How Pharma Companies Can Scale Up Their Knowledge Discovery with Semantic Similarity Search 

Pharma has deep roots in human history with centuries of folk pharmaceutical knowledge offering a hit-and-miss range of natural remedies. But the industry as we know it today actually emerged in the second half of the 19th century when the world’s first factory for the sole production of medicines was found.

By the late 19th and early 20th century, some chemical companies had already begun using research labs to explore the medical applications for their products. Fast forward to today and the pharmaceutical sector is a global trillion-dollar industry. However, to ensure the safety and efficacy of drugs, the process of drug discovery and development is under extensive scrutiny and control on both national and global levels.

Regulatory Landscape Challenges for Pharmaceuticals

On top of having to comply with stringent and detailed regulatory requirements, Pharma companies also have to respond to a lot of questions from different regulatory agencies and do it within a short period of time. Typically, such companies have archives with a significant number of already answered questions. But the huge amount of data gathered over the years is in various formats (mostly unstructured) and stored on various systems, which makes the retrieval of the information costly, time-consuming and inefficient.

Essentially, whether a regulatory authority will view a company as trustworthy and competent will depend largely on these responses to compliance questions. Submitting an incomplete and untimely response can result in creating a negative impression, impact the company’s finances and even lead to regulatory action. Therefore pharmaceutical companies are in dire need of a system that goes beyond the conventional search technologies, which are increasingly failing to address their needs.

Scaling Up Information Extraction for Greater Value and Less Response Time

To enable pharmaceutical companies to quickly process large volumes of questions from regulatory agencies, Ontotext has developed a smart solution, which makes information extraction easier, faster and much more efficient.

First of all, this solution is able to ingest large amounts of various documents in various formats and to automatically extract and classify pairs of questions and answers. Then, the content of the questions is semantically indexed, which enables the system to compare the new questions to any previous questions stored in the database.

From this processed data a knowledge graph (KG) is created. It represents the relationships between the different elements of the document and empowers a semantic search. This type of search goes beyond the traditional keywords and is more intuitive and context-aware, which allows it to disambiguate concepts. Because of its highly interlinked nature, it can also recognize multiple references to one and the same entity.

The Power of Semantic Text Similarity

Ontotext’s solution uses one of the latest features of their signature semantic graph database GraphDB – the semantic similarity plugin. Thanks to this pugin, GraphDB’s semantic text similarity search matches words across documents that co-occur with other words in the same context. Then it returns the top 10 most similar Q&A pairs from the database. As a result, Pharma company analysts save a lot of time and effort and can easily reuse company knowledge.

As pharmaceutical manufacturers strive for a high standard of public trust, while fostering innovation and working to enhance public health, they are turning more and more to AI-based solutions like Ontotext Semantic Similarity Search in Documents. Such technologies will empower them to face up to some of the industry changes in regulation well into the future.

Although this particular solution was developed for a very specific Pharma Regulatory use case, the system’s functionality applies to all types of domains because it is based on a generic technology.

Are you facing similar problems?