The Current Status
Statistical analysis output validation takes time and effort but does not guarantee a high-quality deliverable. Errors are still likely to go uncorrected. When submitting a study to the regulator, the expectation is that the output validation tasks -which are lengthy and expensive timewise – will be done with impeccable accuracy.
The validation work is carried out by the pharmaceutical companies internally or outsourced to a Clinical Research Organization (CRO), or in some cases performed by both organizations.
The FDA reported that in 2020 (during the DIA conference), 21% of all submissions contained inconsistencies and errors. That’s about 1 out of 5 submissions containing errors.
Each submission contains hundreds, even thousands of tables, and the SAS output validation process is nearly always completely manual. This complex process, typically executed by the SAS programmers and biostatisticians, is ill-defined, very time-consuming, and error-prone.
In Beaconcure’s case, automating the quality control validation checks does not require pre -standardization of processes. This means that the input of data can be heterogeneous, and the machine will automatically arrange the content as standardized output.
Automating manual processes has multiple benefits. Computers can work twenty-four hours a day, seven days a week, if they are connected to energy and Internet sources.
- Algorithms perform tasks with absolute accuracy, exactly as coded to do.
- The delivery of results is predicted and can be planned in advance (timewise).
- Accessibility to automated data is far better and is easier to understand and manipulate further.
- Size matters less. Large quantities of data can be processed with high efficiency and consistently
- Consequently, automated processes take less time than those performed manually.
Automation of QC processes, for the sake of data validation, means one thing at the end of the day: accuracy. If a machine is to replace a human, it must perform better and faster. Performing better means the machine must pass the mark of human data accuracy in the same time or less. It must do so with absolute superiority in its methodology to reflect reliability and prove to be effective.
Automation enables work to be executed faster and more efficiently. Data can be easily reached, queried for purpose, and analyzed. The outputs of a large quantity of data are substantially improved, making the final product better and the delivery faster.
Delivering studies for regulatory approval faster or answering regulatory questions without compromising the accuracy.
Whether on the critical path or not, validating data automatically can provide shorter timelines and cycles of review that promote early delivery to the regulator. This process happens to seek earlier approval.
The Human Aspect
As a direct result of reducing the necessary time for a task, it saves resources and reduces resource fatigue. Employees can spend less time on those projects, giving them the time to focus on high-level tasks that need a human presence. Other than that, there’s also more space for employee development toward completing one’s more scientific tasks that cannot be automated.
Moreover, the evolution of the study can be easily analyzed and tracked. It offers both the management and regulator team transparency into the data validation and authoring processes. In return, it can provide great reliability.
Looking at the environmental impact, automation reduces the carbon footprint as everything is done digitally and paperless.
The benefits of using automation to replace manual QC checks include resource savings, transparency, better effective communication between task owners, better access to historical and current data while reducing overall risks of errors and inconsistencies.