enhancing medical evidence discovery through interactive pattern recognition and process mining
TRANSCRIPT
ENHANCING MEDICAL EVIDENCE DISCOVERY THROUGH INTERACTIVE PATTERN
RECOGNITION
V. TRAVER, A. MARTINEZ- ROMERO, J.L. BAYO, P. SALA, P. CARVALHO, J. HENRIQUES, M. G. RUANO, A. BIANCHI, C. FERNÁNDEZ- LLATAS
Madrid, 8th April 2016
Corresponding author: Vicente Traver [email protected]
Surrounded by heterogenous clinical data, medical staff needs to takethe right decision supported by medical evidence
Translational research is now happening in the healthfield, making use of business processes and workflows
in daily practice
Are we using evidence based medicine to take our decisions?
How are we defining our clinical processes? Is it done univocally?
Are the clinical pathways the solution? Are they understandableby the clinicians?
Are they used in real clinical practice?
Using Workflows to describe Clinical Pathways• Facilitate the praxis of health professionals• Improvement of quality of care• Unify criteria• Help the administrative management of clinical processes
DESIGN OF CLINICAL PATHWAYS
• Automation and Traceability• Measure and OptimizationProcess
Standardization
• Complete, Unambiguous formal processes • Expressivity Vs Understandability
Workflows as automation Language
• Complete and explicit definition needed• Differences between actual and perceived process• High timing consuming• High knowledge of representation language
needed
Difficulties in the process design
using workflows
PROCESS MINING
• Not possible from typical databasesObtain a model from the execution logs
• Set of workflow execution logs used to train the modelCorpus
• Parallel Activity-based Log Inference Algorithm (PALIA)
Our Vision: Activity-Based Workflow Mining
…27/05/2013 19:31:18.65 => i:9 Fin Accion: FirstTriageResult Res: ACUTE27/05/2013 19:31:18.68 => i:9d4fcabf Nodo: Get BP -> InicioAccion: getBloodPressure27/05/2013 19:31:18.69 => i:9d4fcabf Nodo: Get BP -> InicioReloj: clk427/05/2013 19:31:18.71 => i:9d4fcabf Nodo: Get Temp -> InicioAccion: getTemperature27/05/2013 19:31:18.72 => i:9d4fcabf Nodo: Get Temp -> InicioReloj: clk327/05/2013 19:31:18.83 => i:8229056d Fin Accion: FirstTriageResult Res: ACUTE27/05/2013 19:31:18.88 => i:8229056d Nodo: Get BP -> InicioAccion: getBloodPressure27/05/2013 19:31:18.90 => i:8229056d Nodo: Get BP -> InicioReloj: clk427/05/2013 19:31:18.91 => i:8229056d Nodo: Get Temp -> InicioAccion: getTemperature27/05/2013 19:31:18.93 => i:8229056d Nodo: Get Temp -> InicioReloj: clk327/05/2013 19:31:19.47 => i:1a5c92f6 Fin Accion: getTemperatureResult Res: FEVER27/05/2013 19:31:20.19 => i:aa3380da Fin Accion: getTemperatureResult Res: OK27/05/2013 19:31:20.27 => i:aa3380da Nodo: TEMP OK -> InicioAccion: 27/05/2013 19:31:20.38 => i:aa3380da Fin Accion: getBloodPressureResult Res: OK27/05/2013 19:31:20.40 => i:aa3380da Nodo: BP OK -> InicioAccion: 27/05/2013 19:31:20.41 => i:aa3380da Nodo: Quality Test -> InicioAccion: QualityTest…
PROCESS MINING
Process Mining can infer the real deployment of processes• Computer Aided Design of Clinical Pathways• Clinical Pathways tracing • Detect BottleNecks• Cost Effectiveness study of Processes• Measure adherence of Clinical Pathways instances
Processes Deployment Support
Individualized behavior Modeling
Discovery
Conformance
EnhancementAnalysis of processes through Process Mining
