THANK YOU FOR SUBSCRIBING
Pharma Tech Outlook | Monday, July 04, 2022
the six stages of pharmacovigilance case automation outline how implementing intelligent automation such as AI and ML could benefit pharmaceutical companies.
FREMONT, CA: To obtain information regarding adverse events (AEs) and manage patient safety, pharmacovigilance (PV) is essential. The AE case processing segment, a crucial component of PV, now confronts many difficulties. Adverse event reporting is costly, time-consuming, and subject to human error, which could have a negative influence on patient safety. PV case processing is progressively being automated, although it's not always clear what that entails or where it starts. This article compares the pursuit of self-driving automobiles by the automotive sector to that of other regulated industries with considerable safety concerns and marketing expenditures. When reinterpreted for pharmacovigilance, the published standards for motor vehicle automation at successive levels of the automotive industry offer a useful framework.
Stay ahead of the industry with exclusive feature stories on the top companies, expert insights and the latest news delivered straight to your inbox. Subscribe today.
Stage zero, or manual case processing, excludes any automation support. The whole case intake and processing procedure are purely manual, and the case submission process and any decision-making that results from it are solely handled by human users. Even if warning or intervention technologies improve this performance, the process execution is still entirely manual. This may be all that is ever required for organisations with modest volumes of adverse occurrences to balance compliance and cost, but there are numerous tested digital technologies to automate process steps and advance to the next stage.
The only support is provided in the first level of autonomy, and human user monitoring is still a crucial element. If humans apply this to the creation of autonomous or self-driving cars, stage one would be an interaction between a human driver and the system. Stage one, for instance, would enable a self-driving car to do tasks like steering or controlling the car's acceleration and deceleration (eg, blind-spot alerts). Various degrees of this level of automation have been present in PV for many years. Auto-narratives and letter creation are good examples where proven systems transform structured case data by manually picking from pre-defined templates tailored to match company-specific characteristics rather than investing time and money manually creating these free-text descriptions.
Human users continue to monitor case processing at this stage of automation and take appropriate action when necessary, although fewer instances need manual involvement. In our image of the autonomous vehicle, automation enables the system to take over multiple tasks at once, such as steering and acceleration/deceleration. For instance, automating data extraction and identification from source documents speeds up case processing; doing this work in advance improves the ability to identify follow-ups and reduces duplication. Another situation where AI can do the work-intensive tasks is with bulk literature screening. Smart algorithms can digest more stuff in less time with more accuracy than PV specialists could by reviewing peer-reviewed literature and reports for AE signs for hundreds of hours. Although there will be false positives requiring inspection, it is interesting to note that an automated PV system utilising natural language processing (NLP) is more likely to find arbitrary mentions of ailments or products than a human reader. When to follow up, confirming data, and initiating queries for missing information can all be improved by training an AI/ML system. Alternatively, this can be done algorithmically, as it is when related to null taste values.
The life sciences sector gains from autonomous case processing in several ways. By automating labour-intensive and expensive tasks, businesses can save time and money by turning the conventional case processing workflow on its head. Companies must first assess where they stand currently to create specific PV automation targets that are appropriate for their level of risk tolerance.
More in News