University of Kansas School of Medicine

transcript

Clinical and Translational Informatics Capabilities for the University of Kansas Medical Center

Russ Waitman, PhD Associate Professor, Director Medical Informatics

Department of Biostatistics September 29, 2011

Outline

What is Biomedical Informatics? What are the Clinical Translational Science Awards? Informatics Aims: focus on storing and getting Tools for storing information: CRIS and REDCap Tool for viewing/getting information: HERON/i2b2 Oversight Process Information Architecture Observations Milestones

Questions

Background: Charles Friedman The Fundamental Theorem of Biomedical

Informatics: A person working with an information resource is

better than that same person unassisted. NOT!!

Charles P. Friedman: http://www.jamia.org/cgi/reprint/16/2/169.pdf

Background: William Stead The Individual Expert

William Stead: http://courses.mbl.edu/mi/2009/presentations_fall/SteadV1.ppt

Evidence

Patient Record

Synthesis & Decision

Clinician

5 Human Cognitive

Capacity

The demise of expert-based practice is inevitable

2000 2010 1990 2020

Structural Genetics: e.g. SNPs, haplotypes

Functional Genetics: Gene expression

profiles

Proteomics and other effector molecules

Decisions by Clinical Phenotype

William Stead: http://courses.mbl.edu/mi/2009/presentations_fall/SteadV1.ppt

Background: Edward Shortliffe Biomedical Informatics Applications

Basic Research

Applied Research

Biomedical Informatics Methods, Techniques, and Theories

Imaging Informatics

Clinical Informatics Bioinformatics Public Health

Informatics

Molecular and Cellular Processes

Tissues and Organs

Individuals (Patients)

Populations And Society

Edward Shortliffe: http://www.dentalinformatics.com/conference/conference_presentations/shortliffe.ppt

Background: Edward Shortliffe Biomedical Informatics Research Areas

Edward Shortliffe: http://www.dentalinformatics.com/conference/conference presentations/shortliffe.ppt

Biomedical Knowledge

Biomedical Data

Knowledge Base

Inferencing System

Data Base

Data Acquisition

Biomedical Research Planning & Data Analysis

Knowledge Acquisition

Teaching Human Interface

Treatment Planning

Diagnosis Information Retrieval

Model Development

Image Generation

Real-time acquisition Imaging Speech/language/text Specialized input devices

Machine learning Text interpretation Knowledge engineering

“It is the responsibility of those of us involved in today’s biomedical research enterprise to translate the remarkable scientific innovations we are witnessing into health gains for the nation.”

Clinical and Translational Science Awards A NIH Roadmap Initiative

• Administrative bottlenecks • Poor integration of translational resources • Delay in the completion of clinical studies • Difficulties in human subject recruitment • Little investment in methodologic research • Insufficient bi-directional information flow • Increasingly complex resources needed • Inadequate models of human disease • Reduced financial margins • Difficulty recruiting, training, mentoring scientists

Background: NIH Goal to Reduce Barriers to Research

CTSA Objectives: The purpose of this initiative is to assist institutions to forge a

uniquely transformative, novel, and integrative academic home for Clinical and Translational Science that has the consolidated resources to:

1) captivate, advance, and nurture a cadre of well-trained multi-

and inter-disciplinary investigators and research teams; 2) create an incubator for innovative research tools and

information technologies; and 3) synergize multi-disciplinary and inter-disciplinary clinical and

translational research and researchers to catalyze the application of new knowledge and techniques to clinical practice at the front lines of patient care.

NIH CTSAs: Home for Clinical and Translational Science

Trial Design

Advanced Degree-Granting

Programs

Participant & Community Involvement

Regulatory Support

Biostatistics

Clinical Resources

Biomedical Informatics

Clinical Research

Ethics

CTSA HOME

Other Institutions

Industry

Dan Masys: http://courses.mbl.edu/mi/2009/presentations_fall/masys.ppt

Bench Bedside Practice

Building Blocks and Pathways Molecular Libraries Bioinformatics Computational Biology Nanomedicine

Translational Research Initiatives

Integrated Research Networks Clinical Research Informatics NIH Clinical Research Associates Clinical outcomes Harmonization Training

Interdisciplinary Research Innovator Award Public-Private Partnerships (IAMI)

Dan Masys: http://courses.mbl.edu/mi/2009/presentations_fall/masys.ppt

Reengineering Clinical Research

KUMC CTSA Specific Aims 1. Provide a HICTR portal for investigators to access clinical and

translational research resources, track usage and outcomes, and provide informatics consultative services.

