Review Reports - Ultrasound-Assisted Extraction of Antioxidant Compounds from Pomegranate Peels and Simultaneous Machine Learning Optimization Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The study explores ultrasound-assisted extraction (UAE) of bioactive compounds from pomegranate peels by optimizing ethanol concentration, ultrasound power, and extraction time. Using response surface methodology and machine learning models, specifically the Random Forest model, the authors identified ethanol concentration as the main factor influencing the yield of antioxidants and phenolic compounds. The paper shows how combining green extraction technologies with artificial intelligence tools can improve efficiency and support the sustainable use of agri-food waste. By integrating artificial intelligence into the optimization of extraction processes, the study brings an innovative and modern approach to the field of natural resource valorization.

I have minor comments :

There are minor typing errors – see also the PDF.
The text should be standardized regarding ultrasound-assisted extraction, which sometimes appears as UAE and sometimes as UBAE.
In Materials and Methods, section 2.5 “Bioactive Compounds Quantification,” for the quantification of TPC, AAC, etc., the cited reference is an article that itself cites the original method. For clarity and easier understanding, I recommend either to briefly describe the method used or to also cite the original reference (see also comments in pdf).
In table 4 add standard deviatrion for the content determinations.
In table 4 I suggest rearranging the table columns so that the composition parameters (e.g., TPC, TFC, TAC, AAC) appear first, followed by the antioxidant activity results (FRAP, DPPH). In addition, I suggest standardizing all content results to mg/g dw (or clearly indicate the conversion used) to ensure easier comparison between parameters. It is also recommended to report values with two decimal places for clarity and consistency.
Table 7 contains some empty cells. I recommend filling these cells with NA ( or -) where data are unavailable or the correlation is undefined.

Comments for author File: Comments.pdf

Author Response

We would like to thank the reviewer for his/her careful reading and insightful comments.

I have minor comments :

1. There are minor typing errors – see also the PDF.

PDF comments:

Abstract Line 16: Both UBAE and UAE are mentioned as extraction methods in the text; for clarity and consistency, they should be referred to in a uniform way

We acknowledge the reviewer’s concern and have revised the manuscript for consistency. UAE is now used throughout.

Line 154: “labeled X₁”

The typographical error was corrected, as requested.

2. The text should be standardized regarding ultrasound-assisted extraction, which sometimes appears as UAE and sometimes as UBAE.

The authors understand the reviewer’s concern about consistency, thus revised the manuscript accordingly. UAE is the term that was kept in the manuscript.

3. In Materials and Methods, section 2.5 “Bioactive Compounds Quantification,” for the quantification of TPC, AAC, etc., the cited reference is an article that itself cites the original method. For clarity and easier understanding, I recommend either to briefly describe the method used or to also cite the original reference (see also comments in pdf).

We recognize the reviewer’s concern. To avoid unnecessary duplication of widely used methods, we kept the descriptions concise but clarified them as recommended.

4. In table 4 add standard deviation for the content determinations.

All standard deviation values were added in Table 4, as suggested.

5. In table 4 I suggest rearranging the table columns so that the composition parameters (e.g., TPC, TFC, TAC, AAC) appear first, followed by the antioxidant activity results (FRAP, DPPH). In addition, I suggest standardizing all content results to mg/g dw (or clearly indicate the conversion used) to ensure easier comparison between parameters. It is also recommended to report values with two decimal places for clarity and consistency.

The columns were rearranged and all values are now presented with two decimals, as recommended.

6. Table 7 contains some empty cells. I recommend filling these cells with NA ( or -) where data are unavailable or the correlation is undefined.

The typographical error was corrected, as requested. The empty cells of Table 7 should have a “-“ in them.

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,

We congratulate your well-done job on the complex data from wet lab to the implementation of AI for theoretical prediction. Your work offers the advantageous side of machine learning to solve big probabilities in the extraction techniques. However, please be advised that some considerations on some parts, including the title, abstract, introduction, as well as the discussion, need to be taken into account to improve the report. Please find the attached file to retrieve the details.

Good luck.

Comments for author File: Comments.pdf

Author Response

Dear authors,

Good luck.

Brief summary

An article entitled “Ultrasound-Assisted Extraction of Bioactive Compounds from Pomegranate Peels and Simultaneous Machine Learning Optimization Study” authored by Martha Mantiniotou, Vassilis Athanasiadis, Konstantinos G. Liakos, Eleni Bozinou, and Stavros I. Lalas provided optimised ultrasound-assisted extraction (UAE)’s parameters for Pomegranate peels (PPs) phytochemicals recovery. Interestingly, the study carried out an artificial intelligence approach using machine learning to predict the antioxidant and phytochemical responses by the UAE. Indeed, the it appears as the novelty of this study that could be considered as an empiric reference for those who work on the related field. This approach had highlighted both major and minor factors affecting the outcome, including solvent concentration and ultrasonic power & extraction, time respectively. Some points as presented in the table below could be thought to help improve the report to be more informative for readers from broader backgrounds.

We would like to thank the reviewer for his careful reading of our work and his/her favorable feedback.

Specific comments

Title: The term of Bioactive Compounds is most likely to be bias from medicinal chemistry. Bioactive seems too general, while your concern is only antioxidant.

