CONTINUOUS CLINICAL PATHWAYS EVALUATION BY USING AUTOMATIC LEARNING ALGORITHMS
3 likes1,146 views
This document discusses the use of automatic learning algorithms for the continuous evaluation of clinical pathways to improve healthcare standardization and efficiency. It highlights the challenges of designing workflows for clinical processes, the application of activity-based process mining, and the development of a clinical pathways process miner tool for analyzing deviations and costs. The authors propose an iterative approach to refining clinical pathways through process mining techniques, enabling continuous improvement and error correction.
CONTINUOUS CLINICAL PATHWAYS EVALUATION BY USING AUTOMATIC LEARNING ALGORITHMS
- 1. CONTINUOUS CLINICAL PATHWAYS EVALUATION BY USING AUTOMATIC LEARNING ALGORITHMSCarlos Fernández-Llatas, Teresa Meneu, Jose Miguel Benedí andVicente Travervtraver@itaca.upv.es
- 2. IntroductionUsing Clinical Pathways for Health Care standardizationWorkflow TechnologyActivity-Based Process Mining for Clinical Pathways automatic LearningClinical Pathways Process MinerConclusions and Future WorkIndex
- 3. Clinical pathwaysFormal care protocolsEfficient and coordinated use of resourcesFrom the evidence based medicine approachAdvantagesFacilitate the praxis of health professionalsImprovement of quality of careUnify criteriaHelp the administrative management of clinical processesUsing Clinical Pathways for Health Care standardization
- 4. Using Clinical Pathways for Health Care standardizationProblems
- 5. Complexity
- 6. Subjectivity
- 9. Without adequate ICT support Workflow TechnologyWorkflowTheautomation of a business process, in whole or part, during which documents, information or tasks are passed from one participant to another for action, according to a set of procedural rulesWorkflow Management Coalition Glossary
- 10. Workflow TechnologyWorkflowFormal description of a process designed to be automatedReadable by expertsCan be automatically executed by using workflow enginesWorkflows can help Clinical Pathways Standardization
- 11. Workflow TechnologyDifficulties in the process designComplete and explicit definition neededDifferences between actual and perceived processLarge time process designHigh knowledge of representation language needed
- 12. Process MiningProcess MiningObtain a model from the execution logsCorpusSet of workflow execution logs used to train the modelCan be used in an iterative way to refine continuously clinical pathwaysEvaluation of deployed Clinical GuidelinesHuman readable and modifiable inferred model
- 13. Workflow TechnologyEvent Based Process MiningInference from system eventsInsufficient for clinical pathways inferenceGettemperatureMedicateYesFeverNoDo nothing
- 14. Activity-Based Process Mining for Clinical Pathways automatic LearningActivity Based Process MiningInference from Activities Corpus with information of start, duration and result of actions07/05/2010 10:22:59 => i:1 BeginAction: Admission07/05/2010 10:34:01 => i:1 EndAction: Admission Res: OK07/05/2010 10:35:03 => i:1 BeginAction: Triage07/05/2010 10:46:06 => i:1 EndAction: Triage Res: InHospital07/05/2010 10:56:07 => i:1 BeginAction: TNS07/05/2010 10:56:08 => i:1 BeginAction: TMP07/05/2010 11:23:10 => i:1 EndAction: TNS Res: OK07/05/2010 11:33:12 => i:1 EndAction: TMP Res: Fever07/05/2010 11:43:19 => i:1 BeginAction: TMP07/05/2010 11:50:19 => i:1 EndAction: TMP Res: OK07/05/2010 11:51:24 => i:1 BeginAction: QualityTest07/05/2010 12:00:28 => i:1 EndAction: QualityTestRes: OK07/05/2010 12:15:29 => i:1 BeginAction: Discharge07/05/2010 12:20:30 => i:1 EndAction: DischargeRes: OK
- 15. Activity-Based Process Mining for Clinical Pathways automatic LearningPALIA (Parallel Activity-based Log Inference Algorithm)
- 16. Input: Activity based log samples
- 17. Output: TPA (Timed Parallel Automaton)
- 18. 4 Subalgorithms
- 19. Parallel prefix aceptor tree
- 20. Parallel Merge
- 21. Onward Merge
- 22. Elimination of repeated transitions and not useful statesAlgorithmdesignedwithin Carlos Fernandez’ thesis
- 23. TPA (Timed Parallel Automaton)Based on parallel finite automaton approachAlphabet enriched with discrete temporal labelsAble to describe complex transitionsDescription of parallelism and synchronization patternsDouble function of transition:g: N+ N+d: Q Q where: Q is a set of nodes N Activity-Based Process Mining for Clinical Pathways automatic Learning
- 24. Activity-Based Process Mining for Clinical Pathways automatic LearningParallelprefix aceptor treeActivity-Based Process Mining for Clinical Pathways automatic LearningParallelMergeActivity-Based Process Mining for Clinical Pathways automatic LearningOnwardMergeActivity-Based Process Mining for Clinical Pathways automatic LearningEliminate repeated transitions and not useful statesClinical Pathways Process MinerTool for Process Mining over deployed Clinical Pathways Analysis of deviations on Clinical Pathways executionsComparison between real execution and designed pathwayDiscovery of new Clinical Pathways patternsAnalysis of Costs produced of Clinical Pathways deviationsCost evaluation of inefficient casesBottle-necks detection.Currently testing with experts in Heart Cycle projectClinical Pathways Process Miner
- 28. Process Mining Techniques can be used to help in the Clinical Pathways design processA Clinical Pathway can be approached in an iterative wayErrors and new patterns can be detected and corrected in each iterationThe process can be continuously improved designing more and more cost-efficient Clinical PathwaysConclusions
- 29. CONTINUOUS CLINICAL PATHWAYS EVALUATION BY USING AUTOMATIC LEARNING ALGORITHMSCarlos Fernández-Llatas, Teresa Meneu, Jose Miguel Benedí andVicente Travervtraver@itaca.upv.es