Big Data in radiation therapy: challenges and opportunities.

Area of Science:

• Oncology
• Medical Informatics
• Data Science

Background:

• Radiation oncology data exhibits Big Data characteristics (Volume, Variety, Velocity, Veracity) due to disparate sources and privacy concerns.
• Challenges in data utilization include poor quality, privacy, interoperability, and large volumes, hindering comprehensive model development.
• Existing data collection methods like clinical trials and registries have limitations in addressing these Big Data challenges.

Purpose of the Study:

• To explore strategies for effectively utilizing radiation oncology Big Data for research and innovation.
• To advocate for the adoption of Findable, Accessible, Interoperable, Reusable (FAIR) data principles in oncology data management.
• To propose distributed learning as a solution for overcoming data centralization challenges.

Main Methods:

• Classification of radiation oncology data using the 4Vs of Big Data.
• Discussion of current data collection and utilization approaches: clinical trials, registries, and data stores.
• Proposal of distributed learning techniques for analyzing FAIR data.

Main Results:

• Radiation oncology data is characterized by significant Big Data attributes, posing substantial challenges.
• The FAIR data principles are crucial for enabling effective data sharing and analysis.
• Distributed learning offers a viable method to learn from dispersed FAIR data without centralization.

Conclusions:

• Effective utilization of radiation oncology Big Data requires addressing challenges related to quality, privacy, and interoperability.
• Implementing FAIR data principles is essential for maximizing the value of collected data.
• Distributed learning platforms are vital for advancing personalized medicine through rapid learning from diverse data sources.