Back

Deep Learning Solutions for Pneumonia Detection: Performance Comparison of Custom and Transfer Learning Models

Zhong, Y.; Liu, Y.; Gao, E.; Wei, C.; Wang, Z.; Yan, C.

2024-06-21 public and global health
10.1101/2024.06.20.24309243 medRxiv
Show abstract

Withdrawal StatementThe authors have withdrawn this manuscript because of the following reasons: Significant Methodological Errors: Upon further review of our work, we discovered critical methodological errors in the data preprocessing stage that affect the reliability and reproducibility of the results. These errors could mislead other researchers if left unaddressed. Inaccurate Data Interpretation: In addition, we have found that some of the interpretations of the results, especially concerning model performance, were not adequately supported by the data. This misinterpretation has led to conclusions that do not align with our current understanding of the study. Ethical Considerations: After further reflection, we have recognized the need to clarify certain ethical aspects of our data collection process, especially regarding the dataset used. This issue needs to be addressed in compliance with ethical guidelines before the findings can be deemed credible and citable. New Research Developments: Finally, new developments in related research have emerged, which contradict some of the conclusions presented in our manuscript. As we plan to pursue a different direction that incorporates these new findings, we feel that it is necessary to retract the current manuscript. After careful reconsideration, my co-authors and I have identified several issues that we believe warrant the withdrawal of the manuscript. Therefore, the authors do not wish this work to be cited as reference for the project. If you have any questions, please contact the corresponding author.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.