Some weeks you make a lot of progress, and some, well, you don’t. Last week was one such week. While there was some progress, it was not as much as previous weeks. I’ll just give you the highlights:
I sent our materials to a Venture Capitalist (VC) around mid-week last week. I was expecting at a minimum, nothing, and of course, was hoping that we would land an intro call or some interest. To my surprise, I did receive a response. However, it was not the one that I had hoped to hear. In short, he told me that the problem of data management and transfer, in his experience, makes the go-to-market plan for an EHR extremely difficult. While this isn’t news to me, it did remind me of the larger challenges beyond UX and UI design of an app like this. I certainly don’t fault him for seeing this as a challenge and it’s one that I don’t have an answer to right now. However, I don’t believe that this means it’s impossible, just that there is a creative solution that we need to find.
While this is a setback and I’d be lying if I said it wasn’t discouraging at the moment, I enjoy hearing this because it paints a very clear picture of where the greatest room for innovation and creativity lies.
I still believe that the combination of block-based note taking and Natural Language Processing (NLP) is still part of our core differentiator. I have been doing more research and design around how that can integrate in a fairly unobtrusive way into the system. I’ve attached a screenshot below of what I believe a potential solution could look like.
The core of NLP relies on interpreting not just keywords but the intent and context of those keywords. It’s not enough for the computer to understand the word “hypertension” or the phrase “high blood pressure” the computer needs to be able to understand why it’s being used and how a clinician needs to be able to act upon that information.
Below I’ve attached a screenshot of a card that may pop up to the right of the note. The computer would highlight the phrase “high blood pressure” and show the left card in the margin of the page. The computer then gives the clinician the option to see the patient’s blood pressure history inline without navigating to another screen. A future iteration could include a suggestion to add a prescription of a blood pressure med for the patient depending on prior data.
This is why NLP is so difficult. If all we were doing were building this for hypertension, then, well, most of the work is done. However, not everything is as cut and dry. Hypertension can be seen by the data that is collected through a blood pressure cuff. Many other diagnoses are not as cut and dry.
I do believe that this is a great solution, even if it is a version 1. However, gathering the data and creating the system to power this is rather complex and difficult.
This week I’ll be working on a few things: