Personalization of Optimal Chemotherapy Dosing Based on Estimation of Uncertain Model Parameters Using Artificial Neural Network

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have done an excellent job with a paper that is of relevance to our journal and readers. The following queries must be addressed to improve it in the spirit of constructive criticism:

1. The authors must do a better job at emphasizing the relevance of their study in the introduction. If this is a novel approach or type of study for a dreaded complication and problem, they must state it so. If not novel, or does it contribute anything new compared to prior literature?
2. While the mathematical model is very well explained and appears to be flawless based on my interpretation, we are still missing a more clinical application for the discussion. In other words, how can our readers who may not be experts in mathematical models and engineering, and who may be dedicated clinicians benefit from this study and its practical implications?
3. While the results are well explained in the discussion, I recommend preparing a simple visual abstract or graphical algorithm for practical recommendations for our readers.
4. We are missing a limitations paragraph in the discussion.
5. We are also missing a paragraph on future directions and ideas for further research in particular with respect to clinical applications.
6. The conclusion is quite long and needs to be summarized. A lot of what the authors introduce in the conclusion section really belongs in the discussion.

Congratulations to the authors. I look forward to reviewing the revised manuscript if our Editor-in-Chief agrees.

Author Response

See the attached file

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

In this study, the authors proposed a new method for personalizing chemotherapy dosing by combining artificial neural networks with model-based optimization, where the training dataset was generated by simulating the state responses of different virtual patient groups to capture the variability between subjects. Furthermore, the state responses were parameterized using exponential functions to reduce the dimensionality, and a multilayer perceptron artificial neural network was trained to estimate patient-specific model parameters based on the response data of a single chemotherapy dose. This article presents a novel and rational approach to the problem of personalized chemotherapy dosing in this area of cancer treatment.

Some issues that need to be corrected are listed as follows:

1. In Section 1 (Introduction), it lacks description of your technical innovation points and research hypotheses. Besides, for Literature Review, it is obvious that the cited related research work has not kept up with the latest international progress.

2. In Section 2 (Materials and Methods), the assumptions of the studied model should be clearly defined, and the rationality of your assumption and its practical application should also be explained. 

3. In Section 3 (Results), the parameter settings used in the studied artificial neural network models are not clearly listed, which makes the repeatability of the experiments questionable. Besides, there is a lack of reasonable selection basis and sensitivity analysis for selected parameters. It is recommended to explain the basis for selecting the initial values and analyze the sensitivity of different initial values to the calculated results.

4. For your numerical results, no performance comparison with other studies was conducted. You should compare your method with the latest related works and highlight the innovation and scientific contribution of the current article.

5. For your training samples selection, the description of your data preprocessing is not detailed and precise enough. For example, the division of training set and validation set in the studied dataset is not explained clearly. Additionally, no separate validation set was created, so it is impossible to verify whether the proposed model showed signs of overfitting.

6. You need to double-check all mathematical formulas and complex derivations, as there are some typos in the current manuscript.

Author Response

See the attached file

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The authors developed a personalized chemotherapy dosing strategy using a multilayer perceptron artificial neural network to estimate patient-specific tumor and drug response parameters, enabling optimized treatment that reduces uncertainty and minimizes drug use compared to conventional population-based approaches. The analysis is detailed, and the results appear promising. I recommend a minor revision.

 

Suggestions for Improvement:

• Introduction: It would be beneficial to emphasize the disadvantages of chemotherapy and the advantages that precision medicine could offer.
• Line 75: The literature survey needs to be more detailed. The authors should clearly identify the scientific gaps and explicitly state what this paper aims to address.
• Section 2: A biological explanation of natural killer cells, cytotoxic T cells, and tumor cells would be helpful.
• Line 262: Why was the MLP neural network chosen for this problem? Has it been used in similar applications before?
• Line 299: Are the parametric uncertainties chosen arbitrarily, or is there a specific rationale behind them?
• Line 318: Provide a more detailed explanation of how data augmentation is performed.
• Results: The results appear promising, but how do they compare to existing methods in the literature? This comparison should be included.

Author Response

See the attached file

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have done an excellent job at improving their manuscript in a diligent and respectful manner. I agree with publishing the paper in its current state.

Author Response

We sincerely appreciate the time and effort the reviewer has dedicated to evaluating our manuscript. The insightful comments and constructive suggestions have significantly improved the quality and clarity of our work. 

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have well responded to my previous questions and made significant improvements. Although the quality of this submission has been significantly improved, the text duplication ratio (21%) is too high. I suggest that the authors revise the paper again to reduce its duplication rate, and this submission could be accepted for publication as long as it meets the format requirements of Applied Sciences.

Comments on the Quality of English Language

The text duplication rate (21%) is too high, so I suggest the authors revise their paper again to reduce the duplication rate.

Author Response

Dear Reviewer,

We are aware that two specific paragraphs concerning the model and its assumptions were adopted from the literature and cited, which resulted in an increased text duplication ratio. To reduce the overlap and lower the duplication ratio, we have rephrased these paragraphs while maintaining their exact technical content.

See the revised manuscript.

Unfortunatelly we do not have access to plagiarism-checking software or a subscription to such a service, so we are unable to check for text overlap in the revised manuscript on our own.

We hope that we have correctly identified the problematic paragraphs and that no further refinements will be necessary.

If otherwise, please indicate us specifically.