Thursday 27 March 2025
Scientists have long struggled to accurately classify genes based on their expression levels, a crucial step in understanding complex biological processes and diagnosing diseases. This challenge arises because gene expression data is often high-dimensional, meaning it contains many more features (or variables) than samples. As a result, machine learning algorithms can get overwhelmed and struggle to make accurate predictions.
A team of researchers has now developed a new approach that tackles this problem by combining two powerful tools: Boruta, a feature selection algorithm, and LIME, an explainable artificial intelligence technique. The resulting method, called BOLIMES, is designed to identify the most informative genes while providing insights into how these genes contribute to the classification of gene expression data.
The key innovation behind BOLIMES is its ability to iteratively refine the feature subset by combining Boruta’s robustness with LIME’s interpretability. Boruta filters out non-informative genes by comparing each feature against its randomized counterpart, ensuring that only valuable information is retained. LIME then ranks the remaining genes based on their local importance to the classifier, providing a clear picture of how each gene contributes to the classification outcome.
In testing BOLIMES on multiple datasets, the researchers found that it outperformed traditional machine learning methods in terms of both accuracy and feature selection quality. The method was able to identify smaller subsets of genes that were sufficient for accurate classification, reducing the risk of overfitting and improving model interpretability.
The implications of BOLIMES are significant, particularly in the context of precision medicine. By accurately identifying the most informative genes associated with a particular disease or condition, clinicians can develop more targeted treatments and improve patient outcomes. Moreover, the method’s ability to provide insights into gene expression patterns can help researchers uncover new biological mechanisms and identify potential therapeutic targets.
While BOLIMES is still an early-stage approach, its potential is vast. As scientists continue to explore the complexities of gene expression and develop new machine learning algorithms, this innovative method will likely play a key role in advancing our understanding of complex biological systems and improving human health.
Cite this article: “New Approach Combines Feature Selection and Explainable AI to Accurately Classify Gene Expression Data”, The Science Archive, 2025.
Gene Expression, Machine Learning, Feature Selection, Boruta, Lime, Bolimes, Artificial Intelligence, Precision Medicine, Bioinformatics, Genomics
