The average vaccine discovery time takes a minimum 14 months, about 0.5 billion dollar investment, and over 2000 people trying hundreds of formulas to see if it kills the virus. While there are millions of people waiting for a vaccine without knowing how to help.
What it does
We are combining AI and gamification to predict which protein sequences from the new coronavirus are the best candidates for a COVID-19 vaccine. We gathered data from the virus of the SARS 2003 outbreak (also a coronavirus) and turned it into a puzzle game. By playing the game, users are describing new features of experimentally validated vaccine candidates. This information is used in an AI model to predict new vaccine candidates.
Our model is based on state-of-the-art vaccine prediction models to which we are incorporating gamification. By enriching experimentally-validated data through human-based computing (data augmentation), we are increasing the accuracy of vaccine prediction and saving time and reducing costs in pre-clinical trials. By allowing anyone to become a scientist in the process of vaccine discovery, we are helping to fulfill a need for action that is very present right now. Our platform connects researchers in need of human-based data processing with users who contribute as crowd players.
How we built it
In our prototype phase, we’ve turned a research tool (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1479323/) into a puzzle-game. Each solved puzzle is teaching the AI to find a pattern among the amino-acid chemical properties in a protein sequence. Each puzzle is generated from existing protein-based vaccines created for SARS 2003 outbreak, a virus also in the coronavirus family. Therefore, we can use similar formulas to COVID-19. The science behind AnalysisMode
Accomplishments that we're proud of
- Released a new landing page interface experience, reducing the visitor bounce rate from 73% to 30%. Therefore reducing our marketing costs by -120%. https://www.analysismode.com/
- Newly published article, on Luxembourg’s business magazine. https://www.siliconluxembourg.lu/analysismode-tool-helps-scientists-discover-the-vaccine-for-coronavirus/
- A beta game version live
- Contacted by a journalist from “O Estadao”, a major press in Brazil, with weekly 33M unique visitors.
- Initiated discussion with Pratica Capital venture capital and two angel investors in Luxembourg.
- Increased daily unique visitors by 325% from 90 to 383.
- Marketing campaign reduced Cost per User Acquisition to 0.01$ (landing page view), the campaign had 89 shares ($0.13/share).
- 81.75% of the users are now visiting the site by phone, as an effect of the UI improvement allowing the puzzle-game to played on smaller screens.
- Found +15 new supporters for the project - the names will be announced soon on the website.
- Reached a 25K gameplays from 98 countries
- Increased retention to 6,28% (average on apple and google stores is 2%)