Building the Electronic Data Infrastructure: Lessons from Indiana PROSPECT

Select to access the PowerPoint® presentation (4 MB).

Slide 1

Building the Electronic Data Infrastructure: Lessons from Indiana PROSPECT

Paul Dexter, MD
Chief Medical Information Officer, Wishard Health Services
Regenstrief Institute Scientist

Supported by AHRQ Grant R01 HS19818-01

Dr. Dexter has no conflict of interest.

Slide 2

Enhancing health care IT infrastructure

• A learning health system.
• Coordinated clinical, research, and quality improvement efforts.
• Outcomes important to patients.
• CER, PCOR.
• Leveraging EHRs and other data sources.
• Rapid, comprehensive, hypothesis-generating results.

Slide 3

Local opportunities and challenges

• A large health information exchange.
• Creation of new software.
• Study enrollment challenges, PBRN.
• Investigator access to preliminary data.
• Integration of clinical and genetic research.
• Capture of patient reported outcomes.
• Improved support of standards.

Slide 4

Specific aims

• Enhance existing information technology infrastructure:
  • Support providers, caregivers, and researchers by providing new tools for communication and co-management.
  • Provide de-identified access to the INPC database for CER work.
  • Capture and store health care outcomes important to patients and their caregivers.

Comparative effectiveness clinical trial of medication treatment for behavioral symptoms of Alzheimer's disease.

Slide 5

INPC Data

~ 80 hospitals signed up.

~46 hospitals "live."

• ~ 1,400 interfaces.
• ~ 12 million individuals.
• ~ 4 billion structured results.

• Also includes:
  • Laboratories.
  • Radiology centers.
  • Public health.
  • 5 large payors.

Image: A map of Indiana shows the locations of INPC data contributors.

Slide 6

Image: A screenshot shows a sample patient record entry with observation data sorted by chronological date for the last two years.

Slide 7

Image: A screenshot shows a sample patient record entry showing information on the ordered dosage for Simvastatin for the patient as well as a list of the patient's medical conditions and other medications.

Slide 8

CPOE Application

Image: A chart depicts the process for the CPOE Application. The first level begins with the following list:

• " Application Domain submits Event Rule Requests and Invokes Named Rule Requests.
• " Alert Hub manages Alert Presentation.
• " User Actions handled by Application Domain.

To the right is a yellow box containing the following text:

• " Diagnosis Added.
• " Orderable Concept Selected.

A line connects this box to a white box below it, containing the following text: "Sync/Async Event Nothing Retrieved." A double-headed arrow points between two boxes, one yellow and labeled "Note Template" and the other red and labeled "Named Rule Request." On the far right of the first level is a blue box captioned "Alert Hub" containing icons representing alerts and the following notes: "*Aggregates Alerts. *Determines where and how to display alerts in the UI."

The second level begins with the following list:

• " Receives:
  • o Event Rule and Named Rule Requests.
• " Delegates Rule Response Handling to Registered Subscribers from the Application Domain.

To the right is a large, green box captioned "CDS Hub." Arrows point down from "Sync/Async Event Nothing Retrieved" and "Named Rule Request" on the first level to "CDS Hub." An arrow also points up from "CDS Hub" to "Alert Hub" on the first level. To the right of "CDS Hub" is a large, yellow box captioned "Rule Response Subscribers" containing icons representing alerts and three ovals labeled "CPOE," "View Chart," and "Notifications." An arrow points from the ovals in "Rule Response Subscribers" to "CDS Hub" and, through it, to "Alert Hub" on the first level.

The third level begins with the following list:

• " Determines which rule to call.
• " Calls different rule engines and receives/passes back rule responses to CDS Hub.
• " Sets a "base" rule priority.

To the right is a large, red box captioned "Rule Hub." Within it are a large, green box captioned "Rule Service (Invokes Rule Runners)" and, below it, a greenish-yellow box captioned "Rule Runners" containing a series of smaller white boxes captioned "DDI 'service'," "DAI 'service'," "PDQ 'service'," "DROOLS 'service'," and "VAX RULES 'service'." Arrows point down from "Rule Service" to each of these Rule Runners. Two arrows point down from "CDS Hub" on the second level to "Rule Service." A third arrow points up from "Rule Service" to "CDS Hub" and, through it, to the ovals in "Rule Response Subscribers" on the second level and "Alert Hub" and "Named Rule Request" on the first level.

Slide 9

Automated study recruitment

Image: A screenshot shows a sample order for Rivastigmine, a drug used in the treatment for Alzheimer's Disease, from Englewood Pharmacy. On the right-hand side of the screen is an alert announcing that a comparative effectiveness trial is being conducted on treatments for Alzheimer's Disease. Would it be okay to ask the patient taking Rivastigmine to participate in the study?

Slide 10

Informatics decision support research

Image: A screenshot shows an order entry for a sample patient. Alerts related to Diabetes Drug Adherence appear on the right-hand side of the screen.

Slide 11

Study design tool

Image: A screenshot shows the components of the study design tool as a decision chart.

Slide 12

Study design tool

Image: A screenshot shows the study design tool as programming written for software.

