Permutation-based identification of important biomarkers for complex diseases via machine learning models

Abstract

Study of human disease remains challenging due to convoluted disease etiologies and complex molecular mechanisms at genetic, genomic, and proteomic levels. Many machine learning-based methods have been developed and widely used to alleviate some analytic challenges in complex human disease studies. While enjoying the modeling flexibility and robustness, these model frameworks suffer from non-transparency and difficulty in interpreting each individual feature due to their sophisticated algorithms. However, identifying important biomarkers is a critical pursuit towards assisting researchers to establish novel hypotheses regarding prevention, diagnosis and treatment of complex human diseases. Herein, we propose a Permutation-based Feature Importance Test (PermFIT) for estimating and testing the feature importance, and for assisting interpretation of individual feature in complex frameworks, including deep neural networks, random forests, and support vector machines. PermFIT (available at https://github.com/SkadiEye/deepTL) is implemented in a computationally efficient manner, without model refitting. We conduct extensive numerical studies under various scenarios, and show that PermFIT not only yields valid statistical inference, but also improves the prediction accuracy of machine learning models. With the application to the Cancer Genome Atlas kidney tumor data and the HITChip atlas data, PermFIT demonstrates its practical usage in identifying important biomarkers and boosting model prediction performance.

Precision Oncology: Artificial Intelligence and DNA Methylation Analysis of Circulating Cell-Free DNA for Lung Cancer Detection

“…Recent advances in machine learning (a branch of AI) point to a significant potential for future impact on medical research and practice. It has been noted that AI methods could potentially make significant contributions in the medical field in the following areas: understanding “disease underlying architecture, perform early diagnosis of diseases, and disease progression prediction” ( 43 ).…”

Identification of novel candidate biomarkers and immune infiltration in polycystic ovary syndrome

“…The LASSO logistic regression model selects biomarkers using LASSO in the R package glmnet [ 10 ]. SVM-RFE is an iterative approach combining linear support vector machines with feature selection and backward elimination, which is implemented with the R packages e1071 , kernlab , and caret [ 8 ]. To further assess the diagnostic ability of candidate biomarkers, the receiver operating characteristic (ROC) curve and calculations of its area under the curve (AUC), accuracy, sensitivity, and specificity were performed using the R package pROC .…”

“…Support vector machine-recursive feature elimination (SVM-RFE), a sub-method of machine learning, offers an advantage in explaining the strength and direction of interactions between predictors and outcomes by RFE of non-linear kernels [ 8 ]. CIBERSORT, a gene expression-based deconvolution algorithm, assesses immune cell infiltration signatures [ 9 ].…”

Significance of Immune-Related Genes in the Diagnosis and Classification of Intervertebral Disc Degeneration

“…Each data point as a point in n-dimensional space was plotted. And we were able to identify the most suitable hyperplane to distinguish the two classes (normal and degenerated discs) well [17][18][19]. Finally, we utilized ROC curves and area under the curve (AUC) to evaluate the predictive accuracy of these 11 IRGs [20].…”