Knowledge Graph-Enabled Cancer Data Analytics

Abstract

Cancer registries collect unstructured and structured cancer data for surveillance purposes which provide important insights regarding cancer characteristics, treatments, and outcomes. Cancer registry data typically (1) categorize each reportable cancer case or tumor at the time of diagnosis, (2) contain demographic information about the patient such as age, gender, and location at time of diagnosis, (3) include planned and completed primary treatment information, and (4) may contain survival outcomes. As structured data is being extracted from various unstructured sources, such as pathology reports, radiology reports, medical records, and stored for reporting and other needs, the associated information representing a reportable cancer is constantly expanding and evolving. While some popular analytic approaches including SEER*Stat and SAS exist, we provide a knowledge graph approach to organizing cancer registry data. Our approach offers unique advantages for timely data analysis and presentation and visualization of valuable information. This knowledge graph approach semantically enriches the data, and easily enables linking with third-party data which can help explain variation in cancer incidence patterns, disparities, and outcomes. We developed a prototype knowledge graph based on the Louisiana Tumor Registry dataset. We present the advantages of the knowledge graph approach by examining: i) scenario-specific queries, ii) links with openly available external datasets, iii) schema evolution for iterative analysis, and iv) data visualization. Our results demonstrate that this graph based solution can perform complex queries, improve query run-time performance by up to 76%, and more easily conduct iterative analyses to enhance researchers' understanding of cancer registry data.

Fig. 1.

In this figure, we present a high-level overview of the cancer scientific digital library framework.

Fig. 2.

Hierarchical breast cancer treatment sequence scenarios.

Fig. 3.

In this figure, we present, age-specific TNBC incidence rates (Female, Louisiana 2010–2012). The left figure shows our results and the right figure presents Hossain et al. results [5] (right figure’s credits go to Hossain et al. [5]). According to the 2010’s census in Louisiana, the total number of EA women is 1,463,058 and the total number of AA women is 766,756 (the ratio is almost 2:1), which includes the age group of <30 years. “The estimates are based on the 2010 Census and reflect changes to the April 1, 2010 population due to the Count Question Resolution program and geographic program revisions” [49].

Fig. 4.

In this figure, we present TNBC CDI distribution (Female, Louisiana 2010–2012). The left figure shows our results and the right figure presents Hossain et al. results [5] (right figure’s credits go to Hossain et al. [5]).

Fig. 5.

Age-specific TNBC incidence rates (Female, Kentucky 2010–2012). According to the 2010’s census in Kentucky, the total number of EA women is 1,963,670 and the total number of AA women is 173,032 (the ratio is almost 12:1), which includes the age group of <30 years. “The estimates are based on the 2010 Census and reflect changes to the April 1, 2010 population due to the Count Question Resolution program and geographic program revisions” [49].

Fig. 6.

TNBC CDI Distribution (Female, Kentucky 2010–2012).

Fig. 7.

This figure illustrates an example of the knowledge graph’s easy schema evolution. Here we show a patient with prostate cancer (code: C61.9) and a laterality value of 2 (value for a paired organ). To identify invalid laterality codes, we add an “isValidLaterality” flag dynamically to the graph without substantially changing the software code.

Fig. 8.

In this figure we present, a partial high level visualization of LTR knowledge graph (Rural-Urban Continuum Codes portion).

Fig. 9.

Visualization of the four hypothetical patients information. Although the patient information was simulated to avoid revealing protected health information (PHI) data, the underlying schema is consistent with the actual CTC data.