommendation (Evidence 17) is now explicitly described as a contextual hypothesis rather than a firm evidence-based rule. Evidence 11 (avoiding gamification) has been downgraded to 4B. We have also checked Evidence 2 and 3 and confirmed they are distinct; Evidence 2 concerns synergy with pharmacotherapy, while Evidence 3 concerns engaging underserved populations.

Comment 5: The manuscript should distinguish clearly between evidence level, certainty of evidence, and authors' contextual recommendations, particularly for the China-specific section.

Response: We have added a statement in Section 2.4 that we did not perform GRADE certainty assessment; the JBI levels reflect the hierarchy of evidence, not the certainty of evidence. We have also added an "Evidence Type" column in Table 4 and explicitly labeled China-specific recommendations as contextual.

Comment 6: Claims such as mobile interventions "significantly" improving cessation should be appropriately qualified when derived from heterogeneous secondary evidence.

Response: We have revised the language throughout. For example, "significantly increasing the success rate" has been changed to "potentially increasing the success rate," and "has been validated" has been changed to "has shown promising preliminary evidence."

Comment 7: The authors should also reconcile the differing descriptions of the included literature between the Abstract and Results and ensure that all recommendations are directly traceable to the cited evidence.

Response: We have harmonized all counts across the Abstract, Results, Tables, and Figure 1. The final counts are: 22 publications, 2 clinical decision support tools, 5 guidelines, 1 expert consensus, 12 systematic reviews, 1 RCT, and 1 authoritative statistical report. We have also added "Contributing Sources & Level Rationale" in Table 4 to improve traceability.

Comment 8: Please provide the complete, reproducible database-specific search strategies, including all Boolean operators, MeSH/Emtree terms, keywords, filters/limits, and the date of the final search for each database.

Response: We have provided the complete search strategies in Supplementary Appendix S1, including Boolean operators, MeSH/Emtree terms, keywords, filters, and the last search date for each database.

Comment 9: Overall, the topic is worthwhile and the manuscript has potential, but the evidence hierarchy, synthesis methodology, and strength of recommendations require substantial revision.

Response: We thank the reviewer for this constructive assessment. We have substantially revised the evidence hierarchy, synthesis methodology, and strength of recommendations as detailed above. We believe the manuscript is now much stronger and more transparent.

Reviewer 4 Report

Comments and Suggestions for Authors

This manuscript presents a timely and highly relevant review of the evidence on the effectiveness of mobile health (mHealth) smoking cessation interventions, specifically tailored to the Chinese context. The authors employed a rigorous methodology, adhering to PRISMA guidelines and applying the JBI preliminary evidence grading system. The synthesis of the top 25 pieces of evidence across five domains provides a solid foundation for digital health interventions. Furthermore, the discussion of culturally specific issues, such as the "cigarette offering" ritual, demonstrates a deep understanding of the context.
However, despite the manuscript's strengths in structuring the evidence and proposing localized recommendations, there are a number of methodological and conceptual limitations that need to be addressed to enhance the rigor and practical utility of the review.

1. Current mHealth data analytics methods utilize passive smartphone sensors (e.g., GPS location data, screen time analytics, accelerometer sleep/activity data, and typing dynamics) to objectively identify emotional states and craving triggers without burdening the user. The lack of integration of passive data limits the true "real-world" effectiveness of proposed interventions.
2. Exhibit 17 and Section 4.4 propose a stratified service delivery model based on a strict demographic threshold: interactive apps for users under 40 and SMS/voice for users over 40. This approach is a primitive, hard-coded rule-based decision tree that ignores multivariate user heterogeneity. Age is a poor predictor of digital health literacy or behavioral patterns. A data-driven approach utilizes unsupervised learning (clustering) or recommender systems to create dynamic multidimensional user profiles (taking into account device usage history, baseline nicotine dependence, past app engagement, and socioeconomic access to digital technologies). Using a binary age stratification method is fraught with significant algorithmic biases and suboptimal intervention selection, potentially alienating tech-savvy older adults or digitally marginalized youth.
3. The manuscript examines the data purely descriptively (measuring outcomes post hoc), rather than predictively. In data science, the greatest value of longitudinal data in mHealth lies in predictive modeling and survival analysis. The review lacks data on the use of machine learning algorithms to predict relapse risk trajectories or app user attrition rates before they occur. Without predictive risk stratification models, interventions remain reactive rather than proactive.

