Toward rapid learning in cancer treatment selection: An analytical engine for practice-based clinical data

Area of Science:

• Oncology
• Medical Informatics
• Bioinformatics

Background:

• Electronic Medical Records (EMRs) offer opportunities for Rapid Learning Systems (RLSs) in clinical decision support.
• Melanoma treatment is complex due to patient variability, highlighting the need for advanced evidence-based tools.
• The Melanoma Rapid Learning Utility (MRLU) was developed to address these challenges.

Purpose of the Study:

• To develop and evaluate the Melanoma Rapid Learning Utility (MRLU) as a component of an RLS for melanoma.
• To enable physicians to gain clinical insights through rapid cohort identification and analysis.
• To leverage practice-based evidence for real-time, data-driven treatment planning.

Main Methods:

• Developed a novel clinical decision support system for melanoma using practice-based evidence.
• Generated patient-centered cohorts for on-the-fly stratified survival analyses.
• Compiled a database from clinical, pharmaceutical, and molecular data of 237 metastatic melanoma patients.

Main Results:

• The MRLU successfully identified known melanoma trends, including BRAF mutation frequencies and survival rates with targeted therapies.
• Physician users found the MRLU highly useful (mean score 4.2/5.0) and usable (4.42/5.0).
• Physicians expressed strong interest in integrating such on-the-fly evidence systems into their practice (mean response 4.54/5.0).

Conclusions:

• The MRLU serves as a valuable RLS analytical engine and user interface for melanoma treatment planning.
• The system's design principles can inform the development of RLSs for other clinical disorders.
• Further research is needed to optimize the MRLU's integration into clinical workflows for decision-making.