There’s a lot of excitement at the intersection of artificial intelligence and healthcare. AI has been used to improve advances in healthcare, disease detection and treatment, discover promising new drugs, identify links between genetic diseases, and more.
By analyzing large data sets and finding patterns, virtually any new algorithm has the potential to help patients: AI researchers only need to access the right data to train and test those algorithms. When sharing data, it’s difficult to double-check that researchers are using only the data they need and delete them when they’re done.
Secure AI Labs (SAIL) addresses these issues with technology that enables AI algorithms to run on encrypted data sets that never leave the data owner’s system. Health organizations can control how their data is used while researchers can protect the confidentiality of their data. Neither party needs to see the data or the model to work together.
The SAIL platform can also combine data from multiple sources to create valuable information that drives more effective algorithms. SAIL co-founder and MIT professor Manolis Kellis suggested that one shouldn’t have to chat with hospital managers for five years before they can run machine learning algorithm. The goal is to help patients, help machine learning scientists, and develop new therapies. We want the new algorithms, the best algorithms, to be applied to the largest possible dataset.
SAIL has already partnered with hospitals and life science companies to share anonymous data with researchers. Over the next year, the company expects to partner with roughly half of the top 50 academic medical centers in the country.
Unlock the full potential of AI
As a student at MIT, studying computer science and molecular biology, Kim worked with researchers at the Computer Science and Artificial Intelligence Laboratory (CSAIL) to analyze data from clinical trials, gene association studies, hospital intensive care units, and more.
Kim realized that something was seriously broken when exchanging data, be it in hospitals with hard drives, old file transfer protocols or even when sending it by post. Everything wasn’t well followed, as suggested by Kim.
Kellis, who is also a member of the Broad Institute of MIT and Harvard, has spent years partnering with hospitals and consortia on a variety of diseases including cancer, heart disease, schizophrenia, and obesity. He is struggling to access the same data that his laboratory was working with.
In 2017, Kellis and Kim decided to commercialize the technology they had developed to allow AI algorithms to run on encrypted data. In the summer of 2018, Kim participated in the Delta V Startup Accelerator of the Martin Trust Center for MIT Entrepreneurship. The founders were also supported by the Sandbox Innovation Fund and the Venture Mentoring Service and made several initial connections through their MIT Network.
To participate in the SAIL program, hospitals and other health organizations make some of their data available to researchers by configuring a node behind their firewall. SAIL then sends encrypted algorithms to the servers where the records reside in a process called federated learning. The algorithms process the data locally on each server and transmit the results to a central model that updates itself. Nobody, not the researchers, not the data owners, not even SAIL, has access to the models or data sets.
The approach allows a much broader group of researchers to apply their models to large data sets. To further engage the research community, Kellis’ lab at MIT has started running competitions to provide access to datasets in areas such as protein function and gene expression and to encourage researchers to predict outcomes.
Kellis invited machine learning researchers to train on last year’s data and predict this year’s data. When we see that there is a new algorithm that works better on these community-level assessments, people can adopt it locally in many different institutions and level the playing field. The only thing that matters is the quality of your algorithm, not the performance of your connections. Groups of relevant patient data are often spread across many institutions.
Applying AI models, Kellis hoped that all of these data sets will be open at some point. They can break all silos and enable a new era where all patients with all rare diseases around the world can come together with the push of a button to analyze data.
Enabling the Medicine of the Future
To work with vast amounts of data on specific diseases, SAIL is increasingly seeking partnerships with patient associations and consortia of health groups, including an international healthcare consultancy and the Kidney Cancer Association. Partnerships also align SAIL with patients, the group they most want to help.
Overall, the founders are pleased that SAIL is solving the problems they faced in their laboratories for researchers around the world.
