Prescriptive Maintenance: A Systematic Literature Review and Exploratory Meta-Synthesis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Accept after minor revisions.

Comments for author File: Comments.pdf

Author Response

Comment 1: This article provides a comprehensive and application-oriented systematic literature review and exploratory meta synthesis of Prescriptive maintenance (PsM) in the post Industry 4.0 era, identifying its key driving factors and laying a key foundation for intelligent, interpretable, and action oriented maintenance systems.

Response 1: Dear reviewer, thank you very much for a positive feedback. In the following we provide point-by-point response to the suggestion comments and recommendations for improving the quality of the manuscript.

 

Comment 2: The background exposition section of the paper lacks sufficient analysis of the correlation between policies and the development of PsM, and does not deeply integrate specific policies. The content still lacks comprehensiveness, depth, and specificity, and needs further improvement to more effectively support research.

Response 2: Dear, thank you for the comment. Based on the suggestion, we have revised the introduction section by specifically discussing the correlation between specific policies and development of PsM. We have explored and expanded the background section by incorporating detailed analysis how Industry 4.0 and Industry 5.0 influenced the PsM development by providing examples of policy-driven technological advancements.

Comment 3: In terms of theoretical framework, there is a lack of systematic integration between PsM and existing maintenance theories. Although the article compared the differences between PsM and PdM, it did not fully construct the core theory of PsM and Industry 5.0, and the theoretical integration needs to be improved.

Response 3: Dear reviewer, thank you for the comment. However, although the PsM theories are not addressed in this systematic review, as it was addressed and discussed in previous review by Giacotto, so we did not want to overlap with prior systematic review. The idea is to target the gaps not previously addressed. Hence, although we did not want to go into depth, we have provided core theoretical distinctions, particularly addressing human-centricity, resilience and sustainability driving the evolution of PsM. This also aligns with prior remark on policies and regulatory frameworks driving the evolution of PsM.

 

Comment 4: In terms of research content, the existing analysis mainly focuses on the manufacturing, energy, and aerospace industries, and does not explain the special characteristics and challenges of PsM applications in these industries, which limits the universality of the research.

Response 4: Dear reviewer, thank you for the comment. However, we would kindly ask you to read line numbers 545-592 as we specifically addressed this question and provided an illustrative examples specifically discussing the characterisation of PsM in manufacturing, energy, aerospace and railway sector. In addition, we have provided an in-depth discussion of PsM activities within, and lastly, we have provided transparent and replicable content-based evidence by extracting PsM challenges for each individual study on PsM that is provided in supplementary file. Also, to address the challenges and special characteristics, we provide an exemplification of different challenges by delivering potential solution spaces as future research agenda. Please see completely revised section 4.4.

 

Comment 5: In terms of methodology, the article relies on limited databases, which may introduce subjective bias and affect the comprehensiveness and reliability of the research.

Response 5: Dear reviewer, we thank you for the comment. However, based on our extensive experience, and looking through all other published reviewers, this is the most extensive databases used, as we covered Google Scholar, Open Alex, Web of Science, SCOPUS and Dimensions, highlighting the most extensive search, given that Google Scholar and Open Alex (previous Microsoft Knowledge Academic Graph) alone are widely known as the most extensive sources of scientific records. Nevertheless, to make sure there was no bias and to assure reliability we have highlighted that independent reviewers collected records and merged the results together leading to 0.83 interrater agreement. This assured that study records are reviewed by each individual reviewer and objective review strategy achieved consensus that all relevant studies are included in the review. Also, we have even provided transparent and replicable results providing all study records from screening, after screening and with eligibility criteria by providing reviewers and readers of interest all included and excluded studies in this review.

 

Comment 6: In terms of research perspective, the analysis is mainly based on the existing theoretical framework, and there is insufficient exploration of the economic and organizational factors of PsM implementation, which fails to fully reflect the practical challenges of PsM implementation.

Response 6: Dear thank you for the comment. This has been addressed in the section 4.4 regarding the current challenges and research agenda. Moreover, given that this is a systematic review the evidence collected from prior original primary studies are used for the analysis in depicting the existing challenges of prior primary studies. Moreover, this has also been addressed in the limitations of studies and depicted through factor analysis leveraging MCA-HCPC method. We clearly depict that only studies in cluster 3 are theoretical while cluster 1 and cluster 2 are primary studies with practical case studies and/or experimental validation.

