|
Data Quality
Data Quality Management
Data Quality in a system that tracks the exposure of people to ionising radiation is vital. Data Integrity and Data Quality directly affect the value of the data and its suitability and use for occupational exposure monitoring and reporting. Data Quality and reliability also impact analysis and research outcomes and reputations.
Predictive analytics, profiling, trending and forecasts (such as detecting when a wearer is likely to go past their exposure budget for a given time frame, for example) require a solid starting base, from which data is clean (of high quality) and can be trusted.
Historion introduces an extra level of rigour not commonly included in modern record keeping systems, by including its own integrated Data Quality Management (DQM) tools. Historion’s DQM tools are present in fundamental data importing and processing areas of the software. This means that Historion is not reliant on but can supplement any external IT Department DQM strategies as needed.
Historion DQM tools interrogate data rows as they are imported into the software and can also be used after any data import events. Historion’s inbuilt DQM utilities scale, continuously covering growing data content over time.
Though not directly responsible for generating all the data in Historion, the RSO uses and reports the data contained in the software. In this way the RSO represents the data and relies on it. Equally the RSO can be associated with good quality data and not unfairly associated with poor quality data.
It is important to emphasize that any reference to Reading Data Quality in Historion is in no way representing or referring to Provider laboratory processes and mechanical methods of deriving Doses. Historion only works with data after Providers report it and quality issues within data RSOs receive.
Historion’s DQM tools provide options for the RSO to assess, vet, clean and rely upon data as the quality of the data is continually made visually apparent. The extent to which Data Quality control options are used and the strictness to which the RSO applies the DQM tools is user determined. The Historion DQM tools can be applied rigorously (recommended) or ignored (not recommended).
The DQM Tools present in Historion include;
Dose Readings Gaps Detection
Dose Readings Duplication Detection
Dose Readings Overlap Detection
Wearer Duplication Detection
Wearer Merge/Split Utility
To the extent that it can the software implements automatic Data Quality controls and mechanisms;
Inbuilt page Data Validation and Checking Mechanisms
SQL Server Relational Table integrity controls and rules
Mandatory data fields and data type controls and rules
Causes of Poor Data Quality
It is worth noting the five most common causes of poor Data Quality and inherent risks associated which are applicable to where dosimetry data is generated (the provider) and used (the customer);
1. Data entry by people: the leading cause of poor Data Quality is human error in its many forms.
2. Data migration and conversion projects: migrating legacy data to new systems carries an inherent risk to data quality with the potential for data problems that can remain hidden when systems are new.
3. Mixed entries by multiple users: instructions on what data should be recorded are open to human interpretation and assumptions. People can incorrectly populate content without realizing their interpretation is different to the original intention for fields involved or their understanding and assumptions surrounding content may be different to what is assumed and understood by other users.
4. Changes to source systems: application users are responsible for the consistency of the data they enter and maintain across application and configuration changes. As third parties make changes to how an application functions, they can inadvertently harm the integrity of the data it uses.
5. IT System Errors: As applications grow in complexity and are distributed across more computers, data corruption becomes more common and harder to isolate and fix.
Risks Associated with Poor Data Quality
Poor quality data can lead to customer dissatisfaction, increased operational costs, less effective decision-making, and a reduced ability to make and execute strategy as well as;
Inaccurate mathematical outcomes, such as understated or overstated exposure aggregates
Partial collation of doses due to incorrectly split records, in the case of wearer data
Reduced confidence in data outcomes and reports where poor data quality is apparent
Inappropriate or disproportionate actions in response to understated or overstated outcomes
Incorrect research conclusions based on unreliable or missing data
Wasted time and effort in resolving data quality problems and following up on causes
Legal or financial damage due to incorrect reporting, advice or results publication
Reputation damage
Data Quality Futures
Where Historion fields have a “camel” or “title” case expectation optional Title Case (TCase), automatic Upper Case (UCase) and Lower Case (LCase) auto transformation options will be added.
We will create an automated data quality management (DQM) report. Triggered after import and/or set to be a cyclic DQM report. This can be set to send only if new potential DQM issues are detected after import. Historion support is a suggested recipient of this report as well for collection of data quality and system size statistics where our clients agree with this data collection.
Based on the traffic light system and data quality counters we will create a data quality “score”.
|