Qinsight™: Introduction and Demonstration 2017-08-28T17:57:02+00:00

Qinsight: Introduction and Demonstration

Transcript

Hi, I’m Jeff Saffer. And I’m Vicki Burnett. We are the founders of Quertle and the architects of first big data artificial intelligence platform for discovering hidden insights from the biomedical literature.

At some point, all of us have been frustrated by not finding the information you need. But it is more than frustration; a European Patent Office study showed that up to 30% of R&D money is wasted rediscovering existing information. And, according to the NIH, one of the leading causes for grant proposal failure is proposing work that isn’t actually new. Clearly more advanced methods are required to find value in the overwhelming tsunami of biomedical information.

Hence, we created our BioAI platform and the Qinsight discovery product. BioAI was specifically created to address unique issues encountered with the biomedical and biological sciences. It uses

  • Neural networks
  • Machine Learning
  • Factor Analysis
  • Natural Language Understanding
  • Pattern Matching
  • And more

All fine-tuned specifically for the life and medical sciences. These methods not only lead to efficient discovery of relevant results but also drive concept-based visual analytics that provide exceptional insight and support strategic decisions in ways that simply can’t be done with legacy solutions.

With its artificial intelligence underpinnings, Qinsight provides

  • Unmatched Discovery
  • Summarization of Results – as they related to your interest
  • Connections – that cannot be discovered anywhere else

leading to actionable knowledge. Importantly, Qinsight’s AI is based on the meaning of the actual document text, without having to resort to using surrogate metadata, which can often be misleading.

In our first example using Qinsight, we using a disease-centric search to look for a connection between diseases and nitric oxide using a natural language query. The artificial intelligence recognizes that we are interested in connecting specific diseases to nitric oxide and focuses the results accordingly, rather than cluttering the results with hits from the generic terms “disease”, “syndrome”, etc. In the results, we can immediately see specific diseases such as allergic airway disease, renal dysfunction, rhinitis, inflammation and cancer.

Unlike with other products, with Qinsight, you can search for nitric oxide using NO, as we are doing here, or other synonyms – and rest assured that you will find all the related terms and the different forms of nitric oxide. You can search for – and find – any synonym. This AI-aided recognition is CRITICAL since 20% of the literature about nitric oxide uses only NO and thus cannot be found by other systems that ignore NO or treat it as no (the negative) or NO (meaning number), and so on. Can you afford to overlook these potentially critical references?

Qinsight‘s artificial intelligence not only does a better job of discovering concepts, but also immediately focuses on relevant documents, so you don’t have extraneous results to weed through. Qinsight actually discovers the meaning of the documents, as they relate to your query, to provide highly precise information. Keyword searching simply cannot do that. Metadata analysis cannot do that. Qinsight‘s modern methods are applied to about 40 million documents, with full-text searching for more than 10 million of these documents. And, with Quertle’s growing list of publisher partners, the amount of full-text being searched is ever expanding. And we can include your own licensed content securely within a custom deployment.

Not only do you need relevant results, from the right content, but you also need intuitive ways to explore those results.

In the first result of this example, note that the authors indicate there may not be a connection between atopy and fractional exhaled nitric oxide. Along with several self-explanatory filters, Qinsight is the only solution that offers a unique Negative Statement filter. This allows you to exclude results where the author explicitly said nitric oxide and a disease were not connected, or to focus on such negative associations. Thus, you can understand why your search terms are connected in some instances but not others.

Qinsight uses AI to automatically identify concepts related to your query. Importantly, this is much more powerful and relevant than mindlessly showing you concepts that appear just anywhere in the results documents but are, in fact, unrelated to your query. For our current example,the first set of concepts answers to the question “what diseases are connected to nitric oxide”. You can also see the different contexts for nitric oxide. Additional concepts below cover General Concepts and the types of Actions, such as “cause” or “prevent”, that tie nitric oxide and diseases together.

Key Concepts allow you to follow up on lines of thought you were interested in, and to recognize new concepts that you didn’t even think about. Just click on any Key Concept to limit the results accordingly. The Key Concepts are essentially a summary of why the results documents are important to you. And, the Concept Cloud makes this summarization even easier.

Now, you can see the degree to which each concept relates to your query. Each concept in the Cloud is clickable so that you can immediately get to the underlying documents.

With Key Concepts, although you may be inclined to focus on the more common terms, the literature is dynamic, and less common concepts may be growing in importance. Quertle’s Concept Trends visualization uses artificial intelligence to uncover trends that are specifically related to your query.

The Concept Trends visualization provides a spiral clustering of the Key Concepts, with the size of each bubble reflecting the overall importance of that concept to your query. In addition, you’ll see that some concepts have a red border. This indicates an increasing trend for those concepts. We have hovered over one of these concepts, and see that Ovarian Cancer is growing in importance with regard to its association with nitric oxide. A blue border indicates a decreasing trend.

In this second example, we are demonstrating a gene-centric search to find the genetic contributions to melanoma. Although many genes have been identified, it is difficult to understand the critical interplay among these genes from a list of results. The Concept Connections visualization uses artificial intelligence to tease out these connections.

We are now looking at a portion of the Concept Connections. The connections are now obvious, with darker orange cells indicating a more significant connection, such as the one between matrix metalloproteinase 2 and vascular endothelial growth factor A. The intensity of the blue color along the diagonal indicates overall contribution of the gene to the results set. Importantly, this is NOT just a simple co-occurrence matrix. Rather our AI engine discovers – on-the-fly – relationships from the current results that are meaningful to your query. This provides confidence that the connections are important to your line of investigation.

We have only touched on some of the features and value of Qinsight, and want to end with a note that there is much more. In addition to the focus of the examples used in this video, Qinsight can also easily find documents and answer questions related to any biomedical or health area. And, Qinsight can fit into your workflows. Quertle’s methods can work with any textual data including healthcare and other critical sources. And, we can customize Qinsight for your needs (try that with the other “solutions”).

There is a reason we call it Q INSIGHT! Please contact us for more information. Thank you.