ViNNSL The Vienna Neural Network Specification Language

Download Report

Transcript ViNNSL The Vienna Neural Network Specification Language

Institute for Information
Systems & New Media
Guideline Compliance for Skin Cancer Treatment
Processes:
The Evidence-Based Medical Compliance
Cluster (EBMC2)
Univ.-Prof. Dr. Stefanie Rinderle-Ma
Dr. Stefanie Weber
University of Vienna
Faculty of Computer Science
Research Group Workflow Systems and Technology
[email protected]
Department of Dermatology
Division of General Dermatology
Medical University of Vienna
[email protected]
BPM European Round Table Meeting, Eindhoven, 5 November 2012
Outline
Institute for Information
Systems & New Media
1 Introduction EBMC2
2 Data Sources
3 First Results
4 Summary
©Stefanie Rinderle-Ma, University of Vienna 2012
2
Institute for Information
Systems & New Media
Institute for Information
EBMC2: Evidence-Based Medicine Compliance
Cluster
Systems
& New Media
 Interdisciplinary cluster consisting of participants from
different groups within
University of Vienna
 Medical University of Vienna / General Hospital of Vienna






Funded by University of Vienna and Medical University of
Vienna
http://ebmc2.univie.ac.at/
Combining the knowledge of all participating members and
their domains
Process oriented view of treatment and follow-up
Treatment of melanoma patients as first use case
©Stefanie Rinderle-Ma, University of Vienna 2012
4
Institute for Information
Systems & New Media
Pilot Project: Skin Cancer
 Skin cancer constitutes a global major health problem
 Incidence of skin cancer is worldwide increasing
 Clinically validated, evidence-based guidelines are available
Institute for Information
Systems & New Media
Non-Melanoma Skin Cancer (NMSC)
 NMSC is worldwide the most common cancer among
Caucasian populations
 2 major forms: BCC, SCC
 Most NMSC occur on skin that is regularly exposed to
sunlight or other ultraviolet radiation
 Other risk factors:
 sun
burns
 light-colored or freckled skin
 exposure to arsenic (BCC)
 family history
 genetics (e.g. Goltz-Gorlin syndrome, XP)
NMSC – BCC
P. Altmeyer Enzyklopädie der Dermatologie, Venerologie, Allergologie, Umweltmedizin; Springer
Institute for Information
Systems & New Media
NMSC – SCC
P. Altmeyer Enzyklopädie der Dermatologie, Venerologie, Allergologie, Umweltmedizin; Springer
Institute for Information
Systems & New Media
Institute for Information
Systems & New Media
Malignant Melanoma (MM)
 Accounting for about 5% of all skin cancer cases, MM is the
most dangerous form of skin malignancy causing 90% of of
skin cancer mortality
 Stages: AJCC I - IV (American Joint Committee on Cancer)
 Early detection is crucial
 Risk factors:
 ultraviolet
radiation (UV-A & UV-B)
 sun burns in early childhood
 light-colored or freckled skin
 family history
 multiple nevi
Malignant Melanoma
P. Altmeyer Enzyklopädie der Dermatologie, Venerologie, Allergologie, Umweltmedizin; Springer
Institute for Information
Systems & New Media
Institute for Information
Systems & New Media
1953 - 2002
Male: 1,5 to 12,8
Female: 1,8 to 10,4
Stang, A., E. Pukkala, et al. (2006). "Time trend analysis of the skin melanoma incidence of Finland from 1953 through 2003 including 16,414
cases." Int J Cancer 119(2): 380-4.
Institute for Information
Systems & New Media
Incidence
Mortality
Age std. Incidence Rates Melanoma
Institute for Information
Systems & New Media
Annual increase in incidence 3% - 7%
Garbe, C. and U. Leiter (2009). "Melanoma epidemiology and trends." Clin Dermatol 27(1): 3-9.
EBMC2
Institute for Information
Systems & New Media
Aims of
1) Development of methodology for the process analysis
of medical treatment processes
 to identify existing treatment and follow-up processes and
compare these processes with existing guidelines
(conformance analysis)
 to investigate cancer incidence and prevalence in Austria
 to evaluate the association of melanoma with other tumor
entities
 to explore and develop specific processes for early cancer
detection (e.g., by identifying incidence patterns of high-risk
patient collectives)
 to determinate effective and cost-saving follow-up strategies
conditional on a variety of treatment protocols
EBMC2
Institute for Information
Systems & New Media
Aims of
2) Definition of a data structure
 that allows formal integration of data sources with different
structures and content (e.g. PTDoc application)
for Information
EBMC2: Evidence-Based Medicine ComplianceInstitute
Cluster
Systems & New Media
 First step: Formal modeling based on textual guidelines, e.g., European
Guideline 1