 

Comment 7: The research conclusion section does not provide differentiated implementation paths for enterprises with different technological maturity levels or different decision-making levels, which limits the universality of the research conclusion and fails to reflect its practical application value.

Response 7: Dear thank you for this valuable comment. The purpose of this systematic review is to provide theoretical clarity of PsM by clearly and systematically identify PsM studies published Post I4.0. However, understanding the need for practicality we specifically discuss the importance of technological requirements and decision making levels that play a role in the adoption of PsM approach. Please see revised version of conclusion, highlighted in line numbers 845-864.

 

Again, we would like to thank the reviewer for valuable comments and suggestions for improving the quality of the manuscript. We strongly appreciate your supporting and participation in the revising the manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

The paper presents a systematic literature review of prescriptive maintenance, an important and increasingly popular topic. It is technically strong, employing a range of analytical methods that yield robust quantitative results. However, the theoretical analysis is weak, as shown by limited argumentation and justification, and it lacks a research outlook, which is a core outcome of such a review. The writing also needs improvement: the narrative is sometimes incoherent, the Related Work section feels underdeveloped, and the Methodology section does not fully reflect the work conducted. Therefore, I recommend the paper for acceptance pending major revisions.

In the following I have detailed the most pending issues:

• Introduction, related work and research problem: The current Introduction tries to cover the paper background, related work, and problem statements all at once. Separating these into dedicated sections would help readers follow the story: the motivation, how this work fits into existing literature, and the scope. An Introduction should guide the reader from context and motivations through positioning and contributions, ending with an outline of the paper structure. Condensing all three topics into one section makes it hard to convey their distinct value and dilutes the main contributions.
• Unsupported statements: Some claims are unreferenced, such as the assertion in line 59: “Such ambitions are often disconnected in theory and practice, which is why PsM adoption is limited.”
• Related work section: Related Work currently reads like a problem statement concluding the research gap. However, in a review article, it should provide readers with deeper insight into the topic under consideration: the origins and evolution of prescriptive maintenance, and an analysis of what makes it different from related approaches (condition-based maintenance, predictive maintenance). Explaining the differences between perspective and predictive. It should also cover the application areas where it first emerged and has since been extended, and highlight the cornerstone technologies and tools that enhance prescriptive maintenance capabilities across different domains. Once these foundations are set, the Related Work section can then position this review within the existing literature.
• Research Question section: The research questions and aims are clearly listed. However, the motivation behind these questions, in terms of their benefits and added value to the body of knowledge, is missing. Including this would help readers understand why it’s worth reading such a long document.
• Research Framework (method section): The Research Framework section reads more like a methodology, but it’s missing a clear justification for the multi‐step process shown in Figure 1. Each of the six steps should be explained in detail: outline the motivation behind it, justify the chosen method, and describe exactly how it was executed (for example, applying a Fleiss’s kappa threshold of 0.83). Breaking this section into six subsections will help readers follow and understand each stage the approach before the result analysis.
• Distinction between PdM and PsM in Reviewed Studies: The authors should clarify the criteria used to distinguish predictive (PdM) from prescriptive (PsM) papers during screening. For example, did you label a recommendation as “prescriptive” only if it was automated or if it optimized a specific objective function?
• Research Agenda/outlooks: The section lacks drawing actionable insights from the results. Such as a translating challenges into guidance/opportunities for the scientific community, highlighting the “low-hanging fruits” and key challenges that remain. A focused discussion on which methods or research avenues are most promising, whether to overcome specific limitations in prescriptive maintenance techniques or to expand into new application domains, would greatly enhance the paper practical value.
• Proof reading: Many sentences throughout the paper need rewriting. Some are confusing, and others have grammar errors. This is likely a result of AI proofreading. Here are some examples (Line 156: The assessment of included is performed by…; Line 206: excluded. Lastly, Timeframe and Abstract consider using only studies after….; Line 218: … reviews that are still under review or can be expectedr in the…)
• Terminology: Maintain consistent usage—once you define PsM, use it throughout rather than switching to prescriptive analytics. Also define any acronyms (for example, PriMa) the first time they appear.
• An introductory paragraph for sections: Each main section should begin with a brief introductory paragraph before its subsections (for example, Section 3: Results).

