Upload
arnold-rudorfer
View
572
Download
1
Embed Size (px)
DESCRIPTION
Software-intensive medizinische Systeme stehen in einem immensen Marktdruck. Während sie technologisch und hinsichtlich ihrer Sicherheit kompromisslos innovativ sein müssen, fordern die Krankenhäuser und auch Gesundheitskostenreformen eine immer kürzere Zykluszeit bei gleichzeitig angespannten Budgets. Traditionelle Entwicklungsprozesse, die Innovation und Qualität über einen schwerfälligen Prozess erreichten, sind nicht mehr zeitgemäß. Unsere Studien zeigen, dass in der Medizintechnik die Kosten für Nacharbeiten über den Produktlebenszyklus hinweg durch eine Verbesserung des Requirements Engineering um 30-50% gesenkt werden können. Requirements Engineering hat daher eine Schlüsselstelle als Erfolgsfaktor im Gesundheitswesen. Modellbasierte Vorgehensweisen unterstützt die durchgängige Entwicklung von Anforderungen bis zur Validierung. Durch den erhöhten Abstraktionsgrad zu Beginn bei der anforderungsentwicklung und Analyse sind Problembeschreibungen wesentlich klarer, einfacher und weniger redundant. Das erhöht nicht nur die Entwicklungsgeschwindigkeit, sondern sorgt innerhalb des Projektes für klar verstandende Domänenkonzepte. Modelle helfen bei der Durchgängigkeit von den systemanforderungen zu den Softwareanforderungen und dann zu Design und Validierung. Unser Beitrag zeigt Industrieerfahrungen bei Siemens Healthcare und bietet Orientierung bei der Umsetzung modellbasierter Entwicklung mit Fokus auf Requirements Engineering in medizinischen Systemen. Wir zeigen, wie Anforderungen so formalisiert werden, dass einerseits informell formulierte Anforderungen schrittweise in eine formalere Form kommen und während die Verständlichkeit der Anforderungen erhalten bleibt. Damit sind die Anforderungen präzise formuliert und werden zur Konsistenzsicherung und Vervollständigung genutzt, und sie können weiterhin von Stakeholder verschiedenem Hintergrund verstanden und validiert werden.
Citation preview
2012 Siemens Healthcare Diagnostics Inc. Page 1© 2012 Siemens Healthcare Diagnostics Inc. All rights reserved.
Model-based Engineering in Medical Product Development
ReConf 2012 March 14, 2012
Munich, Germany
2012 Siemens Healthcare Diagnostics Inc. Page 2
Contents
Goals
Brief look on Siemens Healthcare
Business challenges
Model-based Engineering: Issues & Solutions
Recommendations
Further Information
2012 Siemens Healthcare Diagnostics Inc. Page 3
Contents
Goals
Brief look on Siemens Healthcare
Business challenges
Model-based Engineering: Issues & Solutions
Results and Summary
Further Information
2012 Siemens Healthcare Diagnostics Inc. Page 4
Goals of this Talk
Share the experiences using models in engineering gained at Siemens Healthcare
Discuss challenges, needs and requirements to advance model-driven product development
Show the concrete value of model-based engineering
2012 Siemens Healthcare Diagnostics Inc. Page 5
Contents
Goals
Brief look on Siemens Healthcare
Business challenges
Model-based Engineering: Issues & Solutions
Results and Summary
Further Information
2012 Siemens Healthcare Diagnostics Inc. Page 6
Siemens Healthcare
Immunoassay Clinical ChemistryMolecular Hematology Lab AutomationUrinalysis Point of Care
In vitro diagnostics (laboratory systems)
X-Ray ComputedTomography
MagneticResonance
MolecularImaging
Ultrasound
In vivo diagnostics (imaging)
IT Solutions
Oncology
2012 Siemens Healthcare Diagnostics Inc. Page 7
Vector Consulting ServicesYour Partner in Achieving Engineering Excellence
… offers a comprehensive consulting, tools and training portfolio for optimizing technical product development
… serves industries such as automotive, aviation, IT and telecom, energy and environment, medical and railway
… is supporting clients worldwide on efficiency improvement, requirements engineering, ALM/PLM, systems engineering, product management, interim management, and organizational change
… as a group serves companies across the world with over 1000 employees and sales of well over 150 Mio € pa
www.vector.com/consulting
Railway &Transportation
IT & Telecom
Automotive
Aviation & Defense
Energy &Environment
Medical &Healthcare
2012 Siemens Healthcare Diagnostics Inc. Page 8
Contents
Goals
Brief look on Siemens and Vector
Project syngo.via
Business challenges
Lean Requirements Engineering
Results and Summary
Further Information
Contents
Goals