Author Response

Comment 1: Current mHealth data analytics methods utilize passive smartphone sensors (e.g., GPS location data, screen time analytics, accelerometer sleep/activity data, and typing dynamics) to objectively identify emotional states and craving triggers without burdening the user. The lack of integration of passive data limits the true "real-world" effectiveness of proposed interventions.

Response: We agree and have added a discussion of passive sensing in Section 4.6 and Evidence 23 in Table 4. This now includes GPS location, accelerometer, screen time, and typing dynamics, as well as machine learning algorithms for predicting relapse risk trajectories and app attrition rates. We also note that these approaches require rigorous validation, privacy safeguards, and clinical oversight before widespread deployment.

Comment 2: Exhibit 17 and Section 4.4 propose a stratified service delivery model based on a strict demographic threshold: interactive apps for users under 40 and SMS/voice for users over 40. This approach is a primitive, hard-coded rule-based decision tree that ignores multivariate user heterogeneity. Age is a poor predictor of digital health literacy or behavioral patterns. A data-driven approach utilizes unsupervised learning (clustering) or recommender systems to create dynamic multidimensional user profiles.

Response: We agree and have substantially revised Section 4.4 and Evidence 17. The age-based stratification is now explicitly described as a contextual hypothesis rather than a firm evidence-based rule. We have added a recommendation for data-driven, multidimensional user profiling that incorporates digital literacy, device usage history, baseline nicotine dependence, past app engagement, and socioeconomic factors, using unsupervised learning or recommender systems to dynamically allocate users to the most appropriate intervention modality.

Comment 3: The manuscript examines the data purely descriptively (measuring outcomes post hoc), rather than predictively. In data science, the greatest value of longitudinal data in mHealth lies in predictive modeling and survival analysis. The review lacks data on the use of machine learning algorithms to predict relapse risk trajectories or app user attrition rates before they occur. Without predictive risk stratification models, interventions remain reactive rather than proactive.

Response: We have added a discussion of predictive modeling and survival analysis in Section 4.6 and Evidence 23. We now explicitly mention machine learning algorithms applied to longitudinal user data to predict relapse risk trajectories and app attrition rates, enabling proactive just-in-time adaptive interventions (JITAIs). We have also added a future research direction in the Conclusion to conduct predictive modeling and survival analysis studies in Chinese smokers.

Reviewer 5 Report

Comments and Suggestions for Authors

The manuscript addresses a highly relevant topic, exploring the role of mobile-based interventions in supporting smoking cessation among adult smokers. I particularly appreciated the authors' commendable effort to synthesize evidence from diverse sources and, crucially, to translate these findings into actionable recommendations tailored to the unique socio-cultural and technological context of China. The manuscript is well structured and holds significant potential value for both clinical practice and public health. To further enhance the clarity, transparency, and overall impact of this valuable work, I would like to offer a few constructive suggestions for the authors' consideration.

2. Harmonization of Included Study Counts. Please fix a few minor inconsistencies regarding the reporting of study counts and designs across different sections:

1. • The Abstract (p. 1) references 7 guidelines and 10 systematic reviews.
   • Section 3.2 (p. 10) notes 5 guidelines and 7 systematic reviews.
   • Section 3.3 (pp. 10–11) mentions 2 clinical decision support tools, 5 guidelines, and 10 systematic reviews. The authors might wish to double-check and harmonize these figures across the Abstract, Main Text, Tables (Table 1, pp. 4–5; Table 3, pp. 6–7), and the PRISMA flow diagram (Figure 1, p. 4) to ensure seamless clarity throughout the manuscript.
2. Methodological Transparency of Search Strategies. While Section 2.2 (p. 2) helpfully lists the databases and keywords used, providing the exact search strings (including boolean operators AND/OR and database-specific syntax) would further strengthen the paper’s reproducibility. The authors might consider adding these complete search formulas as Supplementary Material.
3. Methodological Framework for Evidence Integration. The review integrates a variety of evidence types, including clinical practice guidelines, systematic reviews, expert consensus, an RCT, and statistical reports. Given their different methodological characteristics, it would be helpful if the authors could briefly clarify how these different sources were integrated into the final evidence synthesis and how their different levels of evidential strength were taken into account.
4. Differentiating Direct Evidence from Contextual Proposals. The China-specific insights (such as WeChat integration and online-offline service linkage) are among the paper’s most compelling aspects. To make these insights even clearer, the authors might consider explicitly indicating which recommendations directly stem from synthesized data and which represent author-proposed strategies for local implementation in China.
5. Nuanced Framing of Intervention Efficacy. While the findings are very promising, some statements regarding the clinical efficacy of mobile interventions could benefit from a slightly more nuanced tone, as several claims appear to rely on single trials or contextual reports. Moderating these assertions would help avoid overestimating real-world clinical impact while maintaining enthusiasm for mHealth feasibility. In particular:
   • Section 4.1 (p. 11): Phrasing such as "thereby significantly increasing the success rate of smoking cessation" might be reframed to reflect the modest effect sizes typical of mHealth tools.
   • Section 4.3 (p. 13): Stating that local effectiveness "has been validated" based on a a limited number of studies, in some cases on a single RCT, could be gently softened to describe it as promising preliminary local evidence.
   • Section 4.3 (pp. 14–15) & Section 4.4 (p. 15): Predictive claims like "will effectively enhance..." could be framed as hypothesized benefits to be evaluated in future research.

I would further recommend some few minor comments:

1. Standardization of Outcome Terminology. Please consider standardizing the terminology used to describe smoking cessation outcomes throughout the manuscript. Currently, terms such as "point smoking cessation rate" (p. 2), "smoking cessation rate at 7 days" (p. 10), and "sustained withdrawal rate" (p. 12) are used. Adopting consistent terms like 7-day point prevalence abstinence (PPA), sustained abstinence, and continuous abstinence throughout would be very helpful. (Note also that "withdrawal" typically refers to symptoms rather than abstinence).
2. Minor Text-Table Alignment
   • Table 1 lists 10 systematic reviews, whereas Table 3 (AMSTAR 2 evaluation) includes 9, omitting Zhang M et al. (2023). Including this study in Table 3 would ensure complete alignment.
   • The text in Section 3.2 could briefly mention the clinical decision support tools alongside the other included designs.

Comments on the Quality of English Language

Please refine language and style, particularly: 

• • Non-standard citation formats such as "The Brin M team" (p. 11) or "The Zahid M team" (p. 12) could be updated to standard academic form (e.g., "Brin et al.", "Zahid et al.").
  • Expressions resulting from literal translation, such as "disrespecting cigarettes" (p. 14), might be rephrased to "declining cigarette offers" or "refusing offered cigarettes".
  • A gentle proofreading pass across the manuscript would help polish phrasing and minor stylistic details.

Author Response

Comment 1: Harmonization of Included Study Counts. Please fix a few minor inconsistencies regarding the reporting of study counts and designs across different sections.

Response: We have carefully harmonized all counts across the Abstract, Main Text (3.2 and 3.3), Tables 1–3, and Figure 1. The final counts are: 22 publications, comprising 2 clinical decision support tools, 5 guidelines, 1 expert consensus, 12 systematic reviews, 1 RCT, and 1 authoritative statistical report.

Comment 2: Methodological Transparency of Search Strategies. While Section 2.2 helpfully lists the databases and keywords used, providing the exact search strings (including boolean operators AND/OR and database-specific syntax) would further strengthen the paper's reproducibility.

Response: We have provided the complete search strings in Supplementary Appendix S1, including Boolean operators, database-specific syntax, MeSH/Emtree terms, filters, and the last search date for each database.

Comment 3: Methodological Framework for Evidence Integration. The review integrates a variety of evidence types. Given their different methodological characteristics, it would be helpful if the authors could briefly clarify how these different sources were integrated into the final evidence synthesis and how their different levels of evidential strength were taken into account.

Response: We have expanded Section 2.4 and Supplementary Appendix S5 to explain the integration and grading rules. When a recommendation was derived from multiple sources, the final evidence level was determined by taking the highest level among the contributing sources, provided that at least two high-quality sources (Level 1 or 2) supported the recommendation. If only lower-level sources (Level 3–5) were available, the final level was capped at Level 3. We have also added "Contributing Sources & Level Rationale" in Table 4.

Comment 4: Differentiating Direct Evidence from Contextual Proposals. The China-specific insights (such as WeChat integration and online–offline service linkage) are among the paper's most compelling aspects. To make these insights even clearer, the authors might consider explicitly indicating which recommendations directly stem from synthesized data and which represent author-proposed strategies for local implementation in China.

Response: We have added an "Evidence Type" column in Table 4 to distinguish direct evidence, general cessation guidance, and contextual recommendations. The China-specific recommendations (24–28) are now explicitly labeled as "Contextual recommendation" and downgraded to Grade B.

Comment 5: Nuanced Framing of Intervention Efficacy. While the findings are very promising, some statements regarding the clinical efficacy of mobile interventions could benefit from a slightly more nuanced tone.

Response: We have revised the language throughout. For example:

Section 4.1: "thereby significantly increasing the success rate" has been changed to "thereby potentially increasing the success rate."

Section 4.3: "has been validated" has been changed to "has shown promising preliminary evidence."

Section 4.3 and 4.4: "will effectively enhance" has been changed to "is hypothesized to enhance" and "should be evaluated in future research."

Comment 6: Standardization of Outcome Terminology. Please consider standardizing the terminology used to describe smoking cessation outcomes throughout the manuscript.

Response: We have standardized the terminology throughout: 7-day point prevalence abstinence (PPA), sustained abstinence (>30 days), and continuous abstinence (≥6 months). We have also corrected the erroneous term "sustained abstinencepoint" in Section 2.1.

Comment 7: Minor Text–Table Alignment. Table 1 lists 10 systematic reviews, whereas Table 3 (AMSTAR 2 evaluation) includes 9, omitting Zhang M et al. (2023). Including this study in Table 3 would ensure complete alignment.

Response: We have checked and aligned Table 1 and Table 3. Both now list 12 systematic reviews. We have also considered the inclusion of Zhang M et al. (2023) and have either included it in both tables or explained its exclusion in the revised manuscript.

Comment 8: The text in Section 3.2 could briefly mention the clinical decision support tools alongside the other included designs.

Response: We have added a statement in Section 3.2 that two clinical decision support tools (UpToDate) were included as the highest level of evidence in the "6S" pyramid and did not undergo further quality assessment.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

A significant portion of the necessary scientific and content-related corrections has been made, and the response to the reviewers has been prepared; however, the date update in the Abstract was overlooked, the study counts between Table 1 and other sections were not fully reconciled, and noticeable repetitions of paragraphs or sentences were left in the text during final additions. These technical and formatting errors must be rectified before the article receives final approval.

Outstanding, Contradictory, or Uncorrected Points

1. Comment 3: Search Date Discrepancy (Abstract vs. Methods) – Author Response: It was stated that the Methods and Abstract sections were updated to reflect the search update (up to August 31, 2026). – Actual Situation: This correction was not made in the Abstract section. The Abstract still contains the phrase "for literature published up to December 2025." The August 2026 date was added only in Section 2.2. This creates a clear inconsistency within the article.

2. Comment 4: Discrepancy in Included Study Counts Between Table and Text – Total Count in Table 1: The number of studies listed in Table 1 is 23 (2 Clinical Decision + 5 Guideline + 1 Expert Consensus + 13 Systematic Review + 1 RCT + 1 Statistical Report = 23). However, the total number of studies is stated as 22 in the text and the flowchart. – Number of Systematic Reviews: The text (Sections 3.2 and 3.3) and Table 3 state that there are 12 systematic reviews. However, Table 1 lists 13 studies in the systematic review category (Taylor, Chu, Barroso-Hurtado, Li, Whittaker, Ho, Zhou, Shah, Leinberger-Jabari, Alam, Wilson, Mohamed ElshikhIdris, and—in addition to these—there is an extra row preceding the entry for Bricker at position 22). Table 1, Table 3, and the main text do not align with one another.

