Implementing Biomedical Informatics Approaches to Facilitate Translational and Clinical Research

Size: px
Start display at page:

Download "Implementing Biomedical Informatics Approaches to Facilitate Translational and Clinical Research"

Transcription

1 Implementing Biomedical Informatics Approaches to Facilitate Translational and Clinical Research Institute of Medicine May 30, 2014 David A. Fenstermacher, Ph.D. Chief Research Information Officer Professor, Dept. of Biostatistics Virginia Commonwealth University

2 Quickly and accurately identifying patients for clinical trials that are based on molecular biomarkers Integrating heterogeneous data from multiple sources to continuously screen patients for clinical trial eligibility and automated reporting Developing processes that prompt clinicians about potential patient eligibility Creating frameworks for evaluating patients at multiple sites for rare biomarker clinical trials

3 Preclinical Assessment In Silico Trial Design Precision Patient ID Post Trial Analytics Use of clinical data, molecular data and tissue to qualify and validate biomarker as a robust subject of clinical research. Evaluation and In Silico design of clinical trial using patient clinical and molecular data as support. Use of patient database and network to rapidly identify patients at trialready sites. Detailed clinical and molecular characterization of non/responders to enrich value of clinical trial results. Pre-clinical validation of biomarker as determinant of response Pre-trial go/no-go decision and enhanced trial design capability Pre-existing, trial-ready populations at established high-profile sites Superior data mining and analytical capabilities

4 357, ,169 16,871 11,592 8,232 2,883 1, ER+ Patients in MCC HRI Female Date Dx Jan 06 Dec 11 Vital Status: Alive Age at Dx ( 50) Primary Site BC BC Histology Stage III and IV Patients with TES

5 Automatically Identify patients (or subjects) Integrate data and provide trial eligibility communication to clinicians prior to patient visits Provide sample size estimates - feasibility Create clinical datasets that may be linked with biospecimen databases or provide annotation for other genomic data

6 System integrates three key areas focusing on the clinical trials recruitment process- automated matching, screening, notification of providers and tracking of eligibility status over time Software system developed to assist in quickly and efficiently screen and identify cohorts /sets of patients meeting specific clinical or nonclinical criteria 3 components of CTED are: Automated Matching Tool MD Alert Tracking system

7 Figure 2 Massey Cancer Center Cancer Research Informatics and Services (CRIS) Core Components, Relationships & Data Infrastructure External Data Feeds for CRIS egate HL7 Dictations radiology, operative notes DC summaries EMR Clinical Notes IP/Professional Billing Pathology Shadow DB IDX Scheduling System Decision Support BMT DB Universal Data Store CRIS Internal Data Store Inpatient discharges Radiology Rpts Surgery Rpts Clinical notes all specialties Surgical Pathology Reports Clinical Pathology Reports IP/OP & Professional Billing IDX Scheduled Visits Primary Data Entry by Clinical Research Staff Analytics Services Datasets Data linkages Analysis Support CRIS Components Automated Cancer Extraction Application (ACE) Links, Screens, Parses and stores data for surveillance and research Clinical Trials Eligibility Database (CTED) Tracking Tool Manual Query Tool Automated Matching Tool for assessing CT eligibility Clinical Trials Data Management System (ONCORE) CT administration and compliance Information System (MCCIS) Claims/linked data Repository VCUHS Cancer Registry Tissue & Data Acquisition & Analysis Core Clinical and Cancer Control Research

8 Provides an automated electronic solution to efficiently identify patients meeting specific inclusion/exclusion criteria (clinical, demographic etc.) Scheduled, automation screens all patients repeatedly to determine eligibility or match status (based on each patient s current clinical status) Screens data for ALL patients across the health care system ( 700,000 patients) Presents that list with linked documentation to the research staff and/or PI Customizable

9 Offers the opportunity to select patients based on combinations of discrete variables as well as easily customizable searching of free text: Demographics Treatments Clinical Lab test results Diagnoses Tumor markers (PSA, CEA, CA19 etc.) Serum chemistries Hematologic parameters

10 Free text documents searched using text strings/terms unique to your study- enables access to information not otherwise (readily) available clinical notes (stage, detailed CA RX, medications) pathology reports (histology, genetic markers stage) radiology dictations (disease progression, mets) discharge summaries Algorithm-derived complex clinical concepts (e.g. Newly diagnosed metastatic disease from radiologic dictations )

11 Links output data from Automatch (patients matching study specific eligibility criteria) with VCU scheduling system each night. If patient on a list from the query is scheduled for next day visit then designated personnel (research nurse, treating physician etc.) receive an alert directing them to a secure website 24hours prior to the scheduled visit Can access the information at any time (even at point of care)

12 Billing Data Clinical Data TCC Database Molecular Data INTEGRATION OF PATIENT S CLINICAL, BILLING AND MOLECULAR DATA PROVIDES: Clinical Trial Matching for Biomarker and non-biomarker Clinical Trials Comparative Effectiveness Research (Outcomes Research) Post Trial Analytics Improved ability to track and understand patient treatment outcomes and costs

13 VCU is currently exploring expanding CTED usage at two other cancer centers VCU s Center for Clinical and Translational Research (CCTR) has signed a MOU with Penn State Hershey to begin creating joint infrastructure for cohort discovery Using I2B2 for data integration CTED algorithms and tools leveraged in collaboration

14 Pre-built data cubes / reports provided by centralized Decision Support group did not have the data elements / granularity required for many of the types of analyses this unit conducted, including assessment of potential number of patients eligible for a given clinical trial. Typically, existing data products were built with broad, current financial and / or operational reports in mind for example, metrics for the entire hospital. And these were entirely unsuitable for creating datasets for researchers to mine for health services research on cancer care, and the enterprise-wide analytics staff did not have cancer-specific knowledge and expertise. Thus, we needed to create a system that could be used by a dedicated team for more granular financial and operational reports and to support clinical and health services research at the cancer center.

15

16 Research inquiry Operational reporting Financial reporting Clinical inquiry and reporting Outreach efforts Maps Business development

17 Enhanced ability to link our data with external / alternate data sources Cleaner data overall far less manipulation required while working within the data provided weekly. Some manipulation is still required when linking with external / alternate data sources, depending on the source. Using MDAS, Analytic Services has fulfilled more than 200 research requests and over 70 investigations within the past three years, as well as filling a similar number of non-research requests: Research Services: Sample size calculations/feasibility for clinical trials Custom datasets: integrated across different sources of patient data (e.g., billing, cancer registry, bone marrow transplant, palliative care, appointment data, clinical data from clinical EMR) identifiable (requires IRB approval) or fully de-identified Analytic findings and interpretations in collaboration with health services researchers Non-Research Services Scheduled and ad-hoc data reports on clinical operations and finances Data-driven maps of cancer incidence, research sites, clinical sites, population characteristics Quality assurance analyses for improving cancer care

18 CTED and MDAS provide data management infrastructures that integrate heterogeneous data, automate reporting, track patients longitudinally and can alert clinicians prior to patient visits Overall, molecular and clinical data generated at the point of care is being used for patient stratification, cohort discovery and multiple research projects Increase knowledge regarding the relationship between therapeutics and molecular biomarkers Expand resources to more chronic and acute diseases beyond cancer

19 Translational Research Informatics Core Charles Geyer, M.D. Lynne Penberthy, M.D. Jonathan DeShazo, Ph.D. J. Brian Cassel, Ph.D. Valentina Petkov Chris Gillam Thomas Neumann Jim McDermott Jon Lowman Dolan Smith Tim Aro Nevena Skoro, MPH R. Michael Sarkissian Karman Tam, MPH Karen Shineer, MSHA CPC