A New Approach to Quantify Soccer Players’ Readiness through Machine Learning Techniques

Round 1

Reviewer 1 Report

The submitted article, focusing on the utilization of ML methods for predicting the preparedness of football players, brings a new perspective on indirect assessment using GPS and IMU. The introduction provides sufficient information, including limitations regarding the evaluation of PL. Data collection, preprocessing, and the application of ML methods are described at a good level. For the purpose of evaluation, the authors have chosen the z-score transformation of PL, which they explain logically. In terms of statistical analysis, the authors employ linear mixed models, which are becoming popular, especially due to their implementation in statistical tools. Personally, I have a question regarding whether the PL results follow a normal distribution. This information could be further included in the article itself. Although the discussion addresses many questions that arise from the text, often self-critically, particularly regarding the absence of fatigue indicators, it would be appropriate for the authors to partially supplement the article with information related to the activities of football players that are dependent on the course of the season/week and could have an impact on PL. Additionally, I recommend improving the quality of the images in the results chapter.

Author Response

Dear Reviewer,

first of all, we would like to thank you for your kind feedback. You will find the new document with the corrections made according to your suggestions.

First, you can find additional information related to the activities of football players (line 422-428) that we highlighted in red. 

In addition, we improved the quality of the images in the results section. 

Regarding the normal distribution analysis of PL results, there is no requirement for the response variable to be normally distributed in linear mixed model, therefore, we did not insert this information. 

 

 

 

Reviewer 2 Report

The paper  analyzes ML approach, used to predict the players’ PL. The research defines digital procedures and tests the algorithms performance to detect readiness in the athletes. The Overall merit in ML studies, the significance of content, the quality and the scientific approach of the paper is high. I suggest to improve the introduction and conclusions. Moreover, it could be better (if it is possible) to relate data to geographic positions (soccer field)

Author Response

Dear Reviewer,

first of all, we would like to thank you for your kind feedback. You will find the new document with the corrections made according to your suggestions.

We improved the introduction and conclusions as you suggested. You can check all changes highlighted in red. Unfortunately, we cannot relate data to geographic positions.

Reviewer 3 Report

Authors should be commended on a well written manuscript. This was a very interesting topic and I look forward to similar publications from this group of researchers. I have minor comments below:

 

Introduction:

 

Line 31-32: Sentence is confusing as written. As reads as internal load is instead considered to be the external load. Please consider rephrasing.

 

Line 33: Maybe useful to provide metrics of internal load to highlight the difficulty of assessing.

 

Methods:

 

Line 87-88: May have missed it but males or female participants?

 

Line 139- First, partial, individual, and rehabilitation sessions should be described in more detail and full explanation of why they were not included. What was the cutoff time for a partial session (what if they only missed the final 5 minutes of training was that session completely excluded), is First the first session of the season or the week or the day?

 

Line 141- Should more categories of sessions be considered? Even within a training session there are different focuses (tactical, conditioning, match simulation).

Author Response

Dear Reviewer,

First of all, we would like to thank you for your kind feedback. You will find the new document with the corrections made according to your suggestions.

Line 31-32: Sentence is confusing as written. As reads as internal load is instead considered to be the external load. Please consider rephrasing.

Line 33: Maybe useful to provide metrics of internal load to highlight the difficulty of assessing.

• According to another reviewer's request, we improved the introduction, therefore, we removed the sentence you mentioned.

Line 87-88: May have missed it but males or female participants?

• We inserted this information (line 84)

Line 139- First, partial, individual, and rehabilitation sessions should be described in more detail and full explanation of why they were not included. What was the cutoff time for a partial session (what if they only missed the final 5 minutes of training was that session completely excluded), is First the first session of the season or the week or the day?

• According to your request, we added this information (line 137-142)

Line 141- Should more categories of sessions be considered? Even within a training session there are different focuses (tactical, conditioning, match simulation).

• For the purpose of this study, we considered only two types of sessions: training and matches. In future studies, we could improve the ML model's accuracy considering this type of information.