NSF Seed Fund Retrospective
I applied to the NSF Seed Fund SBIR Phase I for Digital Health. My project pitch for tramble.health was accepted but my full proposal was declined. I am writing this “retrospective” for me (to bring closure to this eleven-month process) and for anyone considering this program (to share my experience and some tips).
My project that I submitted is tramble.health. The premise is that when your complex chronic conditions and disabilities affect your life, the impact is visible in one or more of your digital data sources: your social media, your location history, your application usage, your spending, etc. By combining these existing personal data sources with personal health information and automating the analysis, this system can display trends, suggest correlations, and propose questions to help you better understand your health and manage your conditions.
“Should I apply to the NSF Seed Fund?”
My answer is in two parts:
(1) Assuming you are eligible and your project is clearly comparable to grants that the program director has funded in the past few years, then I think it is worth it to submit a project pitch (~5 pages, ~1,500 words).
(2) If your pitch is accepted and you are okay with a ≤ 15% acceptance rate, then expect to spend at least a month of calendar time registering with government websites and a long time writing the final proposal (~20 pages, ~7,000 words excluding citations).
What happened?
In my case, I was okay with (2) but I underweighted (1). When I read the other project summaries (while trying to write my summary), I got the first hint that my project probably wasn’t going to be a fit for this program. My project is for any chronic medical condition/disability using patient-reported outcomes, the reviewer panel listed “Lacks focus on any specific clinical condition” and “No evaluation of clinical utility / validity of the outcomes” as weaknesses.
How long did it take?
The NSF Seed Fund “apply” section has a good overview of the entire process; here’s my experience:
March 2, 2020: initial pitch submission (~1500 words)
March 4, 2020: NSF asks for clarification
March 19, 2020: clarification submitted (~500 words)
March 20, 2020: pitch accepted
April — June 2020: registering with various websites
August 2020: focus on writing before remote school starts back up
August 27, 2020: full proposal submitted ahead of September 3, 2020 deadline (~7,000 words)
January 26, 2021: notification of decline, summary and reviews made visible
February 10, 2021: retrospective (this)
The relatively fast pace of the project pitch (two days to get feedback) made me hopeful that it wouldn’t take five months to get feedback on the full proposal, but it still did. (The NSF website says four to six months.) Interestingly, four of the five reviewers submitted their reviews on November 20, 2020, while the fifth reviewer and the panel submitted on January 26, 2021. I was notified on the same day.
Resources I used
Although you can read summaries of funded/accepted projects (here, or here), there are no examples for either the project pitch or the full proposal. (At least, I couldn’t find any.) If the NSF Seed Fund could do one thing to improve the process, it would be to provide examples of project pitches and full proposals that they have accepted in the past.
I emailed a couple of other people who had their projects accepted with some specific questions around IRB/Human Subjects Review and they were helpful. (Yes, the NSF wants you to have an Institutional Review Board do a Human Subjects Review, or have the IRB exempt you from it. The impression I got from the other people is that you can start this IRB process after your full proposal is accepted.)
The NSF Youtube channel and their webinars (announced via email) were somewhat helpful. If you have a very specific question about the process, you can call in and ask. These were sometimes over two hours long so be prepared to spend a lot of time on hold. My takeaways from the webinars was (a) the budget is not that big of a deal even though many applicants stress about it and (b) the acceptance rate is ≤ 15% for full proposals.
Writing tips
Having a bunch of separate Google docs worked pretty well. I noticed the “font requirement” but initially missed the “margin requirement” and so I had to regenerate a bunch of PDFs once I saw it. So check and then update your defaults. This was the first time since college that I had to deal with citing a large number of references and the Paperpile add-on in Google Docs made this relatively easy. One minor issue I had was having to submit the references as a separate PDF in Fastlane (as opposed to with the document).
Parts of the template changed in the six-ish months between pitch acceptance and proposal submission. These were mostly minor (like the sub-area codes for Digital Health changing) but “Letters of Support” seem to have disappeared (although a reviewer referenced them so maybe I just missed them in the tool?).
Do-over?
If I could do things over, I would probably have waited longer to apply (it’s a 12-month window) so I could have more user examples. Even then, I doubt I would be accepted for the same reasons listed before: “Lacks focus on any specific condition” and “No evaluation of clinical utility / validity of the outcomes.”
What did I gain from going through this process? All the writing helped re-convince me that it’s a good idea. I had to do a lot of research into the “business case” which was a newer skill for me. And I did get some positive feedback: the pitch was accepted and there were plenty of strengths cited by the reviewers.
Negative feedback
After reading (and re-reading) the negative feedback, my conclusion is that I would have had to do a very different project to get accepted. Since I don’t want to do that project, I am okay with the rejection.
Based on the negative feedback provided, I don’t plan on making any changes to the product I’m building. (If someone thinks “Lack of focus on a specific condition” is a weakness, and I think it’s a strength, I’m not sure how we could resolve that.) I do plan on making one process change: one of the reviewers claimed that “these data formats will change frequently.” That hasn’t been my experience but it’s a potentially common objection and so I will measure this more closely going forward.
What’s next?
I’m going to keep pushing on my alpha version with the hope of having something publicly demo-able by the end of Q2 2021. You can sign up for the mailing list at https://tramble.health/.