Compound can be a species of chemical grouped in any classes, whereas your work was on phytochemical group. Antioxidant phytochemical group could be an alternative one.

The word “Bioactive” was replaced by “Antioxidant”, as suggested.

Abstract at lines 26-27: What do extraction concentration and solvent composition mean?

The typographical error has been corrected.

Abstract at lines 30-33: See your statement: “Beyond laboratory-scale validation, the proposed AI-assisted framework can be directly extended to industrial UAE workflows, supporting intelligent process monitoring, adaptive optimization, and integration with cyber–physical production systems. This positions the approach as a decision-support tool for sustainable scale-up.” Did it come from any of you attempt to prove the expression? If so, providing relevant data is required in your result and discussion section. Otherwise, please consider not to express any assumption on your abstract.

The statement was removed from the abstract section, as requested.

Introduction at line 71: We are most likely to be sure that previous studies on UAE of PPs were documented. Please provide your thought on the UAE experiments to showcase any gap that urge the importance of your work.

The purpose of this study was to compare the two optimization models in order to understand whether AI models can be applied equally or even individually to such tasks. The introduction has been modified to better express the authors' intention.

Lines 92 – 93: Among the mentioned algorithm types, Random Forest was chosen in your work. It is still unclear on why this work relied on RF.

Seven algorithms were tested (RF, ET, MLP, SVR, KNN, PLS, Ridge). RF consistently outperformed the others, as detailed in Section 3.6.

Line 154: See your phrase: labeled z X1, What does z mean?

There was a mistyping, the word “as” should be there instead of “z”. The text was revised accordingly.

Formula 2: See your formula for TPC estimation. What does C_TP stand for? Please do the same addition of definition of other Cs for the rest of estimation.

C_TP stands for total polyphenol concentration in mg GAE/L dw. Definitions were added into the manuscript, as asked.

Lines 175 – 176: See your statement: A calibration curve of ascorbic acid with concentrations ranging from 0 to 500 mg/L was utilized to assess the results. When reading your similar statement at the lines 179-180 where R² was displayed, a contradiction is found due to inconsistency of the statistical information. When R² values of standard curve matters, these should be explicitly stated on the related expression, including that of gallic acid standard curve, etc.

R² values of all calibration curves used were provided, as suggested.

Lines 223-224: It is interesting that TPC, TFC, TAC, and AAC were included in antioxidant responses instead of the traditional term of phytochemical content. Do you have any justification on this?

This study focuses on the antioxidant properties of pomegranate peels, so, all the responses are examined for their antioxidant potential. All these compounds are proved to possess that kind of properties, and their ability to scavenge free radicals is also supported by the results.

Lines 305-308: Expressing your abbreviation for the parameters is enough as you have once already stated the complete terms above.

The manuscript was revised accordingly.

Lines 441-444: See your statement: RF was selected as the working regressor because it achieved the strongest average.. Did it come up from your own comparison with other algorithms or from general knowledge?

Seven algorithms were tested (RF, ET, MLP, SVR, KNN, PLS, Ridge). RF consistently outperformed the others, as detailed in Section 3.6.

Lines 683-684: See your phrase: reported values for six phytochemical and antioxidant response variables, TPC, FRAP, DPPH, TFC, TAC, and AAC. Here, we found a different term representing the six parameters. It looks like a contradiction from the previous term of antioxidant response only. So, why did you not be strict on mentioning the phytochemical content for those except the FRAP and DPPH?

We removed the term “phytochemical” to maintain consistency and avoid confusion.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

Your efforts in addressing our comments are appreciated. Most of the responses have met the required points. However, to improve the consistency of the report, please make sure that:
(1) tested concentration range and R²value of calibrated curve for cyanidin 3-O-glucoside have been added (lines 203-204).
(2) re-double check for any typo or misspelling.

This little addition will complete your manuscript for further consideration.

Cheers.

Author Response

Dear Authors,

Your efforts in addressing our comments are appreciated. Most of the responses have met the required points. However, to improve the consistency of the report, please make sure that:

Thank you for your valuable comments and for acknowledging our revisions.

(1) tested concentration range and R²value of calibrated curve for cyanidin 3-O-glucoside have been added (lines 203-204).

In our study, the total anthocyanin content (TAC) was determined spectrophotometrically using the molar extinction coefficient of cyanidin 3‑O‑glucoside (ε = 26,900 L/(mol·cm)) together with its molecular weight and dilution factor, as described in Equation (4). This approach directly applies Beer–Lambert’s law and does not require a calibration curve. Consequently, a tested concentration range and value are not applicable for this assay. To avoid confusion, we have added in the text a state: “Quantification was performed using the known extinction coefficient of cyanidin 3‑O‑glucoside; therefore, no calibration curve was required and R² reporting is not applicable.”

(2) re-double check for any typo or misspelling.

We have carefully re-checked the manuscript and corrected minor typographical errors and inconsistencies in unit formatting, symbols, and terminology to ensure clarity and consistency throughout.

This little addition will complete your manuscript for further consideration.

Cheers.

We appreciate your guidance, which has helped us strengthen the manuscript. We believe these clarifications and corrections address your concerns and improve the consistency of the report.