Professor Samuel Scarpino has returned to Northeastern University after a year-long stint as Vice President of Pathogen Surveillance at the Rockefeller Foundation with a mission to solidify Northeastern’s place as a global leader in pathogen surveillance. artificial intelligence and life sciences.
Since November, Scarpino has served as director of AI and life sciences at Northeastern’s Institute for Experiential AI. Previously, he was an assistant professor at the Network Science Institute.
Scarpino recently sat down with News@Northeastern to answer questions about what makes Northeastern a dominant presence in AI. His comments have been edited for clarity and brevity.
What about Northeastern making the university a leader in AI?
The real strength of Northeastern is the high degree of importance it places on transdisciplinary collaboration and partnerships between businesses and universities.
Part of the reason Northeastern is uniquely positioned to lead in this area is the word “experiential,” which means cooperatives, but it also means humans in the loop. We can do so much more with AI and humans than with AI alone or humans alone. The idea is that the AI will detect things that we might miss. Humans can also pick up things the AI might miss.
Can you give us an example?
The UK has procedures where mammograms will be scanned by AI systems. Some of these mammograms will also be reviewed by expert panels.
A computer can read 10,000 mammograms without having to take a break. I can’t remember off the top of my head how much an expert pathologist can read before having to take a break, but that’s not much, is it?
Former UK Prime Minister Theresa May said AI could reduce cancer deaths by 10% a year.
AI is not a silver bullet, but I would definitely commit to preventing 10% of cancer deaths. For the ultimate clinical diagnosis, the clinician would have both the AI reading and potentially the input from the expert diagnostic panel.
You talked about the importance of AI in predicting disease states and the tragedy of the existence of anonymized health data that is not used for clinical purposes. How could it be better used? And what are the privacy implications?
I am not a privacy expert. Part of the reason I joined the Institute for Experiential AI at Northeastern is that it has privacy and ethics experts.
The institute’s multidisciplinary approach is a key reason I expect Northeastern to be a leader in the field of AI. You don’t want to end up with data you can’t use or in a bunch of regulatory or ethical issues.
Northeastern understood that this was an interdisciplinary problem and worked to solve it from the start.
One example of the problems with unusable data is a Centers for Disease Control and Prevention whooping cough household study that began in 2015 and spanned several years. The results of the nasal swabs have never been sequenced.
These bacteria may hold clues to the asymptomatic transmission of whooping cough. Sitting in minus 80 degree freezers in Atlanta is quite possibly the answer to one of the biggest open questions about one of the biggest killer childhood diseases on the planet. And it can never be used unless someone goes back to those 4,000 homes and gets permission to use the data.
You don’t want the data to be open and abused. You also don’t want it to be abused in the sense that the cost, effort, energy and invasion of privacy – which could include blood draws and lumbar punctures – that went into data collection means it cannot be used for anything in the future. .
Dr. Naveen Rao, senior vice president of the Rockefeller Foundation Health Initiative, said the foundation is “grateful for the passion, leadership and innovation” you have brought to its pandemic prevention initiative and looks forward to to “continue our partnership to strengthen the prevention and prevention of pandemics”. response across the global health ecosystem. How important is the global reach of Northeast campuses to your work?
A lot of what Northeastern is going to do is generate network effects, leveraging Boston, leveraging the Roux Institute in Portland, Maine, London, and the West Coast of the United States.
If I work on data for European Union citizens, I often have to store and operate this data in the EU. And by our very nature of having a large global footprint, we have the ability to work with organizations and on datasets that we couldn’t do as easily as if we were exclusively housed in Boston.
At the Rockefeller Foundation, my team focused on creating technology products that bridged the gap between data and action. For example, we built a holiday gathering risk assessment tool in 2021.
We have also done quite a bit of work around sewage monitoring, including a joint partnership with NASA and Emory Ghana to design monitoring parameters in non-drained environments.
What are the projects at the top of the list?
One of the first things I seek to do at Northeastern is help establish a research data commons, where datasets are accessible under agreed-upon rules about how the data will be used.
You would connect to a portal and according to its terms of service, you would have to accept confidentiality constraints, ethical constraints. All of these things are effectively done up front, which means that instead of spending three to six months with the lawyers negotiating what we can and can’t do, it’s pre-packaged as part of your participation. to House of Commons research data.
There are all these siloed datasets right now in the health and life sciences, in the universities. It’s an organized way to network them.
For media inquiriesplease contact media@northeastern.edu.
#Samuel #Scarpinos #passion #artificial #intelligence #Northeastern #world #leader #field