next-generation phenotyping using umls and meaningful use ontologies: snomed ct, rxnorm, and loinc

43
Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC Tomasz Adamusiak MD PhD 7omasz

Upload: tomasz-adamusiak

Post on 17-Dec-2014

1.220 views

Category:

Health & Medicine


4 download

DESCRIPTION

SNOMED CT, LOINC, and RxNorm, fuelled by the Meaningful Use legislation, are poised to become the cornerstone of U.S. health information interchange. SNOMED CT is one of the most comprehensive, multilingual medical terminologies in the world. LOINC is a universal standard for identifying laboratory observations. RxNorm is a standardized nomenclature for generic and branded drugs. All three are integrated within the Unified Medical Language System (UMLS) maintained by the U.S. National Library of Medicine. While physicians rarely have to deal with clinical terminologies directly, these are indispensable for data querying, validation and reconciliation. The Clinical Informatics team at the Medical College of Wisconsin has developed ClinMiner (https://clinminer.hmgc.mcw.edu), a clinical research portal for clinical and diagnostic information on patients in genetics clinics and clinical sequencing programs, as well as other clinical research projects. ClinMiner is a larger system that incorporates data entry forms, patient reports, advanced querying, export and data visualization. Data for the system consists of many clinical and referral documents the patients have accumulated throughout their clinic and diagnostic histories, and are standardized through the three Meaningful Use ontologies: SNOMED CT, RxNorm and LOINC; integrated into a single UMLS perspective that allows for seamless and dynamic translation between the annotating sources, as well as provides a consolidated view of the underlying patient data. This approach is unique in integrating all three terminologies into a single workflow of a clinical application, and in fact is not limited to Meaningful Use, as any terminology integrated within the UMLS can be used to annotate, visualize, and query data. This is of particular significance for reintegrating legacy clinical information, for example, billing data annotated with ICD-9 codes in the process of transitioning to ICD-10. Most importantly, as large resources such as SNOMED CT and the UMLS often remain underused due to their sheer size and complexity, ClinMiner demonstrates that the additional effort is well worth it.

TRANSCRIPT

Page 1: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Next-generation phenotyping using UMLS and Meaningful Use ontologies:

SNOMED CT, RxNorm, and LOINC

Tomasz Adamusiak MD PhD 7omasz

Page 2: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

$13 520 965 732.99 Paid by CMS in EHR incentive payments to EPs 2011 – 2013

Page 3: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Meanwhile across the pond…

Page 4: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

It pays to get started early (259,000 providers so far)

http://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/Downloads/Beginners_Guide.pdf

Page 5: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

2015 will be the defining year in the CMS EHR Incentive Programs

https://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/downloads/EHRIncentProgtimeline508V1.pdf

Pen

alti

es!

Page 6: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Achieving Meaningful Use is “relatively” straightforward

• 15 CMs

• 10 MMs

• 3 Core + 3 Alternate CQMs

• 38 Additional CQMs

Stage 1

Page 7: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Each stage will have its own set of requirements

Final rule, September 4, 2012

Page 8: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Information exchange is at the heart of Meaningful Use

Core Measures:

12. Provide patients with an electronic copy of their health information, upon request

13. Provide clinical summaries for patients for each office visit

14. Capability to exchange key clinical information

Menu Measure:

8. The EP (…) should provide summary care record for each transition of care or referral

https://www.cms.gov/Regulations-and-Guidance/Legislation/EHRIncentivePrograms/Downloads/EHR_Medicaid_Guide_Remediated_2012.pdf

Stage 2

Page 9: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

“unstructured document” is explicitly prohibited in transition of care

Electronic Access

Clinical Summaries

Information exchange in MU Stage 2

Patients Referring provider

Receiving provider

Receiving provider

Page 10: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Structured Summary of Care

Problem list

• ICD–9–CM

•SNOMED CT 2009

•SNOMED CT 2012

Medications

•Any source vocabulary that is included in RxNorm

•RxNorm

Encounter diagnoses

• ICD-10-CM

Laboratory tests

•LOINC 2.24

•LOINC 2.27

Procedures

• ICD-9-CM

•HCPCS + CPT-4

•CDT

• ICD–10–PCS

Page 11: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

The era of non-MU ontologies is over

Expertise in working with medical and pharmacy coding schemes (ICD9/ICD10, HCPCS, CPT4, hospital revenue codes, LOINC, SNOMED and NDC)

Knowledge of clinical terminology and coding standards such as ICD-9, ICD-10, CPT, LOINC, and SNOMED.

This individual will be well-versed in Meaningful Use and standard vocabularies (e.g. RxNorm, SNOMED, etc.)

Experience with healthcare modeling efforts and terminologies such as HL7 v3, SNOMED, LOINC, FDB, CPT

Page 12: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

If only there was something to pull all these terminologies together!

Page 13: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

UMLS – an idea ahead of its time

Donald A.B. Lindberg, M.D.

