Overview
This application provides a Statistica workspace that allows a Statistica administrator to auto-deploy an example set of data and associated configurations that perform quality control analysis to monitor a list of process parameters representing a process design as part of stage 1 of continuous process verification. Using Spotfire Statistica Monitoring and Alerting Service (MAS), the analyses are configured for running and capturing quality control alarms using default control indices for continuous verification (stage 3 of CPV).
Users can then submit the analyses to run and view the alarms generated within MAS using a Spotfire analysis. This is augmented with additional visualizations to review, select, and acknowledge the alarms extending the demonstration of stage 3 of continuous process verification.
With the example data and analyses configured (stage 1) and the continued process verification employed (stage 3), the application then allows users to review and edit the control indices for each parameter to influence the continued process verification providing an example of stage 2 of CPV - process parameter
qualification.
Business Scenario
This application provides an example of visualizing and operationalizing data science in the form of continued process verification using Spotfire Statistica®. The example data provides a simple combination of process parameter measurement data and manufacturing/production data attributed by relation to product, site, unit operation, process parameter, and batch ID. This data is then used to deploy configurations and analyses allowing process engineers to monitor, acknowledge, and affect change in the verification of the example process parameters. It demonstrates different aspects of how to use Spotfire Statistica® to help build a journey to employ continued process verification.
Benefits and Business Value
The key benefit is to demonstrate how Spotfire Statistica® can be used to visualize and operationalize data science. Example data and analysis configurations are made available for automated deployment by a Statistica administrator. The example data is structured to represent a combination of LIMS and ERP data for use in SPC analysis and packaged via a Statistica Spreadsheet for demonstration purposes. The analytics are made available for execution, the analysis alarm results are extracted, the analysis alarms acknowledged, and the process parameters are maintained via Statistica Data Functions applied within the Spotfire® visualization. Process parameter configuration is extracted and maintained within Statistica Enterprise using version and approval controls to support ALCOA+ principles. All of this functionality is implemented within Spotfire® which has formalized procedural as well as functional requirements to ensure that its products can operate within the framework provided by 21 CFR Part 11 and the FDA’s Guidance for Industry documents.
Technical Scenario
This application, continuous process verification (CPV) Stage 2 - Process Parameter Qualification with Spotfire Statistica®, consists of various Statistica workspaces deployed as both automation analysis configurations and Statistica Data Functions for visualization within Spotfire.
Visualizations
With the automation analysis configurations deployed in Spotfire Statistica, the first step in the visualization is available in the Alarm Overview page which allows users to refresh the alarm information and manually submit the analytics to run in the Monitoring and Alarming Service. By default, the Alarm Overview visualization shows cached alarms, and will require refreshing the alarm data where no alarms are captured. Click the Refresh Alarm Information button (top left).
Next, the user may click on the Submit Analyses button (top left) to manually start the process for running the analyses, checking, and logging alarms using the Spotfire Statistica Monitoring and Alerting Service. The MAS service (which must be running) is typically configured to poll the configuration database every 20 seconds and it will take a couple of minutes to run each of the analyses for each site and unit operation configured. Click the Refresh Alarm Information to see the updated analysis run status in the Analysis Scheduling Overview (top left), and eventually, the alarm results displayed in the Alarm Details table (top right) and waterfall chart (bottom center) to see the cumulative results of the alarms across site, unit operation (default), parameter, and alarm code. Use the hierarchy slider on the bottom of the waterfall chart to see the different classifications.
The user may select (mark) an alarm, or grouping of alarms, to investigate and then navigate to the Acknowledge Alarm page. At this point the user should perform root cause analysis to determine and document the issue and its resolution. It is recommended to employ a change control or other similar documentation method to document and provide a reference document for the investigation process and results for closing the alarm(s).
The user may further refine which alarm was investigated and closed, changing the alarm selected (marked) as seen in the Marked Alarms table (top right). With the alarm successfully investigated, the user should provide the comment referencing the document used for the investigation in the alarm comment section (top left) and click the Acknowledge Marked Alarms button (top left). All of the alarms in the Marked Alarms table will be acknowledged and the alarm information will be updated.
Only unacknowledged alarms are shown in the Acknowledge Alarm page. Acknowledged alarms, and the details of the acknowledgement, can be seen back in the Alarm Details table in the Alarm Overview page.
During the alarm investigation, users may open the Process Parameter Qualification page to further review the alarms for a specific parameter. Users first select the product, site, and unit operation containing the parameter (which updates the list of associated parameters), and then select the parameter to investigate.
The parameter calculated control indices are shown along with the individuals control chart based on the range of batches selected for the parameter on the top part of the page. The control limits are calculated based on the calculated process mean +/- (3 * process sigma) and applied to the control chart for use in identifying control limit violations. The calculated process mean and process sigma is also used to evaluate runs\trend violations. Process capability indices are shown based on the specification limits and measurement data to provide additional context of the validity of the calculated control indices. All alarm results are shown for the calculated control indices. Users may change the range of batches to use when calculating the control indices (updating the calculated limits visualizations).
Below the calculated limits visualizations are similar control chart and alarm result visualizations based on "current limits". Current limits are those stored for the parameter in Statistica Enterprise (i.e. /CPV Demo/Product 1/Site 1/Unit Operation 1/Parameter Limits.sta). At first, the current limits are blank (not calculated) and therefore all of the parameter data is used matching the default calculated limits results. These current limits are those used when running and checking the alarms in the previous 2 pages, and with the calculated control limits, can be used to further the investigation of the alarms of a parameter and for comparative review.
Users can change the range of batches to use for calculating control indices (batch filter - top center), click on the Load Calculated Indices button (bottom right) to load the calculated control indices, optionally edit the control indices to set them manually, and click the Save Control Indices button (bottom right) to save and update the control indices for the process parameter (which will update the current limit visualizations once saved).
If the alarm results change to include new alarms based on the change to the control indices, new alarms will then be logged once the analyses are resubmitted to MAS (click Submit Analyses button in the Alarm Overview or Acknowledge Alarm page).
The Effective Batch in the current limits defines the place at which to transition the control index to use the new control index. If only one control index is defined, then it will be used for all batches (both before and after the Effective Batch). If more than one control index is defined, the Effective Batch defines the batch at which to change the control index. The above example screenshot shows where one control index is defined, whereas the below example demonstrates how they are applied when 3 are defined.
In this case, the control indices transitioned in the following steps:
1. first control index: LCL=93.5 and UCL=111.99 saved with Effective Batch=P1-S1-BAT75 => Applied to all batches initially.
2. second control index: LCL=94 and UCL=112.22 saved with Effective Batch=P1-S1-BAT120 => Batches P1-S1-BAT1(start) to P1-S1-BAT119 using first control index. Batches P1-S1-BAT120 to P1-S1-BAT230(end) using second control index.
3. third control index: LCL=87 and UCL=112.04 saved with Effective Batch=P1-S1-BAT175 => Batches P1-S1-BAT1(start) to P1-S1-BAT119 using first control index. Batches P1-S1-BAT120 to P1-S1-BAT174 using second control index. Batches P1-S1-BAT175 to P1-S1-BAT230(end) using third control index.
The current control limit table is structured in this manner to ensure that they are applied correctly as the control limit table is merged with the input measurement data in the deployed analysis no matter the range of batches evaluated.
This application demonstrates various aspects of implementing continuous process verification using Spotfire Statistica®.
Operations (configurations)
The following Statistica configurations are referenced assuming they are installed to the default "/CPV Demo" folder in Statistica Enterprise.
The system configurations are those centrally located and referenced for use in Spotfire Data Functions and by the automation analysis configurations for monitoring the process parameters. These are deployed using the auto-deployment Statistica Workspace (performed by a Statistica Administrator during initial install).
