The fast-growing biomedical literature contains a wealth of information which could be extremely helpful for better understanding drug response and choosing the best treatment given a patient’s molecular profile. However, turning this information into actionable insights requires extracting the relevant pieces of knowledge from within scientific texts and transforming them into structured data in an easily searchable repository. Having to scan millions of articles for the required information makes manual approaches futile. On the other hand, automating the process of knowledge extraction is one of the hardest problems in artificial intelligence.  

 

Consider for example the question of the efficacy of some drug for patients with a specific tumor mutation. For each drug-mutation pair we would need to scan the entire biomedical literature looking for articles containing information about the efficacy of the drug for the corresponding mutation. Then, we would need to extract from the relevant texts the specific efficacy claims, along with the data that will determine the “strength” of the claim.

 

Omnisol tackles biomedical knowledge extraction problems using a combined approach – we use sophisticated algorithms for zeroing-in on the relevant texts and identifying the relevant paragraphs within them, and then employ human experts for the task of text understanding. Omnisol’s innovative technology and rich experience in data mining of the scientific literature, together with its results-oriented approach, help to solve hard biomedical knowledge extraction problems quickly and cost-effectively.

​