Researchers designed a multiclass cancer diagnostic tool using AI, MicroRNA
(Nanowerk News) Cancer is one of the deadliest diseases in the world. By 2023, more than 1.9 million new cancer cases and 609,820 deaths are projected to occur in the United States alone. While efforts are being made to improve diagnostic tools, microRNAs are at the forefront of biomedical research.
MicroRNAs, or miRNAs, are a class of small non-coding ribonucleic acids (RNA), which are essential for all biological functions. The main role of miRNAs in the human body is gene regulation. Thus, they regulate various biological and pathological processes, including cancer formation and development. In fact, many cancers are closely related to miRNA function.
The association of miRNAs with cancer development has prompted interest in investigating miRNA expression profiling data as a potentially less invasive diagnostic tool for early detection. Machine learning methodologies have been used to develop high-performance pan-cancer classification models and to identify potentially novel miRNA biomarkers for clinical investigations. However, understanding how these data science techniques correlate with established biological processes to advance integration into the clinical environment is key.
To further explore the feasibility of miRNAs as biomarkers for cancer classification and enhance clinical classification applications, researchers from Florida Atlantic University’s College of Engineering and Computer Science created a multiclass cancer diagnostic model using miRNA expression profiling. Their methodology employs an iterative process that applies several key techniques to an ever-increasing dataset of miRNA expression quantification datasets.
For this study, the researchers assessed how the top miRNA features selected by the machine learning model relate to clinically and biologically verified miRNA biomarkers. They developed a Support Vector Machine and Random Forest machine learning model for cancer classification, and iteratively added cancer classes to the multiclass model. They looked at the relationship between the relevant miRNAs identified through feature selection and the performance metrics of the classification model across 20 iterations. Each iteration adds another key sample site to the multiclass model, increasing the number of cancer types involved.
The researchers examined success metric changes as more cancer types were introduced to a subset, how 20-miRNA signatures changed when more cancer types were introduced to a subset, and characteristics of complete data sets through principal component analysis, a popular technique for analyzing large data sets containing a large number of dimensions or features.
Unlike previous studies, which focused solely on miRNA feature signatures for the final multiclass data set, this study tracks changes in clinical and biological relevance after each addition of cancer tissue type.
The results of the research were published in the journal of the Institute of Electrical and Electronics Engineers IEEE Xplore (“Exploration of the Relevance of MicroRNA Signatures for Cancer Detection and Classification of Multiclass Cancers”), indicating that models with a greater number of cancer classes shift towards focusing on cancer-diverse miRNAs with greater relevance to the functionality being characterized. The study suggests that miRNAs may be highly unique to certain cancer tissues and can be powerful biomarkers for detection and classification; however, currently verified biomarkers lead to a wider range of miRNAs for cancer when detecting cancer.
This study provides insight into the potential relationship between the overall clinical relevance of feature extraction signatures and model success metrics and demonstrates the feasibility of using multi-tissue miRNA cancer signatures as generalizable signatures for detection of a single class of cancer in a number of prominent cancers.
Findings indicated that as the number of cancer classes increased, performance metrics decreased, but the percentage relevance of miRNA feature selection marks increased slightly before stabilizing. In addition, after carrying out principal component analysis, the non-cancerous tissues of all samples had very similar expression visualizations, whereas all the cancerous tissues had unique profiles.
“MicroRNAs hold significant promise for future diagnostic tests due to their direct detection from biological fluids such as blood, urine or saliva as well as the availability of high-quality measurement techniques for miRNAs,” said Oneeb Rehman, correspondent author and Ph.D. candidate in the Department of Electrical Engineering and Computer Science in FAU’s Faculty of Engineering and Computer Science. “This makes understanding and characterizing the biological basis behind a potential miRNA classification tool important for integration into the clinical environment.”
Under Rehman’s tutelage, a team of senior design undergraduate students and co-authors Charles Briandi and Eyan Eubanks, led by Matthew Acs and Richard Acs, from the Department of Electrical Engineering and Computer Science, participated in the study. Hanqi Zhuang, Ph.D., co-author and chairman and professor of the Department of Electrical Engineering and Computer Science, served as the team’s mentor.
“This study, which explores the relationship between microRNA composition and various types of cancer, has important implications for the potential use of miRNAs as biomarkers in both the research and clinical fields,” said Stella Batalama, Ph.D., dean of FAU’s School of Engineering and Computer Science. “What’s particularly impressive about this research is that it engaged a number of our undergraduate students collaborating to investigate better ways to manage diseases that impact millions of people worldwide each year.”
This study used data from the Genomic Data Commons Data Portal, which is sponsored by the National Cancer Institute.