System Configurations
-
/CPV Demo/System/Data/LIMS Data.sta (Statistica spreadsheet)
-
CPV Demo Data (Statistica database connection)
-
Statistica database connection that references the above "/CPV Demo/System/Data" folder in Statistica Enterprise to use the Statistica Spreadsheet "LIMS Data.sta" as a database table.
-
-
/CPV Demo/System/Data/LIMS Data (Statistica data configuration)
-
/CPV Demo/System/Data/LIMS Data (Statistica ad hoc analysis configuration)
-
/CPV Demo/System/Data/First Batch (Statistica data configuration)
-
/CPV Demo/System/Data/First Batch (Statistica ad hoc analysis configuration)
-
/CPV Demo/System/Data/Parameter Options.sta (Statistica spreadsheet)
-
Statistica Spreadsheet document containing the analysis options to use when running the quality control analysis for each of the process parameters and deployed into Statistica Enterprise as a binary document object.
-
-
/CPV Demo/System/Data/System Options.sta (Statistica spreadsheet)
-
Statistica Spreadsheet document containing the system options to use when running the analysis and deployed into Statistica Enterprise as a binary document object.
-
-
/CPV Demo/System/Calc New Limits Analysis.sdm (Statistica Workspace)
-
Analysis workflow used by the /CPV Demo/System/Data Functions/Calc New Limits Analysis.sdm data function to perform the quality control analysis for calculating the new control indices.
-
-
/CPV Demo/System/Current SPC Analysis.sdm (Statistica Workspace)
-
Analysis workflow used by the /CPV Demo/System/Data Functions/Current Analysis.sdm data function to perform the quality control analysis for current control indices.
-
-
/CPV Demo/System/SPC Analysis.sdm (Statistica Workspace)
-
Analysis workflow used by the automation analysis configurations to perform the quality control analysis using current control indices.
-
-
/CPV Demo/System/Save Param Info.sdm (Statistica Workspace)
-
Analysis workflow used by the automation analysis configurations to save updated parameter batch index information.
-
-
/CPV Demo/System/Start Monitoring.sdm (Statistica Workspace)
-
Analysis workflow template used during the auto-deployment to perform the automation analysis workflow for each product, site, and unit operation.
-
Automation Configurations
The automation configurations are deployed for each product, site, and unit operation (i.e. "/CPV Demo/Product 1/Site 1/Unit Operation 1"). They each have their own Statistica Taskset to define what alarms to capture and the scheduling to automate their running (default configured to run on-demand). These are deployed using the auto-deployment Statistica Workspace (performed by a Statistica Administrator during initial install).
-
/CPV Demo/Product 1/Site 1/Unit Operation 1/Start Monitoring (Statistica Workspace Analysis Configuration)
-
Analysis workflow copied from the template in the system configurations and updated for the defined product, site, and unit operation. Steps performed in the automation workflow are as follows:
1. Extract the parameter batch index information (i.e. /CPV Demo/Product 1/Site 1/Unit Operation 1/Parameters.sta).
2. Extract the current parameter limit information (i.e. /CPV Demo/Product 1/Site 1/Unit Operation 1/Parameter Limits.sta).
3. Extract the system options (/CPV Demo/System/Data/System Options.sta).
4. Extract the parameter analysis options (i.e. /CPV Demo/Product 1/Site 1/Unit Operation 1/Parameter Options.sta).
5. Extract the first batch for the given product, site, and unit operation using the "/CPV Demo/System/Data/First Batch" data and ad hoc analysis configurations.
6. Extract the LIMS data for the given product, site, and unit operation and filter the batches using the parameter batch index information, and if the parameter batch index information has not been initialized (blank), use the first batch information (all available batches).
