A support vector machine and a random forest classifier indicates a 15-miRNA set related to osteosarcoma recurrence

Area of Science:

• Oncology
• Genetics
• Bioinformatics

Background:

• Osteosarcoma is a primary bone cancer originating in mesenchymal tissue.
• Metastasis and recurrence are key factors contributing to high osteosarcoma mortality rates.

Purpose of the Study:

• To identify differentially expressed microRNAs (DEmiRs) in osteosarcoma.
• To develop a predictive model for osteosarcoma recurrence.
• To explore potential mechanisms of osteosarcoma metastasis and recurrence.

Main Methods:

• Downloaded and analyzed three miRNA expression profiles from GEO DataSets.
• Screened DEmiRs using MetaDE.ES and constructed a support vector machine (SVM) classifier.
• Validated the SVM classifier's prediction efficiency on independent datasets and constructed a co-expression network.

Main Results:

• Identified 78 significantly DEmiRs.
• Developed a 15-miRNA SVM classifier with high accuracy (84.62%-91.3%) in predicting osteosarcoma recurrence across datasets.
• Found four specific miRNAs (hsa-miR-10b, hsa-miR-1227, hsa-miR-146b-3p, hsa-miR-873) significantly correlated with recurrence time.

Conclusions:

• A 15-miRNA-based SVM classifier shows promise for predicting osteosarcoma recurrence.
• Identified potential mechanisms underlying osteosarcoma metastasis and recurrence.
• Highlighted novel DEmiRs as potential biomarkers or therapeutic targets for osteosarcoma.