workflow modeling for medicine an experience with bpmn2 at intermountain
TRANSCRIPT
Workflow Modeling for Medicine
An experience with BPMN2 at Intermountain
BPMN 2.0
• Business Process Model and Notation 2.0– OMG Standard– Roots are in Graphical Modeling Environments for
Business Processes– Version 2.0 => Computability
• Requires Services– Used for Service Orchestration– Allows Construction of Applications by Integrating
Services• Provides Standard Workflow Components• Provides Integration for Custom Services
Designing a CDSS for Complex Electronic Protocols
• Business Process Model and Notation (BPMN) Engine• Standard UI Generator (WS-Human Task)• Appropriate Services from HER (HSP)• Special Services
– To Manage Data Models– To Provide Added Inferencing Capability
• Physiologic Models• Predictive Models• NLP Services• Custom Java-Based Tools
• UI Client-Tightly Integrated with Hosting EHR
Protocol Execution EnvironmentBPMN2 Engine
Custom BPMN Service Tasks
Data Services
Data Transformation Services Etc…
Human Task Client Framework
Human Tasks
HumanTask Processor
MetaData
User Interface
Generator
Hosting EH
R
Etc. Order
Comm
unications Services
Patient Identification
Services
Data W
rite Services
Data Read Services
Publish/Subscribe Services
(Data D
rive)
Predictive Modeling Services
EHR-Based Resources
BPMN-Frameworks Consume Services(provided by EHR-based environment to workflow engine-HSP)
• Security• Patient Lookup/Retrieval• Clinical Data Access• Clinical Data Storage• Order Query• Order Communications• User Communications
– Client Integration– Alert Delivery
• Etc.
Framework has Internal Services(services used in managing workflow engine activity)
• Start Protocol– Registers patient into protocol, records protocol status.
• Save Protocol– Serializes protocol to storage.
• Get Protocol State– Recovers protocols and protocol states for a given patient
• Alert Service– Build and store Alert Event/attach to relevant user task
• Get active user tasks– Finds protocol associated active user tasks
• Complete user tasks– Signals completion of user tasks/updates protocol data
• Etc.
Two Examples
• Pulmonary Embolism Workup– Built using BPMN 2.0 authoring– Delivered using BPMN 2.0 runtime– Used Activiti (open-source version)
• Pneumonia Protocol– Originally build in Java/Drools– Conversion to BPMN 2.0– Uses Complex Event Processing Framework
Example: A Protocol for the workup of Pulmonary Embolism(an example from the Intermountain Intensive Medicine Program)
• Goal:– Identify Patients undergoing workup for PE– Evaluated Likelihood of PE– Apply a Pulmonary Embolism Workup Protocol
• Approach:– Watch for Initial PE Workup– Suggest Enrollment in PE Protocol– Protocol Provides Diagnostic Suggestions
• Risk-Based Determination of Best Workup Strategy• Feedback of Relevant Data, Risk Scores, Workup Suggestions
Workflow Automation for
Pulmonary Embolism
• Diagnose PE in the ED• Trigger from PE-oriented
procedures– D-Dimer, CT-PA, Vent/Perf
Scan, etc.
• Offer Finding Interpretation and Diagnostic Assistance
Pulmonary Embolism Workflow
Pulmonary Embolism Workflow
Explode Activity
Go to Missing Vitals Screen Go to RGS
Screen
RGS Detail
Example: Pneumonia Care Protocol(an example from the Intermountain Intensive Medicine Program)
• Goal:– Rapidly Screen for Pneumonia Patients in the ED
– Assess Risk of Death– Apply a Pneumonia Care Protocol
• Approach:– Use Probabilistic System to Screen Patients– Suggest Enrollment in Pneumonia Protocol– Protocol Provides Therapeutic Suggestions
• Explicit Admission Suggestions• Explicit Treatment Suggestions
Identify pneumonia patients being seen in the Emergency Department. • Abnormal vital signs • Compatible chief complaint • Positive CXR (determined by natural language processing of dictated report)
Electronically alert ED physician, asking to confirm diagnosis of pneumonia
Search database for HCAP risk factors. • Hospitalization within 90 days • Nursing home residence • Dialysis or wound care center • Prior positive culture for MRSA
Ask ED physician to confirm or refute presence of HCAP risk factors.
Calculate eCURB, PaO2/FiO2, severe CAP criteria• 30 day predicted mortality Additional indications for admission • Hypoxemia, presence of pleural effusion • Co-morbid illness • Inability to take oral medications • Poor social support/homeless • Failure of outpatient therapyRecommendation for Home vs Ward vs ICU admission.
Order cultures/urine antigen studies per guidelines
• ICU all, ward per pre-selected check box, home none • Auto-generated orders
Antibiotics per site of care and HCAP/ MRSA/Pseudomonas risk factors. Prescription generated, first dose administered
Pneumonia Screening Tool
Data Supporting Pneumonia Assessment Clinical Data
Repository
Pneumonia Protocol
Enrollment
Pneumonia Treatment Protocol
Computable Medical Knowledge Reposotory
Chest Xray Reports
Chest Xray Report Processing
(Structured Data Extraction)
Workflow for PneumoniaExample: Community-Acquired Pneumonia
Does the patient have pneumonia?
