Monday, June 22, 2020
9:00 pm – 4:00 pm
Fee $275

Fee includes all course materials, breakfast, and afternoon beverage break.

Cancer registries routinely collect data on every identified cancer case, including patient demographics, primary tumor site, tumor morphology and stage at diagnosis, first course of treatment, and follow-up for vital status (survival). While these programs are extremely impressive in their breadth of coverage and rivaled by few other national or international entities, they are necessarily hampered by a lack of depth of coverage. The reasons for this are multifactorial but are essentially explained by a mismatch between resources available for capturing detailed data and the sheer complexity of patient cases. Electronic medical records (EMRs) have worsened this mismatch considerably, as they are many orders of magnitude larger and much more complex than the paper-based records that they replaced. Despite this, Certified Tumor Registrar (CTR) workforce expansion has not kept pace. In 2011, the National Cancer Registrars Association published a workload and staffing study that found that more than one-third of hospital-based registries had one or fewer full-time equivalents on their staff. Due to this circumstance, registry programs have been thoughtful in their selection of necessary data elements so as not to overwhelm the existing workforce capacity. Nevertheless, there is a strong desire by both the cancer surveillance community and the many secondary users of this data to enhance the depth of coverage by taking advantage of new and emerging health information technology solutions. This need is becoming especially acute given the rapidity with which new important diagnostic, prognostic, and predictive biomarkers are emerging in this era of personalized oncology. In 2019, we held a half-day tutorial at NAACCR, highlighting and illustrating several tools that can be used right now to enhance the quality of registry data and to automate certain tasks. This tutorial was well-attended and the feedback received was very positive. Due to the time constraints the tutorial was primarily didactic. For 2020, we propose to cover the same topics and expand the tutorial to a full day session. Specifically, we propose to introduce and demonstrate: 1) the National Cancer Institute (NCI)’s Observational Research in Oncology Toolbox; 2) the DeepPhe natural language processing (NLP) and visualization software; 3) the NCI-DOE application programming interface (API); and 4) the HemOnc chemotherapy regimen ontology. These are described below in brief detail.

Course Objectives

Upon completion of this tutorial, attendees will be able to: 

• Understand the tools available in the NCI’s Observational Research in Oncology Toolbox and consider how they might be incorporated into existing workflows.
• Appreciate how natural language processing tools such as DeepPhe can aid in the abstraction process.
• Appreciate the utility of APIs and how they may ease the data collection process.
• Develop an understanding of how formal ontologic modeling of chemotherapy regimen concepts can increase the utility of cancer registry information.

Requirements

No particular requirements, but some knowledge of software tools will be helpful.

Content

This tutorial will provide several perspectives on the state-of-the-art and challenges of extracting and visualizing cancer patient data. Presentations will cover technical details but are intended to be broadly inclusive and practical. The emphasis will be on the presentation of existing tools and software that could be utilized by central cancer registries and cancer researchers almost immediately. We propose a 6-hour full-day workshop, with a portion devoted to didactics and a portion devoted to hands-on exploration of the tools described.

Course Organizers

• Eric B. Durbin, Kentucky Cancer Registry

• Harry Hochheiser, Associate Professor

• Valentina Petkov, National Cancer Institute

• Donna R Rivera, National Cancer Institute

• Guergana Savova, Boston Children’s Hospital and Harvard Medical School

• Jeremy Warner, Associate Professor of Medicine and Biomedical Informatics