Brief look on Siemens Healthcare
Business Challenges
Model Management: Issues & Solutions
Results and Summary
Further Information
2012 Siemens Healthcare Diagnostics Inc. Page 9
Business Challenges in Medical Device Industry
Application Life-cycleManagement Tools
Model-basedEngineeringGlobally Integrated
Systems
PlatformsApplication Life-cycle
Management ToolsModel-based
Engineering and ReuseGlobally IntegratedSystems
Platforms
Focus of this talk
Sour
ces:
[1, 2
, 5]
2012 Siemens Healthcare Diagnostics Inc. Page 10
Project Context
Typical Large-Scale Medical Device Projects
Several thousand single product requirements
Several million lines of code
Several hundred developers in many locations worldwide
Many fold clinical applications
Source: H IM
2012 Siemens Healthcare Diagnostics Inc. Page 11
Disclaimer:The content discussed in this presentation needs
to be considered as work in progress.
2012 Siemens Healthcare Diagnostics Inc. Page 12
Contents
Goals
Brief look on Siemens and Vector
Project syngo.via
Business challenges
Lean Requirements Engineering
Results and Summary
Further Information
Contents
Goals
Brief look on Siemens and Vector
Business challenges
Model-based Engineering: Issues & Solutions
Results and Summary
Further Information
2012 Siemens Healthcare Diagnostics Inc. Page 13
Industry Issues and Pain Points
Pain Points Business Impact Intransparent product
structure, domain model partially incomplete
Technology-driven product platform, no link to businessdriversNon-transparent relationship between problem & solution spaceRe-scoping sessions with customers on basis of many specs.
Ambiguity and lack of accuracy in specifications
Textual-based specifications are subject to interpretation Textual use case descriptions work only for smaller projectsNatural language subject to interpretation, inconsistent, incomplete
Growing architecture complexity
Redundancy of architecture components due to lack ofunderstanding of problem- and solution spaceBusiness needs not consistently linked to features Too much variability in software architecture
Insufficient verification and validation efficiency
Test specifications in natural language are executed in a manual way only Thousands of pages of requirements to transform Test cases manually created
2012 Siemens Healthcare Diagnostics Inc. Page 14
Overview: Solutions along the Life-Cycle
D) Architecture Model Mapping
C) Graphical Modeling of Clinical Workflows
B) Feature Model
A) Requirements Meta-Model
E) Model-based System Test
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 15
Pain point 1: Intransparent Product Structure
Solutions:
A. Requirements Meta-modelB. Feature Model
Selected issues: Non-transparent relations
between problem- & solution space
Re-scoping sessions w/ customer on basis of many engineering specs.
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 16
Solution A: Requirements Meta Model
Benefits: Lean artefact infrastructure, no
redundancy Guidance for engineering tasks
with structured input Established link between business
drivers, requirements, design and test
Characteristics:
Provides needed artifacts, their attributes and relationships to each other. Using meta-model, it is also possible to
prescribe the way how / which data are captured .
SWRS
MRS
Market Requirement
SW Feature
SW Req.(Problem)
SW Req.(Solution)
DB
Business PlanMarket Analysis,
SW Platform Roadmap, Business Case, etc.
Stakeholder Request
input for
input for
Specification/Document Structural Element Content Mapping or
Input
AnalysisUse Case Specifications,
Load Profiles,Concept Papers
analyzedby
Problem Space
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 17
Solution B: Feature Model
Highest Level of Feature Model
Graphical ViewHierarchical View
Benefits: Higher level abstraction of grouping
of requirements into sellable units: From 5,000 product requirements to 800+ features (factor ~ 6)
Visual domain model for healthcare workflows (tree & graphical)
Reduction in time to understand aspects of the system
Characteristics:
Hierarchical tree to describe the structure, dependencies and commonalities Lays out the basics for variant
management and impact analysis
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 18
Solution B: Feature Model & Its Relations (1)
Application Use CasesStakeholder Requests
Architecture Model
How can I manage huge amounts ofRequirements?