…27/05/2013 19:31:18.65 => i:9 Fin Accion: FirstTriageResult Res: ACUTE27/05/2013 19:31:18.68 => i:9d4fcabf Nodo: Get BP -> InicioAccion: getBloodPressure27/05/2013 19:31:18.69 => i:9d4fcabf Nodo: Get BP -> InicioReloj: clk427/05/2013 19:31:18.71 => i:9d4fcabf Nodo: Get Temp -> InicioAccion: getTemperature27/05/2013 19:31:18.72 => i:9d4fcabf Nodo: Get Temp -> InicioReloj: clk327/05/2013 19:31:18.83 => i:8229056d Fin Accion: FirstTriageResult Res: ACUTE27/05/2013 19:31:18.88 => i:8229056d Nodo: Get BP -> InicioAccion: getBloodPressure27/05/2013 19:31:18.90 => i:8229056d Nodo: Get BP -> InicioReloj: clk427/05/2013 19:31:18.91 => i:8229056d Nodo: Get Temp -> InicioAccion: getTemperature27/05/2013 19:31:18.93 => i:8229056d Nodo: Get Temp -> InicioReloj: clk327/05/2013 19:31:19.47 => i:1a5c92f6 Fin Accion: getTemperatureResult Res: FEVER27/05/2013 19:31:20.19 => i:aa3380da Fin Accion: getTemperatureResult Res: OK27/05/2013 19:31:20.27 => i:aa3380da Nodo: TEMP OK -> InicioAccion: 27/05/2013 19:31:20.38 => i:aa3380da Fin Accion: getBloodPressureResult Res: OK27/05/2013 19:31:20.40 => i:aa3380da Nodo: BP OK -> InicioAccion: 27/05/2013 19:31:20.41 => i:aa3380da Nodo: Quality Test -> InicioAccion: QualityTest…
Discovery
Conformance
EnhancementAnalysis of processes through Process Mining
…27/05/2013 19:31:18.65 => i:9 Fin Accion: FirstTriageResult Res: ACUTE27/05/2013 19:31:18.68 => i:9d4fcabf Nodo: Get BP -> InicioAccion: getBloodPressure27/05/2013 19:31:18.69 => i:9d4fcabf Nodo: Get BP -> InicioReloj: clk427/05/2013 19:31:18.71 => i:9d4fcabf Nodo: Get Temp -> InicioAccion: getTemperature27/05/2013 19:31:18.72 => i:9d4fcabf Nodo: Get Temp -> InicioReloj: clk327/05/2013 19:31:18.83 => i:8229056d Fin Accion: FirstTriageResult Res: ACUTE27/05/2013 19:31:18.88 => i:8229056d Nodo: Get BP -> InicioAccion: getBloodPressure27/05/2013 19:31:18.90 => i:8229056d Nodo: Get BP -> InicioReloj: clk427/05/2013 19:31:18.91 => i:8229056d Nodo: Get Temp -> InicioAccion: getTemperature27/05/2013 19:31:18.93 => i:8229056d Nodo: Get Temp -> InicioReloj: clk327/05/2013 19:31:19.47 => i:1a5c92f6 Fin Accion: getTemperatureResult Res: FEVER27/05/2013 19:31:20.19 => i:aa3380da Fin Accion: getTemperatureResult Res: OK27/05/2013 19:31:20.27 => i:aa3380da Nodo: TEMP OK -> InicioAccion: 27/05/2013 19:31:20.38 => i:aa3380da Fin Accion: getBloodPressureResult Res: OK27/05/2013 19:31:20.40 => i:aa3380da Nodo: BP OK -> InicioAccion: 27/05/2013 19:31:20.41 => i:aa3380da Nodo: Quality Test -> InicioAccion: QualityTest…
Discovery
Conformance
EnhancementAnalysis of processes through Process Mining
…27/05/2013 19:31:18.65 => i:9 Fin Accion: FirstTriageResult Res: ACUTE27/05/2013 19:31:18.68 => i:9d4fcabf Nodo: Get BP -> InicioAccion: getBloodPressure27/05/2013 19:31:18.69 => i:9d4fcabf Nodo: Get BP -> InicioReloj: clk427/05/2013 19:31:18.71 => i:9d4fcabf Nodo: Get Temp -> InicioAccion: getTemperature27/05/2013 19:31:18.72 => i:9d4fcabf Nodo: Get Temp -> InicioReloj: clk327/05/2013 19:31:18.83 => i:8229056d Fin Accion: FirstTriageResult Res: ACUTE27/05/2013 19:31:18.88 => i:8229056d Nodo: Get BP -> InicioAccion: getBloodPressure27/05/2013 19:31:18.90 => i:8229056d Nodo: Get BP -> InicioReloj: clk427/05/2013 19:31:18.91 => i:8229056d Nodo: Get Temp -> InicioAccion: getTemperature27/05/2013 19:31:18.93 => i:8229056d Nodo: Get Temp -> InicioReloj: clk327/05/2013 19:31:19.47 => i:1a5c92f6 Fin Accion: getTemperatureResult Res: FEVER27/05/2013 19:31:20.19 => i:aa3380da Fin Accion: getTemperatureResult Res: OK27/05/2013 19:31:20.27 => i:aa3380da Nodo: TEMP OK -> InicioAccion: 27/05/2013 19:31:20.38 => i:aa3380da Fin Accion: getBloodPressureResult Res: OK27/05/2013 19:31:20.40 => i:aa3380da Nodo: BP OK -> InicioAccion: 27/05/2013 19:31:20.41 => i:aa3380da Nodo: Quality Test -> InicioAccion: QualityTest…
CONCLUSIONS AND NEXT STEPS
Major challenges to speed up scalability
•Integration with legacy systems•Ease of use•Awareness about possibilities in
clinical daily practice
ENHANCING MEDICAL EVIDENCE DISCOVERY THROUGH INTERACTIVE PATTERN
RECOGNITIONV. TRAVER, A. MARTINEZ- ROMERO, J.L. BAYO, P. SALA, P. CARVALHO, J.
HENRIQUES, M. G. RUANO, A. BIANCHI, C. FERNÁNDEZ- LLATAS
Madrid, 8th April 2016
Corresponding author: Vicente Traver [email protected]
THANKS FOR YOUR ATTENTION