To reduce the manual work and turn it into automation during the data entry of any case and data processing, the biggest struggle would be the conversion of unstructured information to a machine-readable structured format.
Traditional drug safety and necessary systems, especially in the pharma and life science solutions can be improved with the help of automation technologies such as Robotic Process Automation, cognitive computing, machine learning, and NLP-based models that can change the landscape of the process entirely.
Where Can Automation Be Helpful?
Integrating a high level of automation can enhance efficiency while reducing the cost and the number of human errors as a result improving the quality and accuracy of any case processing.
Today's automation solutions can allow monitoring structured and unstructured data from different sources that can help you identify adverse events. The influential and smart NLP technology, integrating with the help of an expert can finalize, examine and remove the events by transferring manual review into automation.
Automation has been used in banking and financial industries since the early 1950s with AI incorporated. AE cases processing solutions were highly complex as they involved significantly more decision points which were restricted and examined under a highly regulated and audited environment.
AE case processing consists of four main activities connected to varied deliverables that further consist of a myriad of decision points to provide an in-depth analytical survey or inspection of the processor issue.
In order to intensify the case processing, data training was essential for machines to understand them. The process was elongated by revisiting thousands of text pages, which was extremely time-consuming. Hence Pfizer piloted a safety data extraction that was sourced from the safety database.
This pilot underwent a viability test of machine learning solutions that focused on case processing. This plot by Pfizer also tested the commercial vendors to extract case critical information from documents and databases and train machine learning algorithms that can identify the complex AE cases. The presence of four elements was extracted and specifically coded into their specific fields, which were further compared (the proposal performance) to identify, which proposal is most suitable to move to the discovery phase.
Result of Information Extraction
Pfizer's AI Center of Excellence scores were compared with three other vendors in terms of F1 scores. The highest score recorded was for reporter type and occupation and the lowest was for AE verbatim. When thoroughly tested, Vendor 1 outperformed all other vendors, inclusive of Pfizer, in all algorithms, case-level validity, and a further improved accuracy.
High-level automation in terms of adverse case processing will only enhance your efficiency in a cost-effective manner. The main areas where automation can be implemented to various areas, such as PV processes, proactive case reporting solutions, and data integrity.
Technology Solution
AI-enabled Natural language Processing (NLP) tools that avail the capabilities of machine learning and Artificial intelligence can discover and extract key elements from both structured as well as unstructured data sources. For integrating end-to-end case processing, global industries can ingrain libraries to assist their code by populating multiple fields in the database.
Ingraining libraries like MedDRA, WHODrug, MeSH can aid systems to identify and gather information for case details such as patient and reporter, adverse event/reaction, suspect or interacting drug, seriousness/non-serious, and other relatable events.
These tools can equally aid medical professionals to focus more on quality and verify the data that has been crunched and delivered by machine learning. This valuable data reduces time consumed in manual tasks of data entry as a result increasing productivity and efficiency.
These technology solutions comprises cognitive layers that create high-level automated narratives for valid case processing.
Benefits of Technology in Adverse Events
Adoption of strategically aligned automation technologies coupled with extensive change management. With the help of multi-tier architecture and combination of customizable solutions can integrate automation platforms without disrupting existing processes across safety databases.
Cost reduction by automating manual tasks.
Accelerate compliance.
Risk analysis and prioritize adverse events.
Encourage proactive decisions for product recalls thereby reducing pressure on the brand.
Reducing repetitive and mundane manual tasks by automating tasks while reducing human errors.
Improve mechanism of feedback and repeatedly enhance quality.
Seamlessly transform unstructured data into structured data for accurate and reliable decision-making.
Develop an eco-system for contextual and data reusability across myriad value chains.
To Conclude -
Integration of automation and artificial intelligence services can enhance accuracy in case processing. The degree of case completeness defines the case-level accuracy. All the more, automated case processing in the pilot can reduce the manual case processing that follows the rule-based approach for identifying a regulatory valid case. A valid case automated processing adheres to binary classification algorithms that only predicts two classes - valid and invalid that leaves no confusion and only accurate data for accurate decision-making.
No comments:
Post a Comment