The following article originally appeared on WSJ.
Whether based on your family history, your choices, your risk factors, or socioeconomic factors, healthcare is deeply personal at the individual level. “But,” says Laurent Rotival, SVP of strategic technology solutions and chief information officer at Cambia Health Solutions, “it’s not at all personal at the systemic level.” It’s a one-size-fits-all approach, when no two individuals are quite the same.
Shifting that paradigm is critical for delivering higher-quality service and better outcomes, with personalized and predictive treatments based on genomics, targeted therapies, and an individual’s health profile. The key to progress lies in the vast quantity of data from sources ranging from medical imaging to electronic medical records. IDC projects the total universe of healthcare data to grow approximately 400% between 2020 and 2025, and with it the potential for better predictive diagnoses, individualized therapies, and new efficiencies that can increase access while controlling costs.
To realize that potential, the data must be made actionable, which is where the storage and computing capacity of the cloud—and cloud-driven artificial intelligence and machine-learning solutions—come into play.
By using AI and ML to drive new insights, automate tasks, and re-engineer processes, organizations from pharmaceutical companies to hospital systems can transform healthcare into something genuinely personal, says Taha Kass-Hout, director of Health AI at Amazon Web Services (AWS). “I see a future where we’re going to learn a lot more about ourselves and be a lot more engaged with ourselves at a level where personalization becomes something democratized worldwide; where anyone around the world will be able to pick the right options for themselves based on whichever environment they live in.”
Lifting Burdens Off Patients and Providers Alike
For enterprises like Cambia Health Solutions—a family of over 20 companies working to make healthcare more economically sustainable and efficient—the first step to reaching that goal lies in a data-driven approach to enable a smarter, more effective use of digital resources, Rotival says.
He points to the member care programs of Cambia’s regional health plans, which employ predictive analytics leveraging an individual’s preferences, claims, clinical, prescription, and immunization records to create a single risk-stratified view of high-risk members. The case-management team proactively contacts these members to offer access to teams of customer service, clinicians, and case managers available through teleservices, phone, and live chats. AWS gave Cambia the GPU computing power to train and serve machine-learning and natural language-processing models with larger amounts of data, much faster. With that, Cambia is capable of building highly innovative solutions which drive personalized care recommendations, increased efficiency, and better outcomes.
Meanwhile, Cambia is exploring AI services such as Amazon Textract and Amazon Comprehend Medical for automating lower-level functions and redirecting human talent toward innovations that add more value for Cambia’s health plan members. These services are capable of pulling medical data from sources like member charts, hospital bills, and nurse notes to build a deeper understanding of an individual’s medical history and risk profile.
Pushing Toward Personalization
“There are about 150 million people living with a chronic condition in the U.S.,” says Anmol Madan, chief data scientist at Livongo, a platform that includes a mobile app that works with smart devices such as connected blood glucose monitors to help people manage medical issues. “Treating people with these conditions accounts for about 90% of the costs associated with our healthcare expenditures, or over $1.1 trillion of direct healthcare costs.” Empowering people to manage diabetes, hypertension, and weight and behavioral health challenges is Livongo’s mission. And it’s one suited perfectly for intelligent solutions.
“Every aspect of our business is effectively a machine-learning problem or an AI problem,” Madan says.
Using machine-learning algorithms, Livongo, which in early August entered into an agreement to merge with a major telehealth company, is able to translate data from blood-glucose readings, physical activity, and meal logs acquired using Livongo-provided devices, plus smartphone data and other important health data, into timely and actionable “health nudges.” These personalized messages around diet, exercise, medications, and more—delivered in real time to members on their connected devices—help them avoid complications that could land them in the hospital, saving the system (and themselves) money.
“Think of a mom of three, living in a rural community and managing her diabetes, whose blood glucose monitor records a high blood-sugar reading. That’s a real-time data input we can combine with her clinical history to predict the best way to support her,” Madan says. “It might be on the phone, with one of our certified diabetes educators. We may learn this individual is better supported by nutrition guidance in our mobile app. If our algorithm predicts adherence to a medication regimen will be a problem, we can connect her to a pharmacist to receive further guidance. So instead of starting with hundreds of different options, we pick the ones that are most relevant based on her data.”
Using Amazon Transcribe, Livongo also derives intelligence from thousands of conversations, both online and by phone, between members and Livongo’s network of health coaches. “We can digitize those, then use natural language processing to really understand what’s happening in our coaching sessions: topics people talk about at an aggregate level and also at an individual level,” Madan says. “It helps us learn and look for patterns at scale, then take that information and personalize the rest of a member’s care.”
In a mobile world where consumers have multiple avenues for engagement, ”the more surfaces you have to understand a member, the more accurate, relevant, and targeted the personalization,” he notes. “Ultimately, when you do that millions of times over, you build really rich machine-learning models, which help you do this even better—and target better clinical outcomes.”
The Cutting Edge of Individualization in Healthcare
Machine learning offers other opportunities for advancement in healthcare, including cutting-edge therapies tailored to individual patients. Among the leaders in this space is Moderna, which uses messenger RNA (mRNA)—instruction molecules used by cells to produce protein—to develop a wide array of medicines, including bespoke cancer treatments as well as potential solutions for infectious diseases, such as its vaccine candidate for COVID-19.
To develop personalized cancer treatments, the company creates a profile of a patient by comparing healthy cells to tumor cells, identifies mutations, and produces a unique mRNA therapy designed to train that person’s immune system to attack tumor cells. Finding the right design would be tedious without machine learning. “To be able to design the optimal messenger RNA therapy, we use a machine-learning model that helps us identify which of the hundreds or thousands of mutations will trigger the most robust immune response,” says Marcello Damiani, chief digital and operational excellence officer at Moderna.
Moderna uses AWS to run everything from manufacturing, accounting, and