This project built an open source tool to contribute to fight the social and economic impact of Covid-19 and slow the spread of CoVID-19 and future respiratory illness epidemics by alerting people when recordings of their breathing presents crackling or wheezing. It could be used in combination with standard self-assessment questions.
The project is still running, and you are welcome to contribute or follow.
In discussions with Dr Ben Goertzel, who heads a partner organisation of Perth IoT called Decentralised Artificial Intelligence Alliance (DAIA), Tom Zorde suggested the idea for a privacy-preserving respiratory analysis app, which I called “ResAlert”. It would use artificial intelligence (AI) to identify anomalies in recordings of human breathing and alert the user. Ben confirmed feasibility and support.
For context, DAIA exists to foster privacy-preserved data sharing between AI services. Technically this is implemented through a blockchain based decentralised marketplace called SingularityNet and Ocean protocol. This will be an important technology for the long-term democratization of the value and control offered by AI and the Internet of Things.
The project has three members thus far:
• Tom Zorde leading the project and product direction; and
• Dr. Deborah Duong leading AI development
• Servet Koçak - Advisory.
These DAIA technologies are not expected to underpin ResAlert in the first stage, but Tom plans that ResAlert will be deployed there once proven and operational to ensure privacy preservation at the highest degree. DAIA conveniently ran a hackathon called CoVIDathon to build privacy-centric solutions for CoVID19 using AI & Blockchain technologies.
Check out the submission video drafted at https://devpost.com/software/resalert
The project delivered a working prototype on 1st June 2020.
ResAlert delivered a deliveed a trained algorith and front end recording functionality as :
• A way to slow spread of respiratory viruses
• A way to reduce impact on medical services
• Anonymous & private feedback to reduce anxiety and empower user
• Alerts to user when abnormalities are evident
• Support alternative self-diagnosis tools and questionnaires such as CDC Coronavirus Self-Checker
• Open source code that the world can use and improve
• Machine learning algorithm, initially trained with publicly available data, that can detect crackles and wheezing in recordings of human respiratory breathing
• A simple user interface to demonstrate operation
• Strict privacy and anonymity
This was again achieved with volunteered skills and knowledge without monetary payment (yes some people can't get their head around that. We followed our proven Perth IoT leadership method where volunteers contribute because there is a strong purpose and value towards greater good).
In this case an Expression of Interest conviened superior talent in return for :
• The opportunity to make a difference to fighting Covid-19 and future respiratory illness epidemics
• The opportunity to make a difference to future architectures for privacy-preservation and decentralised technologies
• Experience and learnings from being involved in a disruptive and unconventional run project working with extraordinary talent.