Using Proteomics Data to Identify Personalized Treatments in Multiple Myeloma: A Machine Learning Approach

Area of Science:

• Biomedical Informatics
• Computational Biology
• Oncology

Background:

• Multiple myeloma (MM) treatment selection is challenging.
• Personalized medicine requires identifying patient-specific drug responses.
• Proteomic profiles offer potential biomarkers for treatment sensitivity.

Purpose of the Study:

• To develop a machine learning (ML) decision support system for personalized MM treatment.
• To predict patient sensitivity or resistance to chemotherapeutics based on proteomic data.

Main Methods:

• Feature selection from proteomic data to identify dominant parameters.
• Classification algorithms (e.g., Random Forest, SVM) were compared.
• Data-balancing techniques were explored due to small cohort size.

Main Results:

• Proteomics data utilization is a promising strategy for MM treatment selection.
• The ML system achieved an average accuracy of 81% in predicting treatment response.
• Pilot study demonstrated feasibility despite a small patient cohort (39 patients).

Conclusions:

• ML-driven analysis of proteomic profiles can guide personalized chemotherapy for MM.
• Further validation with larger cohorts is warranted to refine ML models.
• This approach holds significant promise for advancing precision oncology in MM.