Comments on the Quality of English Language

Need improvement

Author Response

Comment 1: Introduction, related work and research problem: The current Introduction tries to cover the paper background, related work, and problem statements all at once. Separating these into dedicated sections would help readers follow the story: the motivation, how this work fits into existing literature, and the scope. An Introduction should guide the reader from context and motivations through positioning and contributions, ending with an outline of the paper structure. Condensing all three topics into one section makes it hard to convey their distinct value and dilutes the main contributions. Unsupported statements: Some claims are unreferenced, such as the assertion in line 59: “Such ambitions are often disconnected in theory and practice, which is why PsM adoption is limited.”

Response 1: Dear reviewer, thank you for bringing this to our attention. We have extensively reworked the introduction section. In the revised version we also added an introductory paragraph per your Comment 9 (see bottom revision comment), and in every other section. As for the subsections of the introduction, we have highlighted every subsection separately and also reworked parts of the introduction (see highlighted version). As for the unsupported statement, our statement is made from personal experience from practice in response to the prior statements. Given the suggestion you've made, we've decided to remove the sentence and avoid disagreement in supporting such claim.

 

Comment 2: Related work section: Related Work currently reads like a problem statement concluding the research gap. However, in a review article, it should provide readers with deeper insight into the topic under consideration: the origins and evolution of prescriptive maintenance, and an analysis of what makes it different from related approaches (condition-based maintenance, predictive maintenance). Explaining the differences between perspective and predictive. It should also cover the application areas where it first emerged and has since been extended, and highlight the cornerstone technologies and tools that enhance prescriptive maintenance capabilities across different domains. Once these foundations are set, the Related Work section can then position this review within the existing literature.

Response 2: Dear reviewer, thank you very much for your valuable feedback. Explaining the difference between prior CBM and PdM maintenance practices in response to the upcoming interest in PsM, we clearly delinate one from the other by explaining the evolution of PdM and afterwards strategic relevance of PsM practice to make the section more coherent, and setting the stage for Related Work, where we position our Systematic Review amongst other published related reviews on the topic at hand. Please see revised version line numbers 48-63 ("As industrial systems grow more complex and interconnected, maintenance strategies, such as Condition-Based Maintenance (CBM) [5] and Predictive Maintenance (PdM) [6] evolved to keep pace with growing operational demands. However, although PdM is beneficial for forecasting failures, it falls short in determining what actions should be taken, when to take them, and how to optimise decisions under limited resources and environmental uncertainty [7]. In response, PsM emerged as the next evolutionary step, offering targeted and optimised interventions that align with business objectives [8–11]. The strategic relevance of PsM is amplified by its alignment with Industry 4.0 (I4.0) technologies (e.g., Internet of Things (IoT), Machine Learning (ML), Edge and Cloud Computing), and the emerging vision of Industry 5.0 (I5.0) [12]. Leveraging Cyber Physical Systems (CPS) [13], IoT connectivity [14], and Digital Twins [15], PsM decision-support systems can perform real-time and data-driven decision-making across the production lifecycle [16]. Simultaneously, consideration of I5.0 pillars (human-centricity, resilience and sustainability) as core industrial values [17], directs the development of decision-support systems by respecting social and environmental norms [18] that further enhances human-machine collaboration [19].").

 

Comment 3: Research Question section: The research questions and aims are clearly listed. However, the motivation behind these questions, in terms of their benefits and added value to the body of knowledge, is missing. Including this would help readers understand why it’s worth reading such a long document.

Response 3: Dear reviewer, thank you for this valuable suggestion. We have revised and expanded the motivation, aims, and objectives of RQs to enhance understanding of their purpose and contribution to the academic community and practice.