Insights:



1Garbe,
Different aspects: process structure, time, data
Quite rough / general  compromise of different parties
Consequence: variability of real-world processes
C., Peris, K., Hauschild, A., Saiag, P., Middleton, M., Spatz, A., Grob, J.,Malvehy, J., Newton-Bishop, J., Stratigos, A., Pehamberger, H., Eggermont, A.:
Diagnosis and treatment of melanoma: European consensus-based interdisciplinary guideline. European Journal of Cancer 46(2), 270{283 (2010)
Institute for Information
Systems & New Media
Compliance analysis
Process Compliance
Process Mining
Data Integration
S4MDB
GAP-DRG
ACR
Life-Cycle Support for Compliance
Compliance
Source: ©L.T. Ly, Presentation@CoopIS 2011
Institute for Information
Systems & New Media
Monitoring
Outline
Institute for Information
Systems & New Media
1 Introduction EBMC2
2 Data Sources
3 First Results
4 Summary
©Stefanie Rinderle-Ma, University of Vienna 2012
19
Stage IV Melanoma Database




Institute for Information
Systems & New Media
Department of Dermatology, General Hospital of Vienna
Basic medical information on melanoma and treatment in
patients with stage IV melanoma
Information about follow-up treatments in stage IV patients
Major objectives are
 the
progress of patients with melanoma,
 the median survival rate and
 the influence of different treatment options on survival
cf. Michael Binder, Wolfgang Dorda, Georg Duftschmid, Reinhold Dunkl, Karl Anton Fröschl, Walter Gall, Wilfried Grossmann, Kaan
Harmankaya, Milan Hronsky, Stefanie Rinderle-Ma, Christoph Rinner, Stefanie Weber: On Analyzing Process Compliance in Skin Cancer
Treatment: An Experience Report from the Evidence-Based Medical Compliance Cluster (EBMC2). 398-413
Institute for Information
Systems & New Media
Austrian Social Security Institution Data (GAP-DRG)





Hospitalization data
Patient treatments received from general practitioners
Administered medications dispensed at pharmacies
Sick certificates
Major objectives are
 accounting
and
 billing
cf. Michael Binder, Wolfgang Dorda, Georg Duftschmid, Reinhold Dunkl, Karl Anton Fröschl, Walter Gall, Wilfried Grossmann, Kaan
Harmankaya, Milan Hronsky, Stefanie Rinderle-Ma, Christoph Rinner, Stefanie Weber: On Analyzing Process Compliance in Skin Cancer
Treatment: An Experience Report from the Evidence-Based Medical Compliance Cluster (EBMC2). 398-413
Institute for Information
Systems & New Media
New data sources:
Austrian Cancer Register Database (Statistik Austria)





Cancer registry of Austrian citizens
Data sources: cancer report forms (paper based)
Reporting units: hospitals, pathology institutes
Reporting mandatory
Major objectives are
Incidence, prevalence, morbidity and mortality
 Geographical distribution

Japanese Cancer Register Database

Similar to Austrian cancer registry
Institute for Information
Systems & New Media
Data Source Integration and Assessment


Institute for Information
Systems & New Media
Time-related data for process mining
Routing data for decision point analysis
Stage IV Melanoma Database


Few time-related data
Many routing data
Austrian Social Security Institution Data (GAP-DRG)


