Practical Aspects of Implementing and Applying Health Care Cloud Computing Services and Informatics to Cancer...

Area of Science:

• Oncology
• Bioinformatics
• Health Informatics

Background:

• Cloud computing has increased cancer data volume and complexity.
• Navigating cloud platforms for cancer research involves tradeoffs in performance, cost, and complexity.

Purpose of the Study:

• To describe practical challenges of using cloud services for cancer clinical trial matching.
• To evaluate the utility of cloud-based natural language processing for analyzing clinical trial data.

Main Methods:

• Extracted interventional cancer clinical trials from ClinicalTrials.gov.
• Utilized Google Cloud Healthcare Natural Language API to analyze trial titles and eligibility criteria.
• Developed an informatics pipeline to summarize extracted entities and their distributions.

Main Results:

• Analyzed 38,851 cancer clinical trials, finding significant differences in cancer category distribution compared to ClinicalTrials.gov.
• Identified 949 genes, 1,782 lab tests, 2,086 medications, and 4,902 NCI Thesaurus terms with variable detection accuracy (12.8%-89.9%).
• Processed 167,179 text records using 77,702 API calls, taking 33 processing-hours.

Conclusions:

• General-purpose cloud healthcare tools are currently insufficient for automated clinical trial matching in precision oncology.
• Effective extraction and classification of clinical, genetic, and medication concepts are crucial for these tools.
• Understanding practical cloud computing aspects is essential for cancer researchers.