I recently (20220408) participated in a Web 3.0 hackathon and won a bounty prize! For those of you unfamiliar with Web 3.0, the idea is to structure the the Web like an (modern) online videogame, where participants can compete/co-operate to acheive a common goal while chatting and interacting (e.g. with Discord). In addition, "prizes" for achievements are given as game specific currency which lets contributing partiicpants participate in governance decisions.
My entry is Below. Recording here.
Bounty: Novel ideas or use of tools for collective knowledge sharing within social communities - link
I have a rare, terminal (slow growing) cancer and it sucks for four reasons:
- Treatments: It sucks because pharmaceutical companies have zero incentive to invest in treatments for a small (rare) market.
- Data: It sucks because the data needed for “pro bono research” is locked up behind massive paywalls in university hospitals.
- Personalization: It sucks because the collection of rare disease patients is small, making standard cohort (statistical) analysis difficult.
- Reporting: It sucks because when experiments are done, results are often not reported (alltrials.net)
I’ve been working to make this situation suck less.
- Treatments: With the correct data analysis, it’s possible to repurpose/reuse/reposition existing drugs to treat rare diseases.
- Data: And by providing a freely accessible repository of rare disease data, we provide a space for pro-bono research for rare disease patients. This is cool because creating a space where supply (patient data) and demand (grist for research) meet, research is “pushed” to the most underserved areas.
- Personalization: I’ve worked with Clemson to create statistical models for personalized medicine, enabling analysis to be done on “cohorts of 1”. What is a "cohort of 1"? Suppose you take a drug and it doesn't work as advertised? That means you are different from the cohort (group of people) that the drug was tested on. That means you are unique, as are many cancer patients. So how can you use use statistics for drug validation when you only have a statistical sample of one person? Clemson's paper deals with that issue.
- Reporting: I’ve co-created Web 1.0 prototypes supporting this idea three times: in 2018, 2020 and 2022. 2018 and 2020 results showed me that I could (potentially) reuse an existing therapeutic (Bacelein and Valproic Acid) for my own condition. This research was catalyzed by opening up my genomic (DNA and RNA) data and creating a “game” where 17 teams could analyze it. In 3-day centralized versions of the game, players reported results on the third day. However in a decentralized venue, People can take the data and not report results. So the 2022 version of our game uses discord to unify the community and hosts a subset of my genomic data on GitHub under a copyleft license in order to prototype an NFT. The next step is to add a token to compensate patients for contributing data, to compensate researchers for contributing tools and to pay researchers for winning games.
Collective Knowledge: Data input for these games is currently kept on github. However, genomic data can scale to 1 TB per patient. I need a better approach. Report output is currently kept on github as well (index). This brings the output outside of discord.