Version 1
: Received: 21 July 2023 / Approved: 24 July 2023 / Online: 25 July 2023 (05:31:45 CEST)
How to cite:
Mickael, M. A Convolution-Based AI Neural Network Achieves 90% Accuracy in Predicting Cancer Type Based on Copy Number Alterations. Preprints2023, 2023071653. https://doi.org/10.20944/preprints202307.1653.v1
Mickael, M. A Convolution-Based AI Neural Network Achieves 90% Accuracy in Predicting Cancer Type Based on Copy Number Alterations. Preprints 2023, 2023071653. https://doi.org/10.20944/preprints202307.1653.v1
Mickael, M. A Convolution-Based AI Neural Network Achieves 90% Accuracy in Predicting Cancer Type Based on Copy Number Alterations. Preprints2023, 2023071653. https://doi.org/10.20944/preprints202307.1653.v1
APA Style
Mickael, M. (2023). A Convolution-Based AI Neural Network Achieves 90% Accuracy in Predicting Cancer Type Based on Copy Number Alterations. Preprints. https://doi.org/10.20944/preprints202307.1653.v1
Chicago/Turabian Style
Mickael, M. 2023 "A Convolution-Based AI Neural Network Achieves 90% Accuracy in Predicting Cancer Type Based on Copy Number Alterations" Preprints. https://doi.org/10.20944/preprints202307.1653.v1
Abstract
The accurate identification of the primary tumor origin in metastatic cancer cases is crucial for guiding treatment decisions and improving patient outcomes. Copy Number Alterations (CNA) have emerged as valuable genomic markers for predicting the origin of metastases. This research article presents a comprehensive analysis of CNA-based prediction models in metastatic cancer. The study utilizes a dataset comprising CNA profiles from twenty different cancer types and employs advanced AI-based techniques for prediction. The research workflow consists of two models. The first is convolution network-based, while the other is a multi-stage model. In the first stage, a RELU model is developed to differentiate between 16 cancer types, considering their CNA, chromosome location, strand, and overall location. The second stage involves building specialized models to detect differences between cancer types within the same location or organs, such as brain lower-grade glioma and glioblastoma multiforme. Both generalized models achieve an overall accuracy of over 90%, while the specialized models achieve an accuracy of approximately 95% with minimal loss values. The results demonstrate the potential of AI-based approaches utilizing CNA data for improved diagnosis and personalized treatment strategies in metastatic cancer. This research establishes a solid foundation for future advancements in the field, paving the way for more effective and targeted approaches in metastatic cancer management.
Keywords
metastasis; cancer; copy number alterations; AI
Subject
Biology and Life Sciences, Life Sciences
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