How do I scope a product version efficiently?
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 19
Solution B: Feature Model & Its Relations (2)
Application Use CasesStakeholder Requests
Architecture ModelHow do I manage stakeholder requests from different businesses most effectively?
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 20
Pain point 2: Ambiguity and Lack of Accuracy of Specifications
Issues: Textual use case descriptions work only
for smaller projects < ~ 100 requirements
Natural language subject to interpretation, usually inconsistent, incomplete with incorrect version (and conflicting)
Root causes: Textual requirements engineering do not
scale for platform projects Missing versioning No direct access to single requirements Lack of product structure Inconsistently executed change
management process
Level Requts.Object
Manually Embedded graphs
Manual testcase creation
Document
Features
Paragraph
Req. 1(text)
Req. 2(text)
Req. 3(text)
Picture / diagram
Picture / diagram
Picture / diagram
Solutions:
C. Graphical Modeling of Clinical Workflows
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 21
Characteristics: Used to describe clinical workflows that
consist of a collection of steps in a defined sequence together with accompanying specification of pre-/post-conditions, business rules, performance aspects, etc.
Benefits: Increase expressiveness of clinical
workflows to describe dynamic behaviors of clinical workflows Early analysis of stakeholder requests
from customers Improved impact analysis of change
requests Joint modeling sessions to describe the
needs from the customer‘s point of view
Solution C: Graphical Modeling of Clinical Workflows (1)
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 22
Solution C: Graphical Modeling of Clinical Workflows (2)
Application Use CasesStakeholder Requests
Architecture ModelHow can I analyze a feature or a stakeholder request?
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 23
Pain point 3: Growing Architecture Complexity
Solutions:
D. Architecture Model Mapping
Selected issues: Business needs not consistently
linked to features/ requirements; dependencies between features not easily visible
Too much variability in software architecture
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 24
Solution D: Architecture Model Mapping
Characteristics:
Identifies links between features and their
implementation
Explicit modeling of variability in the architecture
Benefits:
Architectural decisions motivated by features
and product-line variability
Enabling reduction of architectural complexity
Support impact analysis for (de-) scoping
sessions Early identification of architectural risks
Improved accuracy of early effort estimates
Reduction of number of scoping sessions
F
F F F
SWF SWF SWF SWF SWF SWF SWF
S
SS SS SS
C C C C C C
Feature Model
Architecture Model
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 25
Pain point 4: Insufficient Verification and Validation Efficiency
Solutions:
E. Model-based System Test
Issues: Test cases often manually
generated from prose use cases Cycle times for system test too
long
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 26
Solution E: Model-based System Test (1)
Benefits: Early identification of requirements
defects through validation by testers Effort reduction for test (cycle time,
cost ~ -30%) and increase of test coverage (*)
Decrease number of defects
Model-based testing is highly structured, reproducible and efficient Increase reuse of development
artifacts Quicker impact analysis by parsing
model for late requirements changes
(*) Based on experience from other projects
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 27
Solution E: Model-based System Test (2)
Application Use CasesStakeholder Requests
Architecture Model
Which application use case is covered by which test case?
Sou
rce:
H IM
2012 Siemens Healthcare Diagnostics Inc. Page 28
Contents
Goals
Brief look on Siemens and Vector
Project syngo.via
Business challenges
Lean Requirements Engineering
Results and Summary
Further Information
Contents
Goals
Brief look on Siemens and Vector
Business challenges
Model-based Engineering: Issues & Solutions
Results and Summary
Further Information
2012 Siemens Healthcare Diagnostics Inc. Page 29
Major Changes to Development forModel-based Engineering
Sou
rce:
H IM
D) Architecture Model Mapping
C) Graphical Modeling of Clinical Workflows
B) Feature Model
A) Requirements Meta-Model
E) Model-based System Test
2012 Siemens Healthcare Diagnostics Inc. Page 30
Benefits from Model-based Engineering
Feature models can help to reduce detail of product requirements through abstraction (~ factor 6) [6]
Reduction of review times by ~ 40% using graphical modeling of product requirements [9]
Reduction of analysis time for requirements changes
Model-based testing reduces testing effort by ~ 30% through automation
Reduction in overall cycle time for development projects driving more efficiency [3, 6]
2012 Siemens Healthcare Diagnostics Inc. Page 31
Experiences with Introducing Model-based Engineering
Model-based engineering means a huge organizational change. Only 10% of organizations in medical have moved that road so far [10]
Modeling competence of product managers and requirements engineers needs to be built up step-by-step thereby facilitating acceptance
Seamless model-driven engineering needs adequate tools support.
