EQLC-EC: An Efficient Voting Classifier for 1D Mass Spectrometry Data Classification

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

After reviewing the manuscript "EQLC-EC: An Efficient Voting Classifier for 1D Mass Spectrometry Data Classification' I have the following comments:

1. While the EQLC-EC model combines established techniques like 1D CNNs and voting mechanisms, the novelty of the approach isn't strongly justified. It would help to clearly outline how this method fundamentally improves upon existing methods beyond incremental accuracy gains.

2. The lack of extensive hyperparameter tuning, as mentioned in the methodology, might lead to suboptimal performance evaluations.

3. The paper mentions improvements over state-of-the-art (SOTA) methods but does not discuss why the selected SOTA models were chosen and how their results were reproduced.

4. Table 7: The table does not account for variability in computational time due to repeated runs or differing initial conditions, which would provide a clearer picture of robustness.

5. Figure 5: The figure does not explicitly address whether any dataset suffers from extreme class imbalances (e.g., minority classes with very few samples). This is important for understanding the challenges faced by the model.

6. The results focus heavily on accuracy and F1-score, neglecting other important metrics like precision, recall, or area under the ROC curve (AUC), which are crucial for imbalanced datasets.

7. Figures like box plots (e.g., Figure 8) are used effectively but lack detailed annotations and visual clarity, such as confidence intervals or outlier explanations.

Author Response

Please see response in the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript is about a so-called EQLC-EC model, an ensemble deep learning method that employs a simplified 1D-CNN architecture and voting mechanisms to achieve high accuracy, efficiency, and generalizability in classifying one-dimensional mass spectrometry data. I find the topic very interesting and was particularly impressed by the method’s speed. It also contains significant novel information that warrants publication and will most probably have an impact on the mass spectrometry community. However, I have some concerns about the manuscript that must be addressed before it can be accepted for publication:

• I really like the Introduction chapter because it is concise and summarizes the topic very well. However, in my opinion, “2. Basic Concepts and Research Status” is too broad and lengthy. I can see two solutions: a) move it to supplementary material or b) incorporate a shortened version into the Introduction.
• The manuscript feels very fragmented, with some sections only a few lines long (e.g., sections 6.3 and 6.4). Many parts resembles a summary, for example, many paragraphs start with an expression followed by a colon (e.g., “Mass Spectrometry Platforms:”). This gives the impression that at least parts of the manuscript were written with the help of AI. For instance, an AI detector (originality.ai) flagged lines 247-260 as “100% confident that text is AI-generated.” The Editor of the journal should decide if this is acceptable for publication in Electronics.
• Some of the conclusions are too broad. For example, lines 841–842 state, “Leveraging the parallel processing power of GPUs, even those commonly found in standard workstations, significantly accelerates computation compared to traditional CPU-based methods.” This is a generally true statement, independent of the current manuscript, and also suggests substantial assistance from an AI-based tool.

Minor comments:

• Table 1: What does “sample size” mean? Is it the number of measurements, the number of MS spectra, or the amount of GB data? This should be clarified by the authors.

Author Response

Please see response in the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Title: Expand the acronyms in the title

Line 15 – spell out ‘CNN’

Line 32 – check the use of spaces between words; this is applicable throughout the manuscript

Paragraph starting Line 61 – no need to include this paragraph; the preceding paragraph is more than enough.

2.1 – a good brief description of how MS works.

Given that many readers would becoming from chemistry, terms well-known in computer science or ML should be given a more explicit definition in the text, for example “high dimensionality”.

Sections 2 and 3 can be revamped and the content can be distilled and included in the introduction section.

Line 307 – provide references to these original authors’ publications.

Figures 3-5 – what does each batch represent? Also these figures should be combined into one for simplicity. The caption should contain enough information for the readers to understand without having to refer to the text.

Line 391  - what are TP, TN and FN?

Table 4 – what does the colour grading represent?