7. Merge the LIMS data with the current parameter limit information.
8. Prepare the analysis options and run options for running the "/CPV Demo/System/SPC Analysis.sdm" analysis workflow.
9. Run the "/CPV Demo/System/SPC Analysis.sdm" analysis workflow to perform the quality control analysis retrieving the list of alarms generated, and based on the "Historical rows to keep" option in the system options, the starting batch to use for the next run (to optimize the automation workflow using batch delta index - evaluate the last N batches only).
10. If there were no errors performing the SPC analysis, run the "/CPV Demo/System/Save Param Info.sdm' analysis workflow to update the parameter batch index information.
11. Log the alarms generated for use in the Monitoring and Alerting Service (please note, alarms will only be logged when run under the context of a Statistica Taskset).
-
-
/CPV Demo/Product 1/Site 1/Unit Operation 1/Product 1-Site 1-Unit Operation 1 (Statistica Taskset)
-
Statistica Taskset to define which analysis to run (/CPV Demo/Product 1/Site 1/Unit Operation 1/Start Monitoring), which alarms to capture for logging (control violations, runs violations from individuals chart, and specification violations), and the schedule to run the analysis (on-demand).
-
-
/CPV Demo/Product 1/Site 1/Unit Operation 1/Parameter Limits.sta (Statistica spreadsheet)
-
Statistica Spreadsheet document containing the current parameter limits and deployed into Statistica Enterprise as a binary document object.
-
-
/CPV Demo/Product 1/Site 1/Unit Operation 1/Parameter Options.sta (Statistica spreadsheet)
-
Statistica Spreadsheet document containing the quality control analysis options to use for the process parameters and deployed into Statistica Enterprise as a binary document object.
-
-
/CPV Demo/Product 1/Site 1/Unit Operation 1/Parameters.sta (Statistica spreadsheet)
-
Statistica Spreadsheet document containing the batch index information for the process parameters and deployed into Statistica Enterprise as a binary document object.
-
-
/CPV Demo/Product 1/Product 1 (Statistica Dashboard)
-
Statistica Dashboard providing a view to the alarms occurring across sites and unit operations for a given product.
-
-
Data Functions
These are deployed using the auto-deployment Statistica Workspace (performed by a Statistica Administrator during initial install).
-
/CPV Demo/System/Data Functions/Acknowledge Alarms.sdm (Statistica Workspace)
-
/CPV Demo/System/Data Functions/Calc New Limits Analysis.sdm (Statistica Workspace)
-
Analysis workflow used in Spotfire as a Data Function to perform the quality control analysis for calculating the new control indices. Includes conditional run (using the Run Workspaces node) to only generate results when valid input data is provided.
-
-
/CPV Demo/System/Data Functions/Current Analysis.sdm (Statistica Workspace)
-
Analysis workflow used in Spotfire as a Data Function to perform the quality control analysis using the current limits. Include conditional run (using the Run Workspaces node) to only generate results when valid input data is provided.
-
-
/CPV Demo/System/Data Functions/Get Alarms for Tasksets.sdm (Statistica Workspace)
-
Analysis workflow used in Spotfire as a Data Function to retrieve the alarms for the configured Statistica Tasksets (how the analyses are scheduled to run) and their current run status.
-
-
/CPV Demo/System/Data Functions/Get Parameter Limits.sdm (Statistica Workspace)
-
/CPV Demo/System/Data Functions/Initialize Data.sdm (Statistica Workspace)
-
Analysis workflow used in Spotfire as a Data Function to retrieve the "LIMS Data.sta", parameter options, and list of deployed Statistica Tasksets.
-
-
/CPV Demo/System/Data Functions/Run Tasksets.sdm (Statistica Workspace)
-
Analysis workflow used in Spotfire as a Data Function to run the deployed Statistica Tasksets.
-
-
/CPV Demo/System/Data Functions/Update Parameter Limits.sdm (Statistica Workspace)
Installation
An installation guide is provided as part of the release on this Exchange page.
Recommended Comments
There are no comments to display.