A Methodological Approach to Extracting Patterns of Service Utilization from a Cross-Continuum High Dimensional Healthcare Dataset to Support Care Delivery Optimization for Patients with Complex Problems
, Yudi Santoso 2, Hanieh Sadri 2, Ken Moselle 3, Abraham Rudnick 4,*, Stan Robertson 5, Ernie Chang 6, Alex Kuo 1, Joseph Howie 2, Gracia Yunruo Dong 7,8, Kehinde Olobatuyi 8, Mahdi Hajiabadi 2 and Ashlin Richardson 9Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
Nice effort. Please pay attention to the following issues:
1. You may reduce the number of keywords. For example, you may omit these ( Clustering, Electronic Healthcare,)
2. Can you re-write objectives? It would be great if you wrote the Line 189-200 first and then finally put your objective.
3. You may use your method with a common diagram.
4. In the Results section, you may put Line 360-362 in the Method section, as you used R and Python for analysis.
5. In the discussion section, you need to discuss more with other relevant studies.
6. Please add your strengths and limitations.
7. You need to explain the importance of your study.
Author Response
- You may reduce the number of keywords. For example, you may omit these ( Clustering, Electronic Healthcare,)
Answer: done
- Can you re-write objectives? It would be great if you wrote the Line 189-200 first and then finally put your objective.
Answer: Most of the introduction has been re-written to provide more clarity on the topic of interest, and part of the objective section has been modified as well.
- You may use your method with a common diagram.
Answer: A diagram has been added to the discussion section summarizing the proposed methods and corresponding results.
- In the Results section, you may put Line 360-362 in the Method section, as you used R and Python for analysis.
Answer: Both Analysis and Result have been combined and the section has been renamed accordingly
- In the discussion section, you need to discuss more with other relevant studies.
Answer: added
- Please add your strengths and limitations.
Answer: added
- You need to explain the importance of your study.
Answer: added
Thanks very much for your feedback
Reviewer 2 Report
Comments and Suggestions for Authors
Methodological Considerations in Extracting and Analyzing Patterns of Service Utilization for Patients with Complex Problems to Optimize Care Delivery
Comment:
1. The title is confusing and suggested for modification.
2. In Line 117, ML promotes health care regarding diagnosis/classification / therapeutic.
3. In line 123, employing machine learning algorithms and which algorithm is applied is very general.
4. Strongly suggested to take a chronic disease and propose a process flow of activities (with probabilities) needed for the patient. Ex: 1. Chronic kidney disease patient and his routine procedure or activities. Ex:2. Pregnancy to postpartum stages.
5. Graphical representation of results is recommended to exhibit comparison.
6. Explicitly mention what is needed through ML algorithms and propose one ML algorithm based on your study.
7. Abstract and Conclusion needs statistical outcome.
8. Format the paper as per the Journal style.
Comments on the Quality of English Language
No issues
Author Response
- The title is confusing and suggested for modification.
Answer: proposed title: Methodological Approach to Extracting Patterns of Service Utilization from Cross-Continuum High Dimensional Healthcare Dataset to Support Care Delivery Optimization for Patients with Complex Problems
- In Line 117, ML promotes health care regarding diagnosis/classification / therapeutic.
Answer: The introduction has been reframed to prevent any confusion with regards to the diagnosis approach
- In line 123, employing machine learning algorithms and which algorithm is applied is very
Answer: The section on Method has been expanded and additional figures and details have been added to the manuscript to provide a better explanation of the ML algorithm used
- Strongly suggested to take a chronic disease and propose a process flow of activities (with probabilities) needed for the patient. Ex: 1. Chronic kidney disease patient and his routine procedure or activities. Ex:2. Pregnancy to postpartum stages.
Answer: The existing cohort was replaced by the schizophrenia cohort and more steps have been added to explain the proposed methodology
- Graphical representation of results is recommended to exhibit comparison.
Answer: A graphical representation of the communities of services generation process and a sample result was added to the discussion section
- Explicitly mention what is needed through ML algorithms and propose one ML algorithm based on your study.
Answer: Community Detection Algorithm Using Louvain Algorithm is the one that was used in our analysis. However, this algorithm was slightly modified by introducing an iterative process to access the challenges presented by the dataset we have been using. More details, figures and examples were added into the method section to provide more details on the algorithm used.
- Abstract and Conclusion needs statistical outcome.
Answer: this specific study is very thin on statistics. We have added some basic numbers to the discussion section. We have two papers building on the concept of PSUs introduced in this paper that are heavy on statistics. These papers have been submitted for review and include: (1) Disparities in Access to Services, as Evident in Patients Journeys: Illustrating a Nuanced Approach in Assessing Healthcare Equity Using Patterns of Service Utilization Across the Full Continuum of Care and (2) Use of Patterns of Service Utilization and Hierarchical Survival Analysis in Planning and Providing Care for Overdose Patients and Predicting the Time-to-Second Overdose
- Format the paper as per the Journal style.