Many time-related data
Few routing data
cf. Michael Binder, Wolfgang Dorda, Georg Duftschmid, Reinhold Dunkl, Karl Anton Fröschl, Walter Gall, Wilfried Grossmann, Kaan
Harmankaya, Milan Hronsky, Stefanie Rinderle-Ma, Christoph Rinner, Stefanie Weber: On Analyzing Process Compliance in Skin Cancer
Treatment: An Experience Report from the Evidence-Based Medical Compliance Cluster (EBMC2). 398-413
Outline
Institute for Information
Systems & New Media
1 Introduction EBMC2
2 Data Sources
3 First Results
4 Summary
©Stefanie Rinderle-Ma, University of Vienna 2012
25
Institute for Information
Systems & New Media
Meta Data Enrichment of the Stage IV Melanoma
Database



Data did not meet requirements for process analysis due to
lack of temporal references
Extension and data recollection
Adding meta data to compensate missing structural
information
 Missing
temporal information
 Imprecise temporal information


Restating the Data Model
Development of a generic and process-oriented data
structure to integrate different data sources
cf. Michael Binder, Wolfgang Dorda, Georg Duftschmid, Reinhold Dunkl, Karl Anton Fröschl, Walter Gall, Wilfried Grossmann, Kaan
Harmankaya, Milan Hronsky, Stefanie Rinderle-Ma, Christoph Rinner, Stefanie Weber: On Analyzing Process Compliance in Skin Cancer
Treatment: An Experience Report from the Evidence-Based Medical Compliance Cluster (EBMC2). 398-413
Institute for Information
Systems & New Media
Import and recollection
Source : R. Dunkl@CAiSE 2012
Institute for Information
Systems & New Media
recollection
Source: EBMC2 Intermediate report (2012)
Institute for Information
Systems & New Media
Source: EBMC2 Intermediate report (2012)
Mapping to MXML
Institute for Information
Systems & New Media
Source: EBMC2 Intermediate report (2012)
Institute for Information
Systems & New Media

10 patients from Stage IV Melanoma Database

Extending and adding data through recollection

Using the restated data model for integration

Mining a process model with the heuristic miner
from the Process Mining Framework ProM

Evaluation with domain experts to verify its
compliance with medical reality
cf. Michael Binder, Wolfgang Dorda, Georg Duftschmid, Reinhold Dunkl, Karl Anton Fröschl, Walter Gall, Wilfried Grossmann, Kaan
Harmankaya, Milan Hronsky, Stefanie Rinderle-Ma, Christoph Rinner, Stefanie Weber: On Analyzing Process Compliance in Skin Cancer
Treatment: An Experience Report from the Evidence-Based Medical Compliance Cluster (EBMC2). 398-413
Institute for Information
Systems & New Media
Data errors
Imprecise temporal information
cf. Michael Binder, Wolfgang Dorda, Georg Duftschmid, Reinhold Dunkl, Karl Anton Fröschl, Walter Gall, Wilfried Grossmann, Kaan
Harmankaya, Milan Hronsky, Stefanie Rinderle-Ma, Christoph Rinner, Stefanie Weber: On Analyzing Process Compliance in Skin Cancer
Treatment: An Experience Report from the Evidence-Based Medical Compliance Cluster (EBMC2). 398-413
Institute for Information
Systems & New Media
cf. Michael Binder, Wolfgang Dorda, Georg Duftschmid, Reinhold Dunkl, Karl Anton Fröschl, Walter Gall, Wilfried Grossmann, Kaan
Harmankaya, Milan Hronsky, Stefanie Rinderle-Ma, Christoph Rinner, Stefanie Weber: On Analyzing Process Compliance in Skin Cancer
Treatment: An Experience Report from the Evidence-Based Medical Compliance Cluster (EBMC2). 398-413
Outline
Institute for Information
Systems & New Media
1 Introduction EBMC2
2 Data Sources
3 First Results
4 Summary
©Stefanie Rinderle-Ma, University of Vienna 2012
Institute for Information
Systems & New Media





Development of methodology for process analysis of
medical treatment processes
Development of application (PTDoc) for formal integration
of data from different sources with varying data structure
International cooperations – experiences in collecting
“good” data
Improving data quality for process mining
Skin cancer as pilot project – applications in other chronic
diseases desirable
Institute for Information
Systems & New Media
Conclusion:
Process mining promises to be a useful tool for auditing
existing data on treatment processes. This might help to
facilitate that medical experts can easily follow the treatment
of a patient using the offered process oriented visualizations.