 

Comment 4: Research Framework (method section): The Research Framework section reads more like a methodology, but it’s missing a clear justification for the multi‐step process shown in Figure 1. Each of the six steps should be explained in detail: outline the motivation behind it, justify the chosen method, and describe exactly how it was executed (for example, applying a Fleiss’s kappa threshold of 0.83). Breaking this section into six subsections will help readers follow and understand each stage the approach before the result analysis.

Response 4: Dear reviewer, thank you for the comment. In the revised version we have renamed the section, completely changed the introductory part of the Methodology by summarising the structure of the modules explaining study retrieval, search strategy, data extraction, analysis of meta data, analysis of content data and analysis of findings. Each individual module of our methodology is separately and explained in-detail. Also, we have provided an explanation of interrater agreement, suggesting the agreement of included studies before screening and after screening using Fleiss kappa, which is a score to measure agreement among authors for including the articles. Usually a Cohen kappa is used for two reviewers, however with three reviewers Fleiss kappa is more appropriate. Please see completely revised section of methodology. Line numbers 162-337.

 

Comment 5: Distinction between PdM and PsM in Reviewed Studies: The authors should clarify the criteria used to distinguish predictive (PdM) from prescriptive (PsM) papers during screening. For example, did you label a recommendation as “prescriptive” only if it was automated or if it optimized a specific objective function?

Response 5: Dear reviewer, thank you for this valuable suggestion. As we have dedicated extensive efforts to delineating this, and to avoid overlap between studies of PdM and PsM, some suggest that they are dealing with PsM by only citing PsM or stating prescriptive maintenance without actually addressing post-prognostic activities. We have provided differentiation and problem of PdM versus PsM in the introduction, please see line number 48-63  ("As industrial systems grow more complex and interconnected, maintenance strategies, such as Condition-Based Maintenance (CBM) [6] and Predictive Maintenance (PdM) [7] evolved to keep pace with growing operational demands. However, although PdM showed to be particularly useful for forecasting failures, it falls short in determining what actions should be taken, when to take them, and how to optimize decisions under limited resources and environmental uncertainty [8]. In response, PsM emerged as the next evolutionary step, offering targeted and optimised interventions that align with business objectives [9–12]."). In addition, we have also stated in our aims "This SLR aims to advance the theoretical clarity of PsM by: (i) systematically identifying PsM studies published post-I4.0 period; (ii) critically appraising quality of studies dealing specifically with prescriptive aspects and post-prognostic (and diagnostic) recommendations and decision-making; ", but also in the methodology by describing eligibility criteria for isolating specifically PsM studies in line numbers 214-218 "Lastly, considerable efforts are dedicated to studies conflating PsM with PdM maintenance unless they explicitly address prescriptive actions (e.g., decision-making, optimisation, scheduling).". This was also explicitly discussed to provide a more narrow understanding and characterisation of PsM in section 4.2 Prescriptive Maintenance actions line number 586-594 "The prescriptive dimension of PsM is characterised by the ability to convert predictive insights (e.g., failure diagnosis, RUL) into actionable recommendations. These recommendations specify what activities should be taken, when, by whom, and under what constraints. In contrast to traditional maintenance strategies, PsM extends this pipeline by embedding decision-making logic into maintenance execution. This marks a decisive shift from mere prognostics to prescriptive intelligence. To capture this dimension, we synthesised a diverse body of empirical and conceptual studies across industrial domains to show how PsM is operationalised in practice (Table 5).". Lastly, as you can see in table 5 we did not only cover automated execution and self-regulation type of activities but rather we determined them based on different type of actions that align with any of the operational recommendations "what actions should be taken, when to take them, how to optimize decisions, by whom, and under what constraints.".

 

Comment 6: Research Agenda/outlooks: The section lacks drawing actionable insights from the results. Such as a translating challenges into guidance/opportunities for the scientific community, highlighting the “low-hanging fruits” and key challenges that remain. A focused discussion on which methods or research avenues are most promising, whether to overcome specific limitations in prescriptive maintenance techniques or to expand into new application domains, would greatly enhance the paper practical value.

Response 6: Dear, thank you very much for bringing this to our attention. We have decided to extensively work on this part of our findings. This is now completely reworked. Namely, we considered extracted data, i.e., challenges and problems, and offer research avenus and agendas that could potentially be of interest for future exploration. In addition,  for each challenge that was encountered from exhaustive reading of prior studies, we exemplify through a practical scenarios, that can help reader better understand both the problem and solution space, ultimately provoking research for embarking into particular agenda.

 

Comment 7: Proof reading: Many sentences throughout the paper need rewriting. Some are confusing, and others have grammar errors. This is likely a result of AI proofreading. Here are some examples (Line 156: The assessment of included is performed by…; Line 206: excluded. Lastly, Timeframe and Abstract consider using only studies after….; Line 218: … reviews that are still under review or can be expectedr in the…).

Response 7: Dear reviewer, thank you for pointing out the spelling errors and grammatical mistakes. We strongly appreciate it. We have completely revised the manuscript and checked it for spelling and grammatical errors.

 

Comment 8: Terminology: Maintain consistent usage—once you define PsM, use it throughout rather than switching to prescriptive analytics. Also define any acronyms (for example, PriMa) the first time they appear.

Response 8: Dear reviewer, thank you for bringing this to our attention. We apologise for omitting acronyms in the manuscript. We have reworked this accordingly. We also reworked prescriptive analytics and used consistent PsA.

 

Comment 9: An introductory paragraph for sections: Each main section should begin with a brief introductory paragraph before its subsections (for example, Section 3: Results).

Response 9: Dear, thank you for the suggestion. We have reworked the manuscript by adding a brief introductory paragraph to each section, preceding its subsections. Please see the revised version of the manuscript.

Again, we would like to express extreme gratitude to valuable comments, insights and recommendations aimed at improving the quality of the manuscript. However, we would also like to draw attention that different types of revisions requested by reviewers was hard to isolate and we tried to the best of our efforts to respond to every reviewer by considering all comments with of utmost importance. Again, we thank all reviewers' comments.

Reviewer 3 Report

Comments and Suggestions for Authors

The article investigates predictive maintenance tools and aims to map the functional space for predictive maintenance across various industrial applications. The focus is on analyzing literature published in the post-Industry 4.0 era. The article aligns well with the scope of the journal; however, several revisions are necessary before it can be considered for publication. Please see my detailed comments below:

1. It is recommended to include the year range of the reviewed articles (as stated in Section 2.2) in the Abstract to provide better context.
2. Consider adding a summary table at the end of Section 1.2 to clearly present the motivation behind the review.
3. Please include a short paragraph highlighting the key differences between this review and existing reviews in the literature.
4. On page 10, line 311, please add a reference to Figure 4 (related to VosViewer) for clarity and completeness.

Author Response

Comment 1: It is recommended to include the year range of the reviewed articles (as stated in Section 2.2) in the Abstract to provide better context.
Response 1: Dear reviewer, thank you very much for the suggestion. The abstract has been revised accordingly.

Comment 2: Consider adding a summary table at the end of Section 1.2 to clearly present the motivation behind the review.
Response 2: Dear, thank you very much for bringing this to our attention. Given that another reviewer suggested to provide a motivation for research questions, we considered to merge the rationale and motivation behind both the prior reviews and motivation for proposing each individual research question. We have revised the section accordingly. In this way, we consider it will be better aligned by considering simultaneously gaps in the prior reviews and need for addressing such gaps through proposed research questions. Please see revised version of the manuscript, line numbers 132-155.

Comment 3: Please include a short paragraph highlighting the key differences between this review and existing reviews in the literature.
Response 3: Dear, thank you very much for the comment. We have revised second to last paragraph before 1.3 and offer distinction between this review and prior reviews on PsM. Please see line numbers 115-123.

Comment 4: On page 10, line 311, please add a reference to Figure 4 (related to VosViewer) for clarity and completeness.
Response 4: Dear reviewer, thank you very much for the comment. The refence to the figure has been added.

Again, we would like to thank the reviewer for the valuable comment and contribution for improving the quality of our manuscript.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The introduction and related work sections could have been restructured and developed in a more effective way (cf. Comment 1 and 2). While the revisions bring better clarity, the related work still lacks depth in terms of historical grounding and domain-specific adoption. However, I do not see this as preventing the paper from being considered for publication.