Answer: the format of the paper has been adjusted to the journal requirements
Reviewer 3 Report
Comments and Suggestions for Authors
The topic is interesting. I have few following concerns:
1. According to the Authors’ intention, “the paper represents a marked departure from more purely diagnostically based methods for service system improvement.”
However, they must describe Diagnostic approaches.
How does it work?
What are the findings of other researchers?
2. In Line 84, This assumes that under ideal conditions as set out above, cohort-specific real-world.
What is the meaning of Ideal Condition?
Can authors write and define the criteria for Ideal Conditions?
As an International reader, it could be easier to understand.
3. In section 1.2, the Authors provide the literature and their findings.
However, it could be written in a better way to understand. For example, using a table could be a better option.
Currently, it isn't easy to understand.
4. In a Machine learning study, the success of the study mainly depends on the “Data Preprocessing” and selection of “features.”
From the current state of the study, it isn't easy to find out these two conditions.
How the Data is selected for the final study?
What kind of features are included in the study?
Why do the authors believe the selected features are the best to describe their study purpose?
How do the authors compare their selection process as better than previous studies?
Thus, the authors must include all these answers in their study. The current form of the manuscript is pretty challenging for a reader to understand.
5. The authors must provide a better explanation for Figure 3.
Such as how they deal with the Null nodes?
How are the null nodes treated in the next Iteration? Etc.
6. It would be better to make a table and describe the basic features of the final data.
Section 4.1 is not easily readable in the current stage.
7. Authors should provide a better explanation of the implications/contribution of the study in the existing literature.
Authors should discuss the Limitations of the current study and how these could be overthrown in a separate Section.
Comments on the Quality of English Language
Moderate English corrections make the paper more legitimate to the readers.
Author Response
Comments and Suggestions for Authors
The topic is interesting. I have few following concerns:
- According to the Authors’ intention, “the paper represents a marked departure from more purely diagnostically based methods for service system improvement.”
However, they must describe Diagnostic approaches.
How does it work?
What are the findings of other researchers?
Answer: The aim of this paper is not to provide a comparative approach with existing diagnostic approaches. The focus is on extracting patterns of service utilization within a high dimensional cross continuum dataset to provide the first step support in evaluating conformance of interventions to cohort-specific clinical practice guidelines. The introduction has been reframed to prevent any confusion with regards to the diagnostic approach
- In Line 84, This assumes that under ideal conditions as set out above, cohort-specific real-world.
What is the meaning of Ideal Condition?
Can authors write and define the criteria for Ideal Conditions?
As an International reader, it could be easier to understand.
Answer: this paragraph was removed as part of the restructuring of the introduction to prevent any confusion
- In section 1.2, the Authors provide the literature and their findings.
However, it could be written in a better way to understand. For example, using a table could be a better option.
Currently, it isn't easy to understand.
Answer: The literature review has been re-written to provide more details and make it more understandable
- In a Machine learning study, the success of the study mainly depends on the “Data Preprocessing” and selection of “features.”
From the current state of the study, it isn't easy to find out these two conditions.
How the Data is selected for the final study?
What kind of features are included in the study?
Why do the authors believe the selected features are the best to describe their study purpose?
How do the authors compare their selection process as better than previous studies?
Thus, the authors must include all these answers in their study. The current form of the manuscript is pretty challenging for a reader to understand.
Answer: Feature Selection, Data Pre-Processing and Data Re-Engineering sections were added to the manuscript to provide more details on these topics
- The authors must provide a better explanation for Figure 3.
Such as how they deal with the Null nodes?
How are the null nodes treated in the next Iteration? Etc.
Answer: More details have been added to explain the iterative community detection approach represented by Figure 3. For the null node, this has been corrected. There were not supposed to be a null node represented in figure 3
- It would be better to make a table and describe the basic features of the final data.
Section 4.1 is not easily readable in the current stage.
Answer: A table has been created to capture the basic features required for graph representation
- Authors should provide a better explanation of the implications/contribution of the study in the existing literature.
Authors should discuss the Limitations of the current study and how these could be overthrown in a separate Section.
Answer: Both of these concerns have been addressed in the updated manuscript. The introduction has been restructured, more references have been added and the limitations of the study has been added to the discussion section.
Thank you very much for your feedback
Round 2
Reviewer 1 Report
Comments and Suggestions for Authors
Thanks for your great work. I have no more comments.
Reviewer 2 Report
Comments and Suggestions for Authors
It is appreciated to use the Journal template and place contents in appropriate places.
Reviewer 3 Report
Comments and Suggestions for Authors
The authors have made significant modifications. I am satisfied with their efforts.