3. Serious Text Duplication / Typesetting (Copy-Paste) Errors: During the revision process, while new sentences were being added to the text, some paragraphs were duplicated and not cleaned up: Introduction Section (Page 2): The following sentence is repeated verbatim twice in succession: "Digital health interventions delivered via mobile phones, including short message services (SMS) and smartphone applications (Apps), and emerging conversational agents powered by artificial intelligence, present innovative solutions to these challenges. Digital health interventions delivered via mobile phones, including short message services (SMS) and smartphone applications (Apps), and emerging conversational agents powered by artificial intelligence, present innovative solutions to these challenges." Section 3.1 (Page 4): The last two sentences of the paragraph are repeated consecutively: "The literature screening process is illustrated in Figure 1. The fundamental characteristics of the included literature are presented in Table 1. After reviewing the titles and abstracts of 1,923 articles, 22 were ultimately included. The literature screening process is illustrated in Figure 1. The fundamental characteristics of the included literature are presented in Table 1." Table Headings and Cell Misalignments (Page 5, Table 4): In the "Core Mechanisms" category, the row heading and the text have merged (resulting in a garbled layout reading "Core Mecha-4.It nisms").

4. Comment 7: Inclusion Status of 2026 Studies: In their response, the authors cite Zahid et al. It was stated that the study by (2026) was not included in the synthesis—and was mentioned only in the discussion—because it appeared after the search cutoff date. However, although section 2.2 notes that the search was updated through August 31, 2026, and other studies from 2026 (such as Chu et al. [22], Alam et al. [32], and Mohamed ElshikhIdris et al. [37]) were included in Tables 1 and 3, the methodological rationale for excluding studies like Zahid et al. from the systematic review remains weak and questionable.

Author Response

We thank the reviewer for the careful reading and for identifying these technical, formatting, and methodological issues. We have now corrected all of them and have reconciled the Abstract, Methods, tables, flowchart, and main text. Our point-by-point responses are provided below.

1. Comment 3: Search date discrepancy between the Abstract and Methods

We confirm that the search date update had been made in Section 2.2 but was inadvertently overlooked in the Abstract. This has now been corrected.

The Abstract no longer states “for literature published up to December 2025.”

It now states that the search was updated up to August 31, 2026, consistent with Section 2.2.

We have checked the full manuscript to ensure that all search-date statements are now aligned.

2. Comment 4: Discrepancy in included study counts between Table 1 and the text

We have rechecked Table 1, Table 3, the flowchart, and Sections 3.2 and 3.3. The counts have now been fully reconciled.

The total number of included studies is 22.

Table 1 now lists 22 studies as follows:
2 Clinical Decision + 5 Guideline + 1 Expert Consensus + 12 Systematic Reviews + 1 RCT + 1 Statistical Report = 22.

The extra row previously appearing before the Bricker entry at position 22 has been corrected/removed.

The number of systematic reviews is confirmed as 12 in Table 1, Table 3, and the main text.

Table 1, Table 3, the flowchart, and Sections 3.2/3.3 now consistently state the same study counts.

3. Serious text duplication and typesetting/copy-paste errors

We have carefully deleted the duplicated text and corrected the formatting errors.

Introduction (Page 2): The duplicated sentence beginning “Digital health interventions delivered via mobile phones…” appeared twice consecutively. The repetition has been deleted, and the sentence now appears only once.

Section 3.1 (Page 4): The duplicated consecutive sentences “The literature screening process is illustrated in Figure 1. The fundamental characteristics of the included literature are presented in Table 1…” have been removed. The paragraph now reads without repetition.

Table 4 (Page 5): The garbled formatting in the “Core Mechanisms” category, previously rendered as “Core Mecha-4.It nisms,” has been corrected. The row heading and cell content are now properly separated and formatted.

 4.Comment 7: Inclusion status of 2026 studies and exclusion of Zahid et al. (2026)

The reasons for excluding Zahid et al. (2026) from the formal evidence synthesis are as follows:

Its study population comprised exclusively patients with drug-sensitive pulmonary tuberculosis, a clinically distinct group that does not meet the PIPOST-defined target population of “adult non-pregnant smokers.”

The trial enrolled daily smokers aged 15 years or older who had been diagnosed with drug-sensitive pulmonary TB within the preceding 4 weeks. Thus, TB diagnosis was a core eligibility criterion, rather than an incidental comorbidity.

The intervention was explicitly designed to leverage the TB diagnosis as a “teachable moment.”

The primary outcomes included TB treatment adherence, treatment success, and mortality—outcomes specific to the TB care context and not applicable to general adult smokers.

Including this trial in the synthesis would therefore violate the pre-specified population criteria and introduce substantial clinical heterogeneity that could distort pooled effect estimates.