Enabling Discovery in High-Risk Plaque using Semantic Web Approaches
1 like823 views
The document discusses the High-Risk Plaque (HRP) initiative aimed at advancing anti-atherosclerosis treatments through collaboration among key stakeholders. It outlines a series of studies focusing on imaging and circulating biomarkers to predict cardiovascular events and the application of semantic web methodologies for data sharing and analysis. Additionally, it emphasizes the integration of bioimage data with public knowledge resources and aims to facilitate complex queries and interoperability in a pre-competitive environment.
Enabling Discovery in High-Risk Plaque using Semantic Web Approaches
- 1. Enabling Discovery in High-Risk Plaque using Semantic Web Approaches C-SHALS 2009 Cambridge MA February 25-27, 2009
- 2. What is HRP? “ HRP” stands for “High-Risk Plaque” HRP is an authoritative industry-funded precompetitive activity to open novel markets for anti-atherosclerosis treatments. It involves all key stake holders: FDA Payor NHLBI/European Public Sector Academia/thought leaders, and Industry Therapeutics Imaging In-vitro diagnostics
- 3. HRP Initiative Activities Circulating Biomarker Pilot Study (Duke—CATHGEN) BioImage Study Circulating Biomarker Study (Copenhagen City Hospital—CGPS, CCHS) Plaque Biology Study (U.Maastricht—CARIM) Health Economic Study Regulatory Initiative
- 4. BioImage Study Objective: To identify imaging and/or circulating biomarkers that predict 3-year cardiovascular events Determine predictive value of biological and/or imaging markers for near-term (1-3-year) outcomes
- 5. Physical measurements Blood pressure, height, weight, EKG, ABI Blood samples DNA, RNA, plasma, serum Cardiac CT and ultrasound (n=6000) For individuals found to be high risk: CT angiogram of coronary vessels MRI of carotids and abdominal aorta FDG -PET Follow up (~3 years) Enrollment (n=7300) Humana members ages 55-80 (men) or 60-80 (women), without known CV disease or serious medical conditions Ongoing event monitoring and survey every 6 months Primary Endpoint: Major CV events (n= ~600) Associate blood biomarkers with imaging data BioImage is a payor-based observational study BioImage Study Design 1Q08 2Q09
- 6. Enabling BioImage Data Mining (POC) Semantic Web methodologies can be applied to HRP datasets These approaches allow for: Sharing the data across all HRP companies in a fully interoperable manner Annotating the data with existing public biological knowledge Enabling faceted browsing to slice & dice data displays for focused biological questions Adopting this approach in a pre-competitive environment will: Enable easier use of image and molecular data, including associating interpretations and annotations Internally jump-start efforts in this emerging standard Demonstrate the utility to the larger community
- 7. Implementing a Semantic Web Approach Convert data format to RDF Improves interoperability Agnostic to the source Easy to adopt with standards Interlinks different types of data & annotated resources Generate an interface using MIT/CSAIL standard Exhibit Determine appropriate filters and lenses needed ask focused questions Scatter plots, tables, timelines, maps Web site hosted at BGM Underlying data files, interface available to member companies
- 8. Demo from the POC
- 9. Lessons Learned from the POC Study Existing BioImage Data maps completely to Semantic Web format (RDF) Large data sets can be viewed and analyzed via enabled browsers, in any combination Configurable views and statistical lenses Metadata can be attached to Image data Still need to enhance access for CT and MRI images
- 10. BioImage for the Semantic Web (BISM) Goals Design and Implement a full-scale Semantic Web Solution for BioImage Data: integration and views Enable complex queries and inferencing on the complete data-set Platform for complex analysis, dashboarding, and reporting Support annotations of data subsets Enable direct integration with public and private molecular and disease knowledge
- 11. BioImage Data Model BIOIMAGEPARTICIPANT POPUL STRATUM ELIGIBILITYSURVEY INITIALSCREENING CTSCAN MRISCAN SUBSTUDY BLOODCHEMISTRY HDL CALCIUM TRIGLYCERIDES GLUCOSE PHYSICAL BLOODPRESSURE BLOODCOLLECTION LDL
- 14. Phase II Develop Database Wrappers around BioImage to link to Semantic Web structures Lightweight web-service for enabling browser-based data viewing and analysis (perform SPARQL queries) Integration of clinical codes standard via Semantic Web (URI) identifiers
- 15. Additional Possibilities Phase II Standardization of HRPI data with any internal data sets (genes, drugs, symptoms, etc) Interface with additional analytic tools Combining data views with other studies even if only partially compatible Ability to associate with various knowledge sources, e.g., Cardiovascular
Editor's Notes
- #5: W/H Ratio (waist-to-hip ratio) ABI (Ankle Brachial pressure Index): The Ankle Brachial Pressure Index (ABPI) is the ratio of the blood pressure in the lower legs to the blood pressure in the arms. Compared to the arm, lower blood pressure in the leg is a symptom of blocked arteries (peripheral vascular disease). The ABPI is calculated by dividing the systolic blood pressure in the arteries at the ankle and foot by the higher of the two systolic blood pressures in the arms.