Continuous assessment and verification of business benefits for model-based engineering is mandatory to maintain sponsorship.
Model-basedEngineering
Processes
Tools Persons
2012 Siemens Healthcare Diagnostics Inc. Page 32
Thank you for your attention!
© Siemens AG 2012. All rights reserved.
2012 Siemens Healthcare Diagnostics Inc. Page 33
Contents
Goals
Brief look on Siemens and Vector
Project syngo.via
Business challenges
Lean Requirements Engineering
Results and Summary
Further Information
Contents
Goals
Brief look on Siemens and Vector
Business challenges
Model-based Engineering: Issues & Solutions
Results and Summary
Further Information
2012 Siemens Healthcare Diagnostics Inc. Page 34
References
1. US Food & Drug Administration, Design Control Guidance for Medical Device Manufacturers; March 11, 19972. US Food & Drug Administration, Quality System Regulation,; January 1, 1997, http://www.fda.org/cdrh/qsr/01qsreg.html 3. Brian Berenbach, Daniel Paulish, Arnold Rudorfer, Juergen Kazmeier, Software Systems Requirements Engineering; Mc-
Graw Hill 2009; http://www.mhprofessional.com/product.php?isbn=00716054794. Renate Loeffler: Formal Scenario-based Requirements Specification and Test Case Generation in Healthcare
Applications, Diplomarbeit, Univ. Paderborn, 10/20095. Medical Device & Diagnostic Industry 101, Carey Smoak, Roche Molecular Systems Inc., Pleasonton, CA, Pharma SUG
2010, Paper 1B016. Arnold Rudorfer, Christof Ebert: Lean Requirements Engineering in Medical Systems, MedConf 2010, Munich, Germany,
October 14, 2010; http://2010.medconf.de/downloads/abstracts2010/T2_T3_V1_vector_siemens.pdf7. Arnold Rudorfer, Christof Ebert: Quality Requirements Engineering in in Medical Systems, OOP11, Munich, Germany,
January 27, 2011 http://www.sigs.de/download/oop_2011/downloads/files/Do2-3_Ebert_Rudorfer_Update_QualityRE_OOP2011.pdf
8. Arnold Rudorfer: Model-based Engineering in Medical Device Development – Issues & Solutions, INCOSE Workshop, Phoenix, AZ, USA, January 31, 2011, http://www.omgwiki.org/MBSE/lib/exe/fetch.php?media=mbse:2011iw_incose_modelmanagement_rudorfer.pdf
9. Brian Berenbach: Comparison of UML and Text based Requirements Engineering, http://www.cs.helsinki.fi/u/vjkuusel/docs/Comparison%20of%20UML%20and%20text%20based%20requirements%20engineering%20Brian%20Berenbach.pdf
10. Christian Denger, Raimund Feldman: State of the Practice in Software Development for Medical Device Production, Fraunhofer Institute Experimentelles Software Engineering, February 2007
2012 Siemens Healthcare Diagnostics Inc. Page 35
Arnold RudorferProgram Manager System PlatformSiemens Healthcare Diagnostics62 Flanders-Bartley RoadFlanders, NJ, 08736 (U.S.A.)
Phone: +1 973 927 28 28 ext. 4732Cell: +1 609 954 2384
Email:[email protected]
2012 Siemens Healthcare Diagnostics Inc. Page 36
Dr. Christof EbertManaging DirectorVector Consulting Services GmbH
Ingersheimerstrasse 24D-70499 Stuttgart
Phone: +49 711 – 80670-1525Fax: +49 711 – 86070-444
Email:[email protected]
2012 Siemens Healthcare Diagnostics Inc. Page 37
Documented Experiences and Best Practices from various Industry Projects
Link to web site McGrawHill
English language:
Software & Systems Requirements Engineering: In Practice
2009
McGrawHill
German language:
Systematisches Requirements Engineering
Fourth edition, 2012
Dpunkt.verlagLink to web site Dpunkt