Clinical Care Pathway Extraction via Temporal Event Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing process mining approaches fail to correlate clinical care pathways with patient outcomes, leading to complex patterns and lack of insight into effective treatment pathways for diseases like congestive heart failure due to issues such as pattern explosion and diversity of events.
Innovation Solution
A method and system for data analysis that constructs patient traces, preprocesses them to reduce complexity, clusters similar traces, mines a process model to identify treatment pathways, and overlays discriminative patterns to correlate care pathways with patient outcomes, providing a visual interface for identifying key pathways associated with positive or negative outcomes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If existing process mining approaches are used to analyze patient event data, then treatment pathways can be extracted, but the data complexity explodes due to multiple concurrent events and diverse event types causing loops and spaghetti-like patterns
Solution Approach 1:
The patent segments the complex patient event data by introducing a temporal abstraction layer that divides events into discrete time points with associated event types. This segmentation transforms the spaghetti-like continuous data into structured temporal segments that can be processed independently, reducing overall complexity while preserving pathway information.
Solution Approach 2:
The patent introduces an intermediary temporal event model that acts as a mediator between raw patient event data and process mining algorithms. This intermediary layer standardizes diverse events into a common temporal framework, enabling existing process mining techniques to work effectively without directly handling the complexity of raw concurrent events.
2Loss of information
If process mining is applied to raw patient event data, then treatment pathways can be identified, but correlation with patient outcomes is not achieved
Solution Approach 1:
The patent merges process mining with outcome analysis by integrating patient outcome data with the temporal event model. This combination allows the system to simultaneously extract treatment pathways and correlate them with patient outcomes, eliminating information loss while maintaining analytical efficiency through unified processing.
Solution Approach 2:
The patent implements feedback by using patient outcomes to refine and validate extracted treatment pathways. Outcome data feeds back into the pathway extraction process, allowing the system to identify which pathways are actually correlated with specific outcomes, thereby improving insight generation efficiency through outcome-driven validation.
3Reliability
If multiple concurrent events are processed in patient traces, then comprehensive treatment information is captured, but pattern explosion occurs making analysis intractable
Solution Approach 1:
The patent applies preliminary action by pre-processing patient event data into a standardized temporal model before pattern extraction. Events are pre-organized by time points and types, with concurrent events structured in a predictable format. This preliminary organization captures all treatment information while preventing pattern explosion during subsequent analysis.
Solution Approach 2:
The patent changes parameters by transforming diverse patient events into a standardized temporal representation with controlled variables (time points, event types, patient identifiers). This parameter standardization maintains comprehensive treatment information while constraining pattern complexity to manageable levels through consistent data structure.
Data Source
AI summary
Systems and methods for data analysis include constructing patient traces as a set of medical events for each patient of a patient population, the patient population being segmented based on patient outcomes. Medical events in one or more of the patient traces are reduced to provide processed patient traces. The processed patient traces are clustered to identify a cluster of patient traces. A process model is mined, using a processor, representing an aggregation of treatment pathways in the patient traces from the cluster. Patterns from patient traces are identified that are discriminative of patient outcomes. At least one of the patterns is represented with respect to the process model to identify treatment pathways correlated with the patient outcomes.