A Review on Recent Progress in Machine Learning and Deep Learning Methods for Cancer Classification on Gene Expression Data
, Muhammad Akmal Remli 2,3,*, Nor Syahidatul Nadiah Ismail 1, Mohd Saberi Mohamad 4, Hui Wen Nies 5 and Nor Bakiah Abd Warif 6Round 1
Reviewer 1 Report
Although the authors reviewed about the type of machine learning methods for cancer classification, the article was submitted as “Article” paper not as “Review”. The type of this manuscript is rather “Review” than “Article”. Moreover, several information in this manuscript were incorrect. The points are listed in below.
[Major point]
- Although there was Figure 1, I can’t find sentences about the figure in main text. Furthermore, in Figure 1, the k-nearest neighbor was categorized as both "supervised" and "unsupervised," but the k-nearest neighbor is a supervised learning method. The authors probably mistook the k-means method and the k-nearest neighbor method, but these methods are completely different and need to be corrected.
- The authors cited 30 articles, which is insufficient for writing a review article in this field. The authors should cite papers on prominent machine learning methods.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This is a review on an interesting and hot topic, the use of machine learning for cancer. The review seems too simplicistic, and should be improved: currently, it is limited to a brief description of ML methods and a few paragrphs describing pulished papers, which were selected apparently without a rigorous method. If a search on Pubmed was performed, or other method, this should be described.
There are two aspects that need to be enriched: a desription of the well known drawbacks of machine learning and AI, such as overfitting, difficulty in validation, need for high quality training datasets etc. can be found here:
Appl. Sci. 2021, 11, 1691. https://doi.org/10.3390/app11041691
Second, in order to make the review more thorough, I recommend to comment on the rapid the increase of interest in ML. A summary with statistics (e.g. increase of papers over years, fields of applications... ) can be found here:
Physica Medica Volume 83 Pagine 221-241 https://doi.org/10.1016/j.ejmp.2021.04.010
- In the discussion, comment on why some ML methods are more commonly used.
- Feature selection is required to prevent overfitting. Describe feature selection algorithms
- comment on the issue of interpretability of ML results, how to solve it
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The manuscript has been much improved and is in a nice condition now.
Reviewer 2 Report
The authors responded to the reviewers' concerns and the manuscript can be accepted.