C. Tilley and J. Willis, The Unified Medical Language System, What is it and how to use it?

Page 14: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Exanthema C0015230

UMLS mappings

rash NOS ICD-10:R21

Cutaneous eruption SCT:112625008

Eruption SCT:1806006

Page 15: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Exanthema C0015230

UMLS mappings

rash NOS ICD-10:R21

Cutaneous eruption SCT:112625008

Eruption SCT:1806006

Page 16: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

The sheer scale makes manual integration impractical

Gene tests

HLA tests

Evaluation and management

Skin tests

Patient information

HPA tests

Everything else

Here Be Dragons

Page 17: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Ontology-based database, query and reporting system

MU Terminologies

• SNOMED CT

• RxNorm

• LOINC

• +

Data types

• Patient and family information

• Demographic information

• Laboratory and genetic test results

• Clinical measurements

• Phenotypes and diseases

• Imaging phenotypes

• Procedures

Page 18: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Data

Warehouse

EHR

Reports

Clinical

Documents

Multiple Data

Sources

Data Integration in

ClinMiner

Standardized with

Meaningful Use

Ontologies

Study -> ETL -> Report

Page 19: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

CCD/ Observ-OM

based

Page 20: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Phenotypes

Medications Labs

LOINC RxNorm

SNOMED CT

UMLS

UMLS Integration

2 900 000 concepts

First

Databank

Micromedex

MediSpan

Gold

Standard

Multum

NDF-RT

Anti-infective

agent

Clinical

finding

SNOMED CT

CONCEPT

Pharmaceutical

product

Disease

SNOMED CT

395 000 concepts

RxNorm

242 000 concepts

Clinical Class

LOINC

PARTS

LOINC Root

LOINC

CLASSTYPES

Laboratory

Class

Radiology Microbiology

LOINC

180 000 concepts

Data semantics in a separate UMLS-driven layer

Page 21: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

MU ready, but also able to reintegrate any data input via UMLS

UMLS

SNOMED CT

RxNorm

LOINC

SNOMED CT

RxNorm

LOINC

ICD-9

MeSH

OMIM

CPT

Page 22: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

B C

A

E D

F

G

H

Page 23: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Oracle Text based google-like search: cystic fibrosis gene carrier

Page 24: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Fuzzy/wildcard matching too!

cron disease myleoid leukemia

Technical note: Oracle Text-based query progression/relaxation with AND/OR/ACCUM/FUZZY operators and inbuilt TF-IDF ranking

Page 25: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Medication reconciliation via National Drug Code (NDC) and RxNorm

31722-331-01 100 tablets of Warfarin Sodium 4 MG Camber Pharmaceuticals

Page 26: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Data-driven customized ontology perspectives

2:0 5:4 3:2

60-80% reduction in graph sizes

Page 27: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Clinical

Avatars ClinMiner entity

MU source

mapping

UMLS

mapping Term label

GENDER F Phenotype None C0015780 Female

GENDER M Phenotype None C0024554 Male gender

RACE African American Asian

Native American

Other Pacific Islander

Unknown

White

Phenotype OMB standard C0085756

C1515945

C0078988

C0043157

C0086409

C1513907

C1532697

African American

American Indian or Alaska Native

Asians

Caucasians

Hispanic or Latino

Native Hawaiian or Other Pacific Islander

Unknown racial group

HEIGHT ClnicalResult LNC:3137-7 C0365282 Body height Measured

WEIGHT ClnicalResult LNC:3141-9 C0365286 Body weight Measured

BSA ClnicalResult LNC:3139-3 C0365285 body surface area measured

INR ClnicalResult LNC:34714-6 C1369580 INR in Blood by Coagulation assay value

SMOKER Y Phenotype SCT:77176002 C0337664 Smoker

SMOKER N NormalPhenotype SCT:8392000 C0337672 Non-smoker

DVT Y Phenotype SCT:128053003 C0149871 Deep venous thrombosis

DVT N NormalPhenotype SCT:413076004 C1446197 No past history of venous thrombosis

AMI Y Phenotype SCT:57054005 C0155626 Acute myocardial infarction

AMI N NormalPhenotype SCT:301121007 C0577811 Myocardial perfusion normal

CYP2C9 GeneticResult LNC:46724-1 C1830800 cyp2c9 gene mutations found [identifier] in blood or

tissue by molecular genetics method nominal

CYP2C92 GeneticResult LNC:56164-7 C2734139 cyp2c9 gene allele 2 [identifier] in blood by

molecular genetics method nominal

CYP2C93 GeneticResult LNC:56165-4 C2734141 cyp2c9 gene allele 3 [identifier] in blood by

molecular genetics method nominal

VKORC1 GeneticResult LNC:50722-8 C1978717 vkorc1 gene mutations found [identifier] in blood or

tissue by molecular genetics method nominal

VKORC1A GeneticResult LNC:50722-8 C1978717 vkorc1 gene mutations found [identifier] in blood or

tissue by molecular genetics method nominal

VKORC1G GeneticResult LNC:50722-8 C1978717 vkorc1 gene mutations found [identifier] in blood or

tissue by molecular genetics method nominal

WARFARIN Medication RxNorm:11289 C0043031 Warfarin

Demo and evaluation:

100 000

clinical avatars x

90 days x

genotype-guided

warfarin dosing

Page 28: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Link to Query Builder

Page 29: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Query Builder

Page 30: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Add term to query

Page 31: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Add term to query

Page 32: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Add term to query

Page 33: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC
Page 34: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Smart reporting

Page 35: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC
Page 36: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC
Page 37: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC
Page 38: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Beyond text search and billing codes

Page 39: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Coming soon

Cross participant query results

• Value limits, negation • Visualization and cohort summary • Ability to add fields to query results – demographics, other measurements or

parameters • Export to Excel

Single participant results

• Short summaries

• Phenotypes • Medications • Labs • Genetic tests

Page 40: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

State of the art ontology development vs. Yahoo! in 1996

Manually-curated ontology

Request new terms

Page 41: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC
Page 42: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Public demo

https://clinminer.hmgc.mcw.edu Use the link or

google clinminer

Click on login User: demouser Pass: demouser

Page 43: Next-generation phenotyping using UMLS and Meaningful Use ontologies: SNOMED CT, RxNorm, and LOINC

Acknowledgments

Marek Tutaj

Stacy Zacher

Clinical Avatars

Vincent A. Fusaro PhD

Peter J. Tonellato PhD

Laboratory for Personalized Medicine Harvard Medical School