Detecting Patient Health Trajectories Using a Full-Body Burn Physiology Model

Round 1

Reviewer 1 Report

Abstract:

The authors meet the requirements regarding the number of words and structure of the abstract. However, the abstract should portray what was done throughout the research, namely Contextualization (exposition of the question addressed and highlighting the objective of the work developed), methods: Briefly describe the main methods or treatments applied, results: summarize the main results of the article and conclusion: Indicate the main conclusions or interpretations. In the abstract presented, the lack of contextualization as well as the conclusions is quite evident.

Furthermore, the verb tense used is not the most correct. In the very first sentence:

 “We leverage a whole-body physiology model of burn inflammation injury to train an algorithm to properly detect patient states” should be “A whole-body physiology model of inflammatory burn injury was used to train an algorithm to correctly detect patient’s states”.

This error is detected throughout the text 63 times.

Keywords: Keywords have some shortcomings. Authors should keep in mind that keywords should be specific to the article, but reasonably common within the subject discipline. Biology for example does not seem to me to be something extremely necessary and specific to this article.

The following words are suggested: Biological system modeling, Full-Body model,

BioGears Engine, Physiology,Computational modeling, Clustering, Interleukin, Burn, Inflamamation

Introduction:

The introduction is well written and enlightens the reader in regard to the context and problem addressed in the work presented in this paper. It is suggested that it be changed and rewritten since it is extremely similar to the one presented in the paper published in 2019: "A Full-Body Model of Burn Pathophysiology and Treatment Using the BioGears Engine*". Submitting to a plagiarism detector the entire introduction shows a 75% plagiarism rate. This is not acceptable. Furthermore, this document should be added to the references.

Materials and Methods:

This section is a bit confused with the contextualization. It is not intended. It should only contain the description of the methodology. We suggest changing the first sentence to the introduction section stating the objective.

General Sugestions:

Correct the verbal tenses by avoiding the use of the “I/We”. Lines 13,16,17,20,41,44,48,50,51,55,56,57,.....

Acronyms should be used only after they are written in full. For example "eXtensible Markup Language (XML)".

Improve the English, opt for shorter sentences that are easily understood by the reader

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

The paper describes a pipeline relying on both mathematical and learning models to classify patients suffering from thermal injury into different outcome classes. The initial part of the manuscript deals with the ODE framework used to generate the synthetic data needed for training the learning model. Indeed, this first part appears to be the more solidly grounded, while the following class definition and ML model are rather weak.

Lacking a real ground truth, patient category is attributed to K-mean clustering on a UMAP data reduction, and the learning workflow relies on a simple NN model within a 4-fold CV. As major flaws, I thus point out the rather arbitrary class attribution, the lack of a robust validation on an independent testing portion of the dataset, the weak reproducibility of the whole learning pipeline due to the too simple 4-CV resampling strategy and the difficult explainability of the resulting model. Indeed, comparison w/ other models/strategies and use of RWD would be also required to ensure scientific soundness to the manuscript which, in its present form, is still at a preliminary stage, with limited contributed novelty. Finally, I would strongly suggest to revise the literature about ML prediction of human thermal injury, since there are a number of relevant publications neither discussed nor even listed in the references.

Minor:

- Missing ref in line 241

Author Response

Please see the attachment. Really great review, i think that we got to everything. Very excited to deliver an update manuscript! 

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The paper have been updated and questions have been answered.

Reviewer 2 Report

All raised issues have been reasonably solved.