Next Article in Journal
The Effect of Blade Angle Deviation on Mixed Inflow Turbine Performances
Previous Article in Journal
On the Safe Deployment of Matrix Multiplication in Massively Parallel Safety-Related Systems
 
 
Article
Peer-Review Record

Deep Learning Algorithm for Heart Valve Diseases Assisted Diagnosis

Appl. Sci. 2022, 12(8), 3780; https://doi.org/10.3390/app12083780
by Santiago Isaac Flores-Alonso 1,†, Blanca Tovar-Corona 2,† and René Luna-García 1,*,†
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Appl. Sci. 2022, 12(8), 3780; https://doi.org/10.3390/app12083780
Submission received: 25 February 2022 / Revised: 25 March 2022 / Accepted: 6 April 2022 / Published: 8 April 2022
(This article belongs to the Topic Artificial Intelligence in Healthcare)

Round 1

Reviewer 1 Report

The proposed paper is of highest interest and is scientifically sound.

Best regards

Author Response

Response to Reviewer 1

Dear Reviewer, 

Thank you very much for your evaluation and comments on the paper “Deep Learning Algorithm for Heart Valve Diseases Assisted Diagnosis”. Attending your suggestion about improving the results, we made some changes by correcting the language style. This changes are marked in the new version in light green colour.

Reviewer 2 Report

The study deals with building classification models for four common valvular pathologies and normal heart sounds. The study is of great importance to the journal as well as to the research community. Overall the paper is very well written and has needed rigor, my only comment is on the handling unbalance dataset. The below statement needs multiple references to support the handling of unbalanced nature of the dataset.

"However, segregating the dataset into a greater number of windows as proposed [10], increases this unbalance, biasing the classifier"

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

This article applies the deep learning classification algorithms to assist the diagnosis of cardiac valve, effectively to solve the defects of artificial diagnosis. The algorithm utilizes the frequency change in the entire cardiac cycle to carry out the binary classification task of normal and abnormal heart conditions. First, with the frequency domain signal processing method, the data of time series is transformed to the image data, and then the deep learning neural networks are used to perform the classification task. Secondly, by using the migration learning method, the task of multiple classifications becomes the binary classification task, which can improve the classification accuracy. Finally, the parallel feature extraction process with multiple models is used to achieve the high-efficiency and high-precision feature extraction.

 

It is an interesting paper, which can be accepted for publication after minor revision. Some comments are given as follows:

 

  1. The introduction part is lack of the introduction of enough relevant works, although the scientific problem is proposed. The contents introduced are not enough, which are unable for readers to understand the research background and the latest research results related to the diagnosis of cardiac valve. It is recommended to add a systematic reference review on the methods of the artificial diagnosis of cardiac valve, which references are actually given in the Discussion.

 

  1. The innovation is not clearly presented. In the paper, the six models discussed are typically the basic neural network models, with a combination model. It is recommended that the authors should discuss the major differences between their model structure and other models in the literature to clearly present the innovation of this work.

 

  1. It is recommended for the authors to discuss the possible future work to increase the integrity and depth of the article.

 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Back to TopTop