2. Create a platform, HERON (Healthcare Enterprise Repository for Ontological Narration), to integrate clinical and biomedical data for translational research.

3. Advance medical innovation by linking biological tissues to clinical phenotype and the pharmacokinetic and pharmacodynamic data generated by research cores in phase I and II clinical trials (addressing T1 translational research).

4. Leverage an active, engaged statewide telemedicine and Health Information Exchange (HIE) effort to enable community based translational research (addressing T2 translational research).

Supporting Aim 1: Clinical Research Information Systems KUMC has purchased Velos eResearch for a Clinical Trial

Management System (CTMS) and an Electronic Data Capture (EDC) System

CTMS functions Define Studies, Assign Patients to Studies Capture Adverse Events, Reports Budgeting, financial planning for studies and invoicing Sample management, regulatory tracking

EDC functions Design and Capture data on electronic Case Report Forms

(CRFs) – ideally in real time. “Patient portal” for surveys and EDC by subjects Export Data for analysis.

CRIS Intro Screen

CRIS: sample e Case Report Form

CRIS: Document Adverse Events

REDCap: Research Electronic Data Capture https://redcap.kumc.edu

It uses the same username and password as your KUMC email. Check out the training materials under videos Case Report Forms and Surveys

For consultation and to move project to production: Register your project with us so we can make sure we don't screw up and drop the ball. http://biostatistics.kumc.edu/projectReg.aspx After you register your project, a CRIS team member, likely Kahlia Ford will

get in touch with you.

Check out other institutions using REDCap and possibly borrow from the master library. http://www.project-redcap.org/

REDCap Disclaimer For clinical trials, CRIS/Velos may be a better fit Multiple years of experience CRIS team builds for you with biostats review Budget for CRIS team and biostats explicity

“Investigator driven” REDCap works if PI takes responsibility for data Scalability: informatics provides consultation and

responsibility for technical integrity; not your dictionary. Underwritten by CTSA right now

Or middle model where informatics can build for you in REDCap. Again, you budget for our team’s time

REDCap Case Report Form Example

REDCap Survey Example

Aim #2: Create a data “fishing” platform

Develop business agreements, policies, data use agreements and oversight.

Implement open source NIH funded (i.e. i2b2) initiatives for accessing data.

Transform data into information using the NLM UMLS Metathesaurus as our vocabulary source.

Link clinical data sources to enhance their research utility.

Develop business agreements, policies, data use agreements and oversight. September 2010 the hospital, clinics and university signed a

master data sharing agreement to create the repository. Executive Committee – decides organization/systems expansion Data Request Oversight Committee – guides implementation and

approves/monitors use. Use Cases:

After signing a system access agreement, cohort identification queries and view-only access is allowed but logged and audited

Requests for de-identified patient data, while not deemed human subjects research, are reviewed.

Identified data requests require approval by the Institutional Review Board prior to data request review.

Contact information from the Frontiers Participant Registry have their study request and contact letters reviewed by the Participant and Clinical Interactions Resources Program

Current Functionality • Single sign-on (CAS) integration with HERON portal

linked off Frontiers home page (Aim 1) • Real-time check for current human subjects training

(LDAP Chalk) • System Access Agreements, Data Use Agreements

and Review Processes implemented in HERON with web pages for monitoring system use

• Demonstration • i2b2 and HERON tools • if time, we’ll do this in real time at the end of the talk

Implement NIH funded (i.e. i2b2) initiatives for accessing data.

i2b2: Count Cohorts

i2b2: Patient Count in Lower Left

i2b2: Ask for Patient Sets

i2b2: Analyze Demographics Plugin

i2b2: Demographics Plugin Result

i2b2: View Timeline

i2b2: Timeline Results

Constructing a Research Repository: Ethical and Regulatory Concerns Who “owns” the data? Doctor, Clinic/Hospital, Insurer,

State, Researcher… perhaps the Patient? Perception/reality is often the organization that paid for the system

owns the data. My opinion: we are custodians of the data, each role has rights and

responsibilities Regulatory Sources:

Health Insurance Portability and Accountability Act (HIPAA) Human Subjects Research

Research depends on Trust which depends on Ethical Behavior and Competence

Goals: Protect Patient Privacy (preserve Anonymity), Growing Topic: Quantifying Re-identification risk.

