The pharmaceutical sector has grown significantly in recent years as a result of digitization and greater research, and it is expected to be worth $1.5 trillion by 2023 according to estimates. The Clinical Pharmacology and Therapeutics Journal has published an research about the use of AI/ML techniques in drug development, regulatory submissions, review, and research.
Pharmacovigilance is crucial for post-marketing safety and effectiveness investigations since it assures the safety profile of biological products throughout medication development. AI/ML is being researched to increase the efficiency and efficacy of the pharmacovigilance process, with multiple areas of application identified. These technologies are helping to automate labor-intensive and repetitive tasks and improve the accuracy and speed of drug safety monitoring.
In this article, we will explore the top five areas where AI/ML technologies have been utilized to improve the entire pharmacovigilance process.
1. OCR & NER Augmented Case Intake
OCR (Optical Character Recognition) and NER (Named Entity Recognition) are two AI techniques that can be applied in pharmacovigilance to automate the case intake process. OCR can extract text from images and scanned documents, such as adverse event reports submitted in paper format.
NER can identify and classify relevant entities, such as drug names, symptoms, and patient details. The combination of OCR and NER can efficiently extract data from multiple data sources containing primarily unstructured data, like medical literature, social media, Electronic Health Records (EHRs), etc.
The process involves converting the scanned document or image into a digital format, using OCR to recognize the text, and NER to extract the named entities, which can be stored in a structured format for further analysis and reporting.
2. ML & NLP for Cognitive Case Processing
Machine learning based cognitive case processing is an automated method of analyzing and categorizing adverse event reports in pharmacovigilance using ML algorithms. The technique involves training an ML model on a large dataset of labelled reports to identify patterns and relationships that correspond to different categories. This model can then be used to classify new reports accurately and consistently, increasing the efficiency of case processing.
NLP techniques can also be used to extract entities from unstructured narratives to further improve the accuracy and speed of the process.
3. NLG-Aided Narrative Writing
Narrative writing is crucial in pharmacovigilance as it provides a detailed account of adverse events and helps in understanding and identifying patterns. However, challenges like inconsistency, resourcing, timeliness, and using multiple data sources make narrative writing difficult. Moreover, follow-up reports may result in fragmented and confusing stories.
To address these challenges, Natural Language Generation (NLG) can be used to automate narrative writing. It ensures consistency and high-quality narratives while reducing the time needed to create them. NLG can quickly extract necessary information, produce narratives and place them in required templates. Audit trails and version control ensure easy access and comparison of different versions of the narrative.
4. RPA-Based Case Workflow
Robotic Process Automation (RPA) is a new technology that uses software bots or artificial intelligence to automate business processes. Workflow automation is the process of optimising sequences of processes through the use of technology and modern computing.
Pharmaceutical companies are interested in implementing automation to improve efficiency, reduce costs, and shorten delivery timelines. RPA can handle repetitive tasks and improve productivity, compliance, and overall quality, reducing the need for human intervention.
For pharmacovigilance case workflows, rule-based automation, RPA, or regular expression-based automation can significantly improve compliance. A study in the Applied Clinical Informatics Journal highlights the potential of workflow automation in solving quality, safety, and efficiency concerns.
5. ML-Assisted Medical Coding
Medical coding involves categorizing similar terms using a validated medical dictionary to obtain a count of all related terms. Machine learning can assist in matching reported terms to dictionary terms, automating the process of coding and reducing the risk of errors. This automation can reduce the effort and cost associated with manual coding.
AI/ML technologies have the potential to transform pharmacovigilance by automating data collection and reporting, enhancing product quality, optimizing treatment regimens, reducing costs, and improving patient safety. The implementation of AI/ML in pharmacovigilance can help pharmaceutical companies present appealing alternatives to traditional workflows, and ultimately improve overall drug safety. However, to ensure the effectiveness of AI/ML models, they must be trained on high-quality and well-annotated data and evaluated on separate datasets.
It is time for the pharmaceutical organizations to rapidly adapt AI/ML use cases and solutions for enhancing pharmacovigilance. DF mSafety AI is one-of-a-kind multi-vigilance solution catering to pharmacovigilance for drugs and other verticals like cosmetics, medical devices, nutraceuticals and vaccines that uses the power of AI/ML and automation to deliver efficiencies and a great user experience in Safety Case Management.
Click here to read more about how the important areas like Literature Monitoring, Seriousness Prediction, Causality Assessment and Signal Detection & Monitoring can be enhanced using AI/ML technologies and the challenges associated with these technologies in our article published in the Pharma Focus America Digital Magazine, Issue 1, 2023.