Lines 542 – 560 – the text here is redundant given it merely describes the same information that’s been presented in Table 4

Line 563 – this line should be reworded to provide clarity on the grading/shading and be included in Table 5 caption

Figure 6 – acronyms in the figure should be explained in the caption

Table 7 – what the “training” and “testing” columns stand for needs to be explained.

Figure 8 – needs polishing/reformatting – the line of text on the bottom should be included in the caption

Figure 9 – Change “paper” to “literature” to be consistent with the caption. I suggest replacing the “x” mark with something neutral to avoid potential confusion.

Figures 10 – 12 – revise the colouring scheme so each item is presented in a unique colour. The standard deviation panes can be removed and the standard deviation incorporated into the main F1 score pane

 

Overall impression

The manuscript gives the impression of a mini-thesis instead of a research article. Though welcome, some of the background can be removed as it may distract readers from the flow of the manuscript. The flow of the manuscript is not clearly presented – the authors attempt to list many of the parallel statements exacerbated the unclear flow of the manuscript.

The manuscript is written for MS but the manuscript treats the MS datasets just as a special type without giving adequate consideration as to why MS is being investigated. As well, potential users of such tool (eg chemists) should show elevated interests if the authors include benefits of this ML model from chemists’ perspective.

Lastly, the presentation of results (figures and tables) needs some work on both clarity and content.

Author Response

Please see response in the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I appreciate the time and effort put into addressing the reviewer comments and improving the manuscript. The revisions provide much-needed clarity on the novelty of your approach, model optimization, and performance evaluation. The additional explanations, refined figures, and well-structured supplementary materials make the study more transparent and comprehensive. Your responses were thorough, and the manuscript now feels well-rounded and scientifically robust. With these improvements, I believe it is ready for acceptance.

Author Response

We thank the reviewers for their time and effort in reviewing our manuscript. We are pleased to note that they have no further comments in this round, and we confirm that all previous comments have been addressed in the revised manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

The maniscript has been significantly improved, I support its publication.

Author Response

We thank the reviewers for their time and effort in reviewing our manuscript. We are pleased to note that they have no further comments in this round, and we confirm that all previous comments have been addressed in the revised manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

Cover letter

Line 25 – Does “instructions” mean to spell “introduction”?

Line 44 – The two lines explaining the color grading should be moved above the table, as part of its caption.

Line 52 – Regarding Table 7, tables’ caption should be above the table. If more description/information is needed for a certain table, work the text into the manuscript. Here for Tabel 7, the description can be incorporated into the paragraph discussion Table 7.

Overall, the revision addressed my concerns well. The above are simple fixes to perfect the manuscript. One thing to note is that when submitting supplementary information, please follow the journal’s instructions so the SI is available where the main manuscript is.

Author Response

Comments 1: Line 25 – Does “instructions” mean to spell “introduction”?
Response 1: Yes, in my previous response letter, I erroneously spelled "introduction" as "instructions."  

Comments 2: Line 44 – The two lines explaining the color grading should be moved above the table, as part of its caption. Response 2: The description of color grading below the table has been simplified and incorporated into the caption as per the request. The updated caption now reads: "Table 6. Computational efficiency and resource utilization of simplified deep learning networks (deeper green indicates lower resource utilization and higher computational efficiency)." This change can be found at line 431 in the main text. The same modification has been applied to Table 8 at line 483 in the main text.  

Comments 3: Line 52 – Regarding Table 7, tables’ caption should be above the table. If more description/information is needed for a certain table, work the text into the manuscript. Here for Tabel 7, the description can be incorporated into the paragraph discussion Table 7.
Response 3: The description of Table 7 has been extracted and its main content has been integrated into the paragraph in the main text at line 437, as required. The same approach has been applied to Table 9 at line 559 in the main text.

Comments 4: One thing to note is that when submitting supplementary information, please follow the journal’s instructions so the SI is available where the main manuscript is.
Response 4: Thank you for this reminder. We have followed the journal’s instructions for submitting supplementary information. The supplementary materials have been uploaded through MDPI’s system, and we have cited them in the main manuscript using the identifier "Supplementary Materials" at the appropriate locations, as per MDPI’s guidelines.