How AI Will Impact Healthcare in 2021 and Beyond

Healthcare is one industry that hasn’t been served particularly well by AI. But due to the innovation of confidential computing, clinical AI algorithms are on the rise and poised to improve healthcare for years to come. Ambuj Kumar, CEO of Fortanix, elaborates on the future of AI in this exclusive article for Toolbox. 

When you look up a product or a service on the web, AI algorithms start targeting you with ads related to that product across websites. These AI algorithms are powerful, to say the least — the recently announced GPT-3Opens a new window has 175 billion independently tuned parameters, for example. Algorithms can optimize which product to show you on which website and also choose specifically colored fonts to maximize the chances of you clicking on a link.  

Advanced algorithms can make us feel invigorated or depressed, or envious by showing exactly the right images on social media at the right time. In fact, they can even be used as a tool for information warfare. But, can AI be good? What is the future of “good” AI technology? 

One industry that has not been served well by technology in general, and AI in particular, is healthcareOpens a new window . If you ask doctors, technologists, or patients, the primary inhibiting factor is sensitivity around healthcare data. If a patient’s data is compromised, lives can be at stake. Further, FDA-approved healthcare algorithms need to work consistently for everyone, whether poor or rich, from California or Missouri, and for all ethnicities.  

Getting clinical approval for a clinical AI algorithm requires vast access to data, which is costly and requires great care for the privacy of patient data. Still, there is one main reason why clinical AI algorithms are on the rise and poised to help improve healthcare in the coming years — for the first time in the world, clinical healthcare AI algorithms are using confidential computing. And this is significant for three main reasons: 

Learn More: 5G Brings Faster Data Transfer and Business Transformation

Patient Data Is Protected With Hardware Enclaves 

Today, confidential computing enables healthcare organizations to develop AI algorithms with patient data in tightly controlled and isolated secure enclaves where they are inaccessible to malware. In this case, the healthcare data can only be used for the intended AI and nothing else, thus eliminating a vast security nightmare.  

The University of California, San Francisco’s Center for Digital Health Innovation recently rolled out an AI project called BeeKeeper AI designed just for this — specifically, to streamline the validation of algorithms and datasets associated with AI capabilities that are embedded in medical devices. Validating these types of algorithms against multiple uniquely distinct datasets owned by multiple organizations is a time-consuming challenge in part because the process is regulated by the U.S. Food and Drug Administration (FDA). Yet, the validation is critical for user safety. 

At a high level, confidential computing creates a trusted environment that provides hardware-based memory encryption to isolate specific application code and data in memory. These secure enclaves increase the security of application code and data and allow multiple organizations to work safely together to validate algorithms while keeping each organization’s data confidential, ultimately protecting intellectual property and patient data. 

Learn More: The Rebirth of Edge Economies Through Digital Infrastructure 

AI Makes Care Delivery More Efficient 

The first use case for UCSF’s technology is for blood transfusion scores. Suppose a patient shows up in an emergency room. The patient’s blood sample can be fed into the AI, which could predict whether the patient will need a transfusion. Hospital staff can get the right blood type ready even before the doctor looks at the patient. 

While the emergency room use case is the organization’s first practical application of the technology, similar AI algorithms can and will be developed for widespread use, including advanced diagnostics and treatment of chronic diseases. The ability to leverage this technology to improve the delivery of care is a win for every stakeholder involved. 

For clinical staff, it streamlines tasks that may otherwise be mundane or time-consuming, freeing them up to focus on other vital tasks and treat more patients. For patients, the delivery of care becomes faster and error-free, in line with what today’s consumers demand across industries, including healthcare. 

Finally, more efficient delivery of care helps improve health outcomes. That means optimized revenue and patient satisfaction for both payers and providers under the increasingly common value-based healthcare model, and, most importantly, healthier patient populations. 

Learn More: Why Palantir’s IPO Reminds Us of Facebook All Over Again

Improved Data Analytics Across Providers 

Healthcare providers are, of course, subject to privacy compliance, most notably the regulations that fall under the Health Insurance Portability and Accountability Act (HIPAA). Yet, there is a somewhat untapped set of data that can be complicated to analyze, such as data shared between two hospitals, because it needs to be done manually for these privacy reasons. 

Most healthcare organizations can’t derive value from patient data effectively because of privacy concerns and data usage regulations. But modern privacy-preserving analytics and confidential computing technology enable the sharing and processing of encrypted datasets from multiple parties, with the datasets only decrypted and analyzed within a secure enclave. Under this scenario, multiple parties can create a contract within the software that runs in a secure enclave, receives keys to decrypt data, runs the analysis, and then encrypts the result. 

What does this look like in the healthcare setting? It could mean combining genetics research datasets with hospital medical data to better predict a disease. Or it might mean using aggregated data insights from across a nationwide or global system to fight fraud that might otherwise go undetected in a single data source. 

Using confidential computing, the aggregate data is never exposed outside of the secure enclave, and the contract is executed as agreed upon by all parties involved, enforcing the data access controls from each party. When applied properly, this method of privacy-preserving analytics is an easy-to-use, efficient, and scalable solution that enables large numbers of parties to create data mashups privately and adhere to regulations. 

All in all, advancements in privacy-preserving analytics and confidential computing prove that technology can be good and quite literally save lives. As more and more health providers in the U.S. and around the world adopt this emerging technology, the result will be an overall improved healthcare system that benefits every level of the industry, patients included. 

Did you enjoy reading this article? Let us know your thoughts on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We would love to hear from you!