Re-identification Risk Example Will the released columns in combination with publicly available data re-identify individuals? What if the released columns were combined with other items which “may be known”? Sensitive columns, diagnoses or very unique individuals? New measures to quantify re-identification risk.

Reference: Benitez K, Malin B. Evaluating re-identification risks with respect to the HIPAA privacy rule. J Am Med Inform Assoc. 2010 Mar-Apr;17(2):169-77.

Constructing a Repository: Understanding Source Systems, Example CPOE

Generic Interface

Engine (GIE)

Laboratory System

Pharmacy System

WizOrder Server

WizOrder Client

Mainframe DB2

HL7 Lab DB

Temporary Data queue (TDQ)

Internal Format

Repackages and Routes

Print SubSystem

document

Knowledge Base, Files

SQL Orderables, Orderset DB

Drug DB

Most Clinical Systems focus on transaction processing for workflow automation

Constructing a Repository: Understanding Differing Data Models used by Systems

http://www.cs.pitt.edu/~chang/156/14hier.html

http://www.ibm.com/developerworks/library/x-matters8/index.html Star Schemas: Data Warehouses

Hierarchical databases (MUMPS), still very common in Clinical systems (VA VISTA, Epic, Meditech)

Relational databases (Oracle, Access), dominant in business and clinical systems (Cerner, McKesson)

Murphy SN, Weber G, Mendis M, Gainer V, Chueh HC, Churchill S, Kohane I. Serving the enterprise and beyond with informatics for integrating biology and the bedside (i2b2). J Am Med Inform Assoc. 2010 Mar-Apr;17(2):124-30.

HERON: Repository Architecture

Extracting, Loading, Transforming Data • Goal: stable monthly process, minimal downtime

• Complete rebuild of the repository, not HL7 messaging. • Two databases: create new DB while old DB is in use. • When the new DB is ready, switch over i2b2 to serve it.

• Initial Files from Clinical Organizations • Export KUH Epic Clarity relational database instead of

Cache/MUMPS. • Monthly file from UKP clinic billing system (GE IDX).

• Demographics, services, diagnoses, procedures, and Frontiers research participant flag.

• ELT processes largely SQL (some Oracle PL/SQL) • Wrapped in python scripts.

HERON De-identification Decisions HIPAA Safe Harbor De-identification Remove 18 identifiers and date shifting by 365 days back Resulting in non-human subjects research data but treated

as a limited data set from a system access perspective. System users and data recipients agree to treat as a limited data set (acknowledging re-identification risk)

To be addressed: For now, we won’t add free text such as progress notes with

text scrubbers (DeID, MITRE Identification Scrubber toolkit) Currently have “obfuscation” turned on.

No sets < 10 and sets randomly perturbed + 3 patients While de-identified, access to timeline functionality provides

individualized patient “signatures”

Transform data into information using standard vocabularies and ontologies

Source terminology Completed planned Notes

Demographics: i2b2 April 2010 Using i2b2 hierarchy. Restricted search criteria to geographic regions (> 20,000 persons) instead of individual zipcodes

Diagnoses: ICD9 April 2010 Using i2b2 hierarchy Procedures: CPT June 2010 UMLS extract scripts developed with UTHSC at Houston

Lab terms: LOINC November 2010 Plan to use i2b2 hierarchy Medication ontologies: NDF-RT December 2010 Physiologic effect, mechanism of action, pharmacokinetics, and

related diseases. Nursing Observations July 2010- NDNQI pressure ulcers mapped to SNOMED CT to evaluate

automated extraction of self reported activity. (Drs. Dunton and Warren.)

Pathology: SNOMED CT February 2011 Providing coded pathology results and patient diagnosis is a critical objective for defining cancer study cohorts in Aim 3.

Clinical narrative 2012 As hospital restructures clinical narrative documentation to use EPIC’s SmartData (CUI) concepts, will determine appropriate standard.

