Call me an old fart, but I think startups like these are good to have. No offense to any of the YC founders, but I think the variety of YC companies that are operating sometimes makes me feel that the "idea bar" is quite low.
Though, I think this EDC/clinical trails data collection is a great place to innovate in. Having worked as a software engineer in a CRO before, I know how slow the development lifecycles are and sometimes even the end users are very slow to adopt. A startup can shake things up by pushing for better data collection and analysis techniques, more automation as well as better end to end phase management.
I actually wrote that knowing very well they are a YC company. What I feel is that there are some awesome ideas/challenges being worked on by YC companies and then there are not so inspiring ones. Maybe a dilution due to more available pool of money.
Somehow I felt YC should only work on really challenging problems, but its a perception that doesn't lead to profits always :).
Sencha was an easy choice to get our MVP in the hand of users. We have more critical features to work on that are orders of magnitude more important to our users.
I'm surprised to see their example visualizations.[1] They look like the defaults in Excel's charts, to me, as opposed to some of the other approaches out there.[2,3] I realize that a lot of their position is unchaining the carefully controlled medical data sources, but the presentation of that data is still a big part.
I'm a dev at Comprehend and have worked first hand on both the query engine as well as quite a few of our visualizations.
We strive to generate insightful documents, which provide value beyond default Excel charts, by providing interactivity which as you pointed out may be poorly captured in screenshots. With that said, our visualization suite consists of both graphs made by cutting edge third party charting libraries as well as in-house custom plots.
On interactivity, three of our most interesting features are drill-down, dynamic real-time highlights and filtering, and global view synchronization.
Drill-down, or click-through, refers to the ability to look into any given presented data point to see where it came from. For instance, if you are plotting the number of medications taken by site, you can right-click on any point or bar in the chart and see exactly which patients make up that data point, what the medications were, and even what symptoms those patients were experiencing. The backing data may come from different tables, different databases, or even a different type of data store all together (such as a flat file or SAS data set).
Comprehend also supports dragging a highlight or filter onto any active report or visualization. Highlights can be generated by the user on the fly; anything from "males over 60 years old" to a custom R function is fair game. Filters are functionally equivalent except instead of highlighting data which matches the predicate, we eliminate data that does not.
The product also supports global applied state, which consists of highlights and filters, which is automatically applied to all active views. This makes it easy to look at the same subset of data in different views to help answer questions and identify trends. We provide other interactive features, such as intelligent tooltips, various exporting options and view transformations, but in my opinion these are the most interesting.
It's easy to confuse Comprehend as a general BI or visualization tool. Although we provide this functionality, the hard tech problem we solve is answering questions where the data lies in disparate data sources.
This high level of interactivity and ground-up support for multiple data stores provide value beyond default Excel charting.
Thank you for this detailed and thoughtful reply (as compared with downvote without comment that I got initially from whomever).
Based on your excellent description here, it sounds much more like Tableau and various other general BI tools (SAS's Visual Analytics, as well), though as you point out, that isn't Comprehend's objective.
The view synchronization (aka brushing and linking) and real-time sorting, filtering, and querying are all powerful features and indeed help elucidate reason from big, noisy data.
In that light, I think Comprehend's 'learn more' page doesn't do any of that justice. My original comment was my honest impression when looking at the site for the first time. The light-box style screenshots on each feature encouraged me to stare at the static UI. I would humbly suggest that it could be reworked to play up all these technical feats you mention (apropos, Heroku's 'how' page or this ACM queue article: http://queue.acm.org/detail.cfm?id=1805128, wherein the examples are all interactive, such as http://hci.stanford.edu/jheer/files/zoo/ex/stats/parallel.ht...).
Though I suspect your potential user base and the buyers probably do get hands-on demos, so the website's feature page may not be terribly important. But that's my thoughts. I wish Comprehend the best of success. The work you've done already is doubtless moving mountains. And assisting health-work is a stupendous, long-term vision.
We absolutely welcome constructive criticism about both our product and how we portray it. We try to keep our screen shots and website feature listing up to date, but currently the team is working hard in preparation for our booth at DIA Philadelphia! Thanks for the well wishes.
Though, I think this EDC/clinical trails data collection is a great place to innovate in. Having worked as a software engineer in a CRO before, I know how slow the development lifecycles are and sometimes even the end users are very slow to adopt. A startup can shake things up by pushing for better data collection and analysis techniques, more automation as well as better end to end phase management.