Should we used the protocol?
Apply Pneumonia Care Protocol.
Pneumonia Detector(a Bayesian Network using 40 clinical factors)
BPDiastolic< 69.569.5 to 82.5>= 82.5
50.634.714.7
71.3 ± 10
Chloride< 103.5103.5 to 105.5>= 105.5
49.423.527.2
104.1 ± 1.8
WBC< 11.8511.85 to 18.75>= 18.75
83.614.22.12
9.68 ± 3.6
Age< 15.515.5 to 45.5>= 45.5
100 0 0
0.5 ± 8.7
RespRate< 19.519.5 to 21.521.5 to 27.5>= 27.5
31.922.525.919.7
22.9 ± 4.6
TempC< 36.7536.75 to 37.4537.45 to 38.05>= 38.05
41.424.411.522.8
37.17 ± 0.8
MeanBP< 85.585.5 to 99.5>= 99.5
48.836.914.3
87.7 ± 11
BPSystolic< 121.5121.5 to 148.5>= 148.5
71.225.53.33
117 ± 16
HeartRate< 85.585.5 to 99.599.5 to 110.5>= 110.5
9.538.209.1073.2
109 ± 13
Sodium< 137.5137.5 to 140.5>= 140.5
39.338.222.5
138.5 ± 2.5
BUN< 13.5>= 13.5
45.754.3
Creatinine< 0.405>= 0.405
41.458.6
SpO2< 92.192.1 to 95.395.3 to 98.4>= 98.4
12.625.842.618.9
95.8 ± 3
BS_CONGESTIONYesNo
0.5999.4
BS_RHONCHIYesNo
0.5099.5
BS_ABNORMALYesNo
4.3095.7
BS_DECREASEDYesNo
2.5297.5
BS_COURSEYesNo
1.0599.0
BS_WHEEZESYesNo
3.0597.0
BS_NO_COUGHYesNo
0 + 100
BS_STRIDORYesNo
0.1899.8
BS_CLEARYesNo
42.357.7
BS_CRACKLESYesNo
0.8699.1
BS_RALESYesNo
0.1499.9
BS_ABSENTYesNo
.031 100
BS_INSPIRATIONYesNo
0.8499.2
BS_TUBULARYesNo
.025 100
BS_INFREQUENTYesNo
0.6999.3
BS_STRONGYesNo
0.8899.1
BS_FINE_CRACKLESYesNo
0.3899.6
BS_NOT_CLEARING_SECREATIONYesNo
0.1099.9
BS_WEAKYesNo
0.1999.8
BS_NON_PRODUCTIVE_COUGHYesNo
1.9298.1
BS_PRODUCTIVE_COUGHYesNo
2.0997.9
BS_MODERATEYesNo
1.5498.5
BS_CLEARING_SECREATIONYesNo
0.5399.5
ChiefComplaintRESPIRATORY COMPLAINTFEVERABD PAINORTHO INJURYCHEST PAINABD PROBLEMSNEURO COMPLAINTFALLTRAFFIC INJURYWEAKNESSENT PROBLEMBODY ACHESCHEST PRESSUREBACK PAINHEADACHESYNCOPEDIZZYCV COMPLAINTSFLANK PAINSKIN COMPLAINTCV PROBLEMSASSAULT RAPEPSYCHIATRICCHEST HEAVINESSSPECIFIC DIAGNOSISDIABETICFLUID NUTRITIONGENITOURINARY PROBLEMPAIN CHESTHEART RACEINFECTIONCONVULSIONSALLERGIESTRAUMAHEART IRRPALPITATIONSHIGH BPRAPID HRLACERATIONIRR HEARTBEATINGESTIONBP HIGHUNCONSCIOUSNESSVAGINAL BLEEDINGMED REFILLUNKNOWNEYE PROBLEMLOW BPCARDIAC ARRESTBP LOWother-
32.715.35.433.353.263.203.112.802.712.452.422.402.332.321.851.761.421.411.390.730.690.680.670.610.430.350.330.320.300.280.270.260.260.240.230.220.200.140.140.120.120.11.092.089.082.066.052.049.044.0400.16
NLP_FINDINGPositiveNegative
39.860.2
BS_FREQUENTYesNo
1.3798.6
BS_EXPIRATIONYesNo
0.9999.0
PNEUMONIAAbsentPresent
86.613.4
Imbed logic, orders into process of care Imbed logic, orders into process of care
Alerting for Pneumonia in the Patient Tracking System(Screening Framework (CEP) Watches Patient Data/Alerts for Pneumonia)
Admission/Discharge Recommendations(Admission determinations based on severity and risk factors)
18
Triage
Diagnostic Testing
Antibiotic Selection
System Produces Individualized Recommendations:
Advanced CDS Delivery Framework• Goals:
• Rapidly develop and deploy clinical protocols• Maintain protocols over time
• Attributes• Standardized CDS management tools• Multi-component inferencing environment• Broad (standardized) data access• Access to key care-oriented services (ordering, etc.)• Multiple, flexible alerting channels• Standardized, component-based client environment• Formal workflow authoring/delivery system• Broad workflow logging system
Questions???
Comments and Questions