Slide 13

Integration and enhancements to eMR-ABC

Image: A screenshot shows the eMR-ABC Home page, which displays newly opened patient records, clinic details, future schedules, and patient follow-up schedules. The patient follow-up schedules section is highlighted in red.

Slide 14

Automated biospecimen tracking

Image: Two photographs of women in lab coats are captioned "Phlebotomist" and "Lab Technician" respectively. Two arrows point from the Phlebotomist to two ovals captioned "Scan patient barcode" and "Scan blood tube barcode." Four arrows point from the Lab Technician to four ovals captioned "Scan centrifuge barcode," "Scan blood tube barcode," "Scan box barcode" and "Scan aliquot barcode." Arrows point from all six ovals to a box captioned "caTrack," which contains a photograph of a PDA with scanner and the text "caTrack Business Logic (Web Service)." A double-headed arrow points to/from "caTrack" and a box on the right, containing the text "caTissue Application." A double-headed arrow points to/from "caTissue Application" and another box below it, containing the text "BioSpecimen Database."

Slide 15

De-identified I2b2 queries

Image: A screenshot shows the I2b2 Query and Analysis Tool; a sample search has been entered.

Slide 16

Image: Three yellow ovals are captioned "Biospecimen Tissue Tracking," "INPC" and "myTrack Molecular Database." The words "Global ID" appear between the first two ovals, and again between the second and third. Arrows point down from all three ovals to a rectangle captioned "Staging Server"; an arrow captioned "De-identification" points down from "Staging Server" to a rectangle captioned "I2B2 Database."

Slide 17

"Vending machine" concepts

• Diabetes mellitus.
• Coronary artery disease.
• Myocardial infarction.
• Heart failure.
• COPD.
• Asthma.
• Hypertension.
• Breast cancer.
• Prostate cancer.
• Lung cancer.
• Colorectal cancer.
• Ovarian cancer.
• Esophageal cancer.
• Stroke.
• Chronic kidney disease.
• GI bleeding.
• HIV/AIDS.
• Schizophrenia.
• Hyperlipidemia.
• Osteoarthritis.
• Rheumatoid arthritis.
• Falls.
• ADHD.
• Etc...

Slide 18

UIMA/cTAKES open source NLP

Image: A chart shows the process of document collection via collection readers and analysis engines to arrived at the common analysis structure consumer. Beneath the chart is a process list of" cTAKES components": Sent. Detector → Tokenizer → Normalizer → POS Tagger → Chunker → NE Recognizer → Negation/Status Marker.

Slide 19

Real-time NLP

Image: A screenshot shows the release order notes for a sample patient with alopecia and occasional diarrhea. An alert on the right-hand side of the screen shows information about possible adverse events that may be caused by one or more medications the patient is taking; alopecia and diarrhea are listed as possible adverse events.

Slide 20

Image: A screenshot shows a release order observations page, which asks questions related to the PHQ-9 quick depression assessment panel.

Slide 21

Integration of LOINC survey instruments

Assessment name:

Brief Interview for Mental Status (BIMS)
Continuity Assessment Record and Evaluation (CARE)
Clinical Care Classification (CCC)
Confused Assessment Method (CAM)
Geriatric Depression Scale (GDS)
Geriatric Depression Scale (GDS)—short version
HIV Signs and Symptoms (SSC) Checklist
howRU
Living with HIV (LIV-HIV)
Mental Residual Functional Capacity (RFC) Assessment Form
Minimum Data Set (MDS) version 2
Minimum Data Set (MDS) version 3
 

Slide 22

Image: A CHICA Pre-Screening Form filled out for a sample patient is shown.

Slide 23

Image: A screenshot shows information for a sample patient.

Slide 24

Image: A screenshot shows a list of Production Feeds under the "Data Feeds" tab.

Slide 25

Image: A lavender cloud is caption "Hadoop Cluster: Subset of RMRS Data." On the upper right are a green cloud captioned "Other Structured and Unstructured Data Sources" and a large data-bin icon captioned "RMRS Oracle Database"; arrows point from both of these to the Hadoop Cluster cloud. Two double-headed arrows point to/from the Hadoop Cluster cloud and a grey box captioned "Batch Systems Job" and a photograph of a woman in a white lab coat captioned "Researcher." In the lower right is a box captioned "Map/Reduce Tools," containing the following text within three smaller boxes: "Mahout: Scalable Machine, Learning Library," "Pig: Scalable Data, Mining Tool," and "Hive: Scalable Data Warehouse Infrastructure, Provides Summarization and Ad Hoc Queries."

Slide 26

Striving for sustainability

Image: A large red circle at the center of the image is captioned "CTSI Informatics and Data Analysis Center." Around it are a number of smaller circles in a circle, captioned "Public Health Informatics," "Health Geographics," "Health System Redesign," "Research Planning and Infrastructure," "Bioinformatics," "Security and Privacy Policy," "High Performance Computing," "Visual Analytics," "Clinical Decision Support and QI," "Data Mining," "Clinical Epidemiology," and "Natural Language Processing."

Slide 27

Thank you