National Center for Biological Ontology

2013 In support of Aim 3 focus on bridging clinical and bioinformatics to advance novel methods.

Other Key HERON decision “Lazy” Load supports alternative views of reality Load with the local terminology first. Map concepts to

standards secondarily in the concept space. Allows multiple ontologies for observations and works

around mapping challenges with contributing organizations

Further technical details described at: http://informatics.kumc.edu/work/wiki/HERON

Linking Clinical Data: FY2012 Sources Supporting Cancer Center Initiative HERON Executive Committee approval June 2011 for

incorporating: University Biospecimen Repository (Aim 3, Cancer Center) Hospital Tumor registry (Aim 3, Cancer Center) University REDCap and Velos Registries and Clinical Trials

systems (Aim 3, Cancer Center) Hospital billing ICD9, MS-DRG, Insurance Status Social Security Death Master File (Aim 4, Cancer Center) Cerner CoPath pathology system (Aim 3, Cancer Center)

Also continue to extract and refine data from Epic EMR

Developing a Rich Description of our Population: Existing and Planned Data Sources for HERON. Existing sources shown in bold underlined text and planned in plain text

An i2b2 query against HERON for currently supported cancer centric data sources

Any neoplasm ICD9 diagnosis (106,000 patients) and a WBC count (121,000) -> 44,000 distinct patients, *require height (123,000) and weight (154,000) -> 35,000 patients, •require Wong-Baker pain scale (84,000) ->14468 patients, •Body Temperature (158,000) -> 14463 patients, •Surgical Pathology Procedures CPT (85,000) -> 12446 patients,

Finally selective seratonin 5-HT3 antagonist antiemetics -> 8517 patients With our improved hardware (Fusionio memory cards), the cohort size is returned in 15 seconds for this 8 group query.

CTSA Aim #3: Link biological tissues to clinical phenotype and our research cores’ results

Support Cancer Center, IAMI, and bridge to Lawrence Research

First focus: Incorporate clinical pathology and biological tissue repositories with HERON and CRIS to improve cohort identification, clinical trial accrual, and improved clinical trial characterization Aligned with existing enterprise objectives to improve

biological tissue repository information systems Clinical trial accrual identified by many as a weak point

institutionally Target both biological research specimens and routine

clinical pathology

Biospecimen Shared Resource Integration

KUH Tumor Registry Validated Outcomes and Observations Tumors, Nodes, Metastasis (TNM) on complete cases Untapped investment: 7 cancer registrars (Tim Metcalf) ~65,000 cases, data since 1950s

North American Association of Central Cancer Registries (NAACCR) file format Will build on work at other NCI designated i2b2 users

(Group Health Cooperative in Seattle, Kimmel Cancer Center in Philadelphia have shared their code/metadata with us)

John Keighley providing invaluable expertise Later, supplement with additional treatment

information not in NAACCR file

Adding Social Security Death Master File Have Death status on approximately 90 million

people. Contains Social Security Number, Name, Date of Birth, Date

of Death, Place of Death Monthly update file from ntis; will sync with releases

Released Friday, September 23, 2011 Matching on SSN plus DOB 177,706 of our 1.8 million people noted as deceased

according to Social Security Administration versus 23,850 from hospital systems.

Future Functionality: IRB and i2b2 1.6 Moving beyond counting to line item data review In August, Karen Blackwell Privacy officials agreed to allow

timeline access under current system access agreement Released Friday, September 23, 2011

i2b2 version 1.5: DataMart Request Form to facilitate our Data Use Agreement

i2b2 version 1.6: Visit enabled queries

i2b2 Modifiers with i2b2 version 1.6 Will have to redo ELT to take advantage

Example: Prostate Cancer and PSA tests

Data Mart Request Form

Murphy SN et al, https://www.i2b2.org/events/slides/i2b2_OpeningTalk_20110628_Murphy.pdf

What do Visits and Modifiers Offer? Visits: I want to know the patient had the lab and the medication in

the same episode of care. Conceptually, i2b2 has had a table for the visit dimension

but the software never exploited the data Modifiers: Is it a billing diagnosis or from the problem list? Is it a

primary or secondary? How to I represent all parts of a medication order (dose,

route, frequency)?

Constrain observations to the same visit

i2b2 Modifiers in the User Interface

HERON <-> REDCap Integration i2b2: excels at data warehousing, knowledge

management, hypothesis exploration REDCap: pretty solid tool for storing and collecting

research data and it’s very user friendly. Goal: if we can integrate the best of both, we will

inherit the advancements in each project. Use cases: Breast Cancer Registry in REDCap integrated with HERON

which holds the biospecimens (REDCap -> HERON) Fulfilling Data Request for Participant Contact Information

(HERON -> REDCap)

Breast Cancer Registry

Similar to Tumor Registry: BSR Personnel create forms and enter data to improve annotation for fields that are difficult to automatically extract from Epic and other clinical systems.

Providing and Auditing Use of Participant Contact Information

Questions, HERON, REDCap demo

University of Kansas School of Medicine

Documents