A Targeted Approach

Simply put, our remarkable platform takes unstructured text (like, for example, physicians’ notes) and creates meaningful, actionable and structured information.  “Natural Language Processing” is a big buzzword these days; our system is an order of magnitude beyond even the most advanced NLP methods. The fact is, while NLP is useful for some basic tasks, it falls far short of what is actually required in a medical or scientific context.

It’s about the relationships.

For example, a lot of medical data is really about relationships, especially with current gene therapy and other state-of-the-art techniques.  That’s where advanced “graph” data techniques and systems come in, discovering the relationships, for example, between biomarkers, the organ, and a disease. Our graph technology is recognized as the best in the industry.  Coupled with other techniques of machine learning, it’s an unbeatable combination.

We’re bendy. 

And the cool thing about this is its flexibility.  With our system, ontologies can be changed on the fly to handle whatever it’s ingesting. We can even work without ontologies. Competitive offerings are hard-coded and require expensive and time-consuming development.

The result is that we can feed our system raw data and have it come out understandable and useable on the other side, without a lot of human intervention.

A Use Case: Multi-Label Classification.

Here’s one cool thing we can do: Take a set of documents (path report, lab notes, patient notes, imaging, whatever) without any labels whatsoever, then classify those documents into one or more sets.  The classifications could be ICD-10 or CPT codes or even patient co-morbidities.

The fallacy of big data.

The term itself suggests nearly unlimited data. In practice, there’s usually not enough data. This comes into play when training the system. A unique aspect of  MST is that we handle imbalanced data sets. Many competitive systems, while looking good on paper, fall short when it comes to tackling real-world imbalanced data.

The bottom line:

We build technology that empowers healthcare organizations to make use of the vast volumes of unstructured data present in electronic health records, medical literature, patient surveys, and more.  Our technology automates processes such as medical coding, provides input for clinical decision support, and clinical research.