There is a lot of excitement at the crossroads of artificial intelligence and healthcare. AI has already been used to improve the treatment and detection of diseases, the discovery of promised new drugs and the identification of genetic and disease links, and much more.
By analyzing large databases and finding patterns, almost any new algorithm has the potential to help patients – AI researchers need to obtain accurate data to train and test those algorithms. Hospital patients are reluctant to share information with research groups about sensitive patients as they may understand. When they exchange data, it is difficult for researchers to make sure that they use only the data they need and that they delete it when it is finished.
Secure AI Labs (SAIL) provides solutions to these problems and allows the implementation of AI algorithms on encrypted databases that are never removed from the database system. Healthcare organizations can control how they use their databases, and researchers can protect the confidentiality of their forms and search queries. Neither party needs to see data or format to cooperate.
SAIL’s platform can gather data from multiple sources, providing a rich understanding of fueling for a more effective algorithm.
“You do not have to worry about hospital executives for five years before implementing your machine learning algorithm,” says MIT professor Manolis Kellis, co-founder of Nen Kim ’16, CEO of the company. 17 “Our goal is to help patients, machine learning scientists and develop new therapies. We want to apply new algorithms – the best algorithms for the largest data processing. ”
SAIL has already partnered with hospitals and life sciences companies to open anonymous data for researchers. Next year, the company hopes to operate about half of the top 50 medical centers in the country.
Release the full capabilities of AI
As a graduate student at MIT studying computer science and molecular biology, Kim worked with researchers at the Computer Science and Artificial Intelligence Laboratory (CSAIL) to analyze clinical trials, genetic association studies, hospital intensive care units, and more.
“Whether it’s hospitals using hard disks, the old file transfer protocol or sending mail, I realized there was something seriously wrong with data transfer,” Kim said. “It was all not well observed.”
Kellys, who is also a member of MIT & Harvard Broad, has spent many years collaborating with hospitals and associations across a number of ailments, including cancer, heart disease, schizophrenia and obesity. He knew that small research teams would struggle to gain access to the data that his laboratory worked on.
In 2017, Kellis and Kim decided to commercialize the technology they were developing to allow AI algorithms to be implemented on encrypted data.
In the summer of 2018, Kim participated in the Delta v Initial Accelerator, activated by the Martin Trust Center for MIT Entrepreneurship. The founders were also supported by the Sandbox Innovation Fund and the Venture Consulting Service, who made various initial connections through their MIT network.
To participate in SAIL’s program, hospitals and other healthcare organizations can attach a node to the back of their firewall so that researchers can retrieve part of their data. SAIL then sends the encrypted algorithm to clients who live in datasets in a process called federal learning. The algorithm transmits the data locally on each server and transmits the results to a revamped central format. No one – researchers, data owners or even sales – has access to templates or datasets.
This approach allows a wider group of researchers to apply their models to larger datasets. Kellis’ lab at MIT has launched competitions to further involve the research community, providing access to datasets in areas such as protein activation and gene expression, and challenging researchers to predict results.
“We invite machine learning researchers to come in last year’s data and make predictions about this year’s data,” Kellys said. If we see that there is a new kind of algorithm that works well in these community level assessments, people can build it locally in different institutions and level the playing field. So what matters is the quality of your algorithm rather than the power of your connection. ”
By allowing a large number of datasets to be anonymized to the overall understanding, SAIL’s technology allows researchers to study rare diseases, where small databases of relevant patients are frequently distributed across many institutions. This has made AI historically difficult to apply to data AI formats.
“We hope that all of these datasets will eventually be open,” Kellys said. “We can cut all the silos and activate a new era together with a single keystroke to analyze data for every patient with rare disabilities around the world.”
Activation of future medicine
Cell is increasingly seeking to partner with patient associations and healthcare groups, including an international healthcare consulting firm and the Kidney Cancer Association, to work with a large amount of data on specific diseases. The partnership coincides with SAIL patients, the team they seek to help the most.
Overall, the founders are pleased to find that SAIL solves problems encountered in its laboratories for researchers around the world.
“The best place to solve this is not an academic project. The best place to solve this is in the industry, where I can provide a platform not only for my lab but for any researcher, ”says Kellis. It is about creating an ecosystem consisting of academia, researchers, pharmacy, biotechnology and hospital partners. I think that vision of the future of medicine will become a reality when all these aspects are mixed.