Sunday, October 18, 2015

Montage and Radiology Data Analytics

I had briefly blogged about Montage mid this year about how excited I was that we have it now at UCLA.

I was fortunate to attend the Montage User Group meeting a couple of days ago and met the people who are behind this software.

Disclosure: I have no financial interest in this company. But, I did eat a couple of burgers and had some fries during the Burger Crawl after the meeting. 

What is Montage?
When I first learned about Montage 5 years ago at a conference (Montage was a vendor), the sales rep pitched it to me as a Google search for radiology reports. It is a search engine for radiology report and so much more.  Montage provides data analytics for administrative (and non-administrative people), a great structured database to query, open API for development, etc.

Why am I excited about Montage?
I do a lot of clinical research and data analysis. Montage organizes the radiology report in a structured format and makes is so much easier for me to find patients with certain conditions, etc. Prior to Montage, we would manually look up on PACS, which was incredibly inefficient as you can imagine. Montage laid the foundation to do research faster. I also use it for learning as well. For example, I can identify patients with appendicitis on MRI or a rare disease and can use the output to practice reading cases. 

I learned from the meeting that two radiologists founded the company. One of the them, Woojin, wrote the code to the first version of Montage. When I asked him if he had a computer science background, he said no.  He knew the clinical need was there and he used Google to learn how to code and went for it. How awesome was that? 

Learning points:
One of the important learning points I got from the meeting was the value and importance of measurable outcomes (vs. a non-measurable ones). Measurable outcomes would include very simple values like number of studies, types of studies, volume over time, RVU, QC metrics, etc. Of course, there are limitations to measurable outcomes. For example, one of the questions that was posed was "who is the best radiologist?". The answer was "Ask the clinicians". Clinical respect and value are often very evident at multi-disciplinary meetings like tumor boards or during clinical rounds (when teams ask for specific radiologists).  If one of the attendings that I work with tell our pancreatic surgeons that she think the lesion is benign, they will not operate or question her. This type of value is not captured in quantitative outcomes. But, until we have better ways to capture it, we have to work with what we have and Montage is certainly a good start.

The second invaluable take home message for me was the importance of rad-path correlation. At a breast imaging conference I attended, the findings of a study presented showed that the most important predictor for an effective radiologists was rad-path correlation.  Getting pathological feedback for our reads is how we get better. Variables such as our case volume and number of years of experience may be important but did not turn out to be as valuable as pathologic correlation. I think we can do more effective rad-path correlation when we get the pathology reports added to our Montage package.

The People:
Jay, the CFO of Montage asked me: "In a horse race, would you bet on the jockey or the horse?". The answer he said is the jockey because if you have a great jockey, then s/he can cajole the horse to her/his needs. In the same way, it seems to me that the team behind Montage is full of talent. Clinical needs drive the mission of the company to develop products that are valuable to many people in radiology: administrators, researchers, residents, etc. 

This team also exemplify the new model for start-up companies. The nine person team work from their respective locations (Sweden, New York, Texas, Penn, etc) at the comfort of their personal needs. The team performs the work remotely in the cloud. They harness existing resources. For example, their application uses Amazon servers (as do IBM, Netflix, among others).  The company harness newer cloud infrastructure coding, like Ansible, to improve output. 

Overall:
I'm super excited about this product and what it can do for us and I'm looking forward to what's in store in their Road Map.  I am really excited we have a medical IT director in my department who recognizes the value and have the insights to add this product to our foundational blocks. 

No comments:

Post a Comment