The convergence of artificial intelligence and healthcare is generating tremendous enthusiasm across the medical community. Privacy-preserving AI healthcare solutions have already demonstrated remarkable potential in enhancing disease diagnosis and treatment, accelerating pharmaceutical development, uncovering genetic connections to various illnesses, and transforming countless aspects of patient care.
Through sophisticated analysis of extensive datasets and pattern recognition, virtually every innovative algorithm holds the promise of improving patient outcomes. However, AI researchers face significant challenges accessing the high-quality data necessary to train and validate these advanced systems. Medical institutions remain understandably cautious about sharing confidential patient information with research teams. Even when data sharing occurs, ensuring that researchers access only essential information and properly dispose of it afterward presents substantial difficulties.
Secure AI Labs (SAIL) has developed groundbreaking technology that addresses these critical challenges by enabling encrypted healthcare AI algorithms to process datasets without ever leaving the data owner's secure environment. This federated learning medical data analysis approach empowers healthcare organizations to maintain complete control over their valuable datasets while allowing researchers to protect the confidentiality of their proprietary models and inquiries. Neither party needs direct access to the underlying data or models to collaborate effectively.
SAIL's innovative platform can also seamlessly integrate information from multiple sources, generating comprehensive insights that fuel the development of more sophisticated and effective algorithms.
"Researchers shouldn't need to spend years cultivating relationships with hospital administrators before implementing their machine learning algorithms," explains SAIL co-founder and MIT Professor Manolis Kellis, who established the company with CEO Anne Kim '16, SM '17. "Our mission is to benefit patients, support machine learning scientists, and facilitate the creation of innovative therapeutics. We want the most advanced algorithms to be applied to the most extensive datasets possible."
SAIL has already established partnerships with leading hospitals and life science companies to provide researchers access to anonymized data. Within the next year, the company aims to collaborate with approximately half of the nation's top 50 academic medical centers.
Unlocking AI's Complete Potential
While studying computer science and molecular biology as an MIT undergraduate, Kim collaborated with researchers in the Computer Science and Artificial Intelligence Laboratory (CSAIL) to analyze data from clinical trials, genetic association studies, and hospital intensive care units.
"I recognized that data sharing mechanisms were fundamentally broken, whether hospitals were using hard drives, outdated file transfer protocols, or even physical mail," Kim notes. "None of these methods offered proper tracking or security."
Kellis, who also serves as a member of the Broad Institute of MIT and Harvard, has spent years establishing partnerships with hospitals and research consortia focusing on various conditions including cancers, cardiovascular disease, schizophrenia, and obesity. He understood that smaller research teams would face significant obstacles accessing the same data his laboratory utilized.
In 2017, Kellis and Kim made the strategic decision to commercialize their technology for enabling AI algorithms to operate on encrypted data.
During summer 2018, Kim participated in the delta v startup accelerator program organized by the Martin Trust Center for MIT Entrepreneurship. The founders also received support from the Sandbox Innovation Fund and the Venture Mentoring Service, establishing numerous early connections through their MIT network.
To join SAIL's program, hospitals and healthcare organizations make selected data available to researchers by establishing a node behind their firewall. SAIL then transmits encrypted algorithms to the servers hosting the datasets through a process known as federated learning. The algorithms process the data locally within each server and return only the results to a central model, which updates itself accordingly. No party—including researchers, data owners, or even SAIL—gains access to either the models or the datasets.
This approach enables a much broader range of researchers to apply their models to extensive datasets. To further engage the research community, Kellis's MIT laboratory has begun organizing competitions providing access to datasets in areas such as protein function and gene expression, challenging researchers to predict outcomes.
"We invite machine learning researchers to train on historical data and predict current results," Kellis explains. "When we identify new algorithm types demonstrating superior performance in these community-wide evaluations, researchers can implement them locally across numerous institutions, creating a level playing field. The only factor that matters becomes algorithm quality rather than connection power."
By enabling numerous datasets to be anonymized into aggregate insights, SAIL's technology also facilitates research into rare diseases, where small pools of relevant patient data are typically dispersed across many institutions. This distribution has historically made applying AI models to such data particularly challenging.
"We envision a future where all these datasets become accessible," Kellis states. "We can break down existing silos and usher in a new era where patients with rare disorders worldwide can connect with a single click to analyze their collective data."
Pioneering Future Medicine
To work with substantial data volumes related to specific diseases, SAIL has increasingly pursued partnerships with patient associations and healthcare consortia, including an international healthcare consulting firm and the Kidney Cancer Association. These alliances also connect SAIL with patients—the primary group they aim to assist.
Overall, the founders express satisfaction that SAIL is addressing challenges they encountered in their laboratories for researchers worldwide.
"This problem requires an industry solution rather than an academic project," Kellis emphasizes. "We're creating a platform not just for my laboratory but for any researcher. It's about establishing an ecosystem encompassing academia, researchers, pharmaceutical companies, biotechnology firms, and hospital partners. I believe integrating all these diverse areas will transform the vision of future medicine into reality."
