Hypergraph Process Analysis for Multi-Stage Variation Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems fail to adequately stabilize multi-stage processes in industries like biologics, pharmaceuticals, and mechanical devices due to dominant sources of variation, leading to inefficiencies and increased costs, as they lack effective data aggregation and contextualization capabilities for real-time analytics and feedback.
Innovation Solution
The implementation of a hypergraph data store and run data store system that maintains process versions with parameterized resource inputs and outputs, allowing for data analysis and correlation across disparate systems, enabling visualization of quality issues and trends, and facilitating process stabilization through a statistics module and process evaluation module.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional systems store files and data without structure, then data storage is simple, but real-time analytics and feedback capabilities are inadequate
Solution Approach 1:
The system segments data into structured components including process runs, parameters, inputs, outputs, and metadata. Each process run is divided into discrete parameter measurements with associated metadata, enabling granular analysis while maintaining overall system organization. This segmentation transforms unstructured data storage into a hierarchical structure that supports both simplicity and analytical capability.
Solution Approach 2:
The patent introduces an intermediary data structure layer between raw data storage and analytical processing. This intermediary structure includes standardized fields for process runs, parameters, and metadata that act as a bridge, allowing conventional storage systems to provide real-time analytics capabilities without requiring complete system redesign.
2Adaptability or versatility
If multi-stage processes use many functional components, then process functionality is enhanced, but data aggregation and reproducibility assessment become rate-limiting
Solution Approach 1:
The system implements a universal data structure that can accommodate multiple functional components across different process stages. The standardized parameter and metadata framework serves all functional components uniformly, enabling data aggregation from diverse sources without requiring component-specific processing. This universality allows the system to handle complex multi-stage processes while maintaining efficient data aggregation through a single standardized interface.
3Reliability
If processes are stabilized by identifying variation sources, then process reliability improves, but resources and time for analysis increase
Solution Approach 1:
The system performs preliminary organization of data into structured process runs with predefined parameter categories and metadata fields before analysis begins. This preliminary structuring of data during the measurement phase eliminates the need for time-consuming data preparation and cleaning during analysis. By establishing the analytical framework in advance through standardized data collection, the system accelerates the identification of variation sources while maintaining high process stability assessment capability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods for process design and analysis of processes that result in products or analytical information are provided. A hypergraph data store is maintained and comprises versions of each process. A version comprises a hypergraph with nodes, for stages of the process, and edges. Stages have parameterized resource inputs associated with stage input properties, and input specification limits. Stages have resource outputs with output properties and output specification limits. Edges link the outputs of nodes to the inputs of other nodes. A run data store is maintained with a plurality of process runs, each run identifying a process version, values for the inputs of nodes in the corresponding hypergraph, their input properties, resource outputs of the nodes, and obtained values of output properties of the resource outputs. When a query identifies one or more inputs and/or outputs present in the run data store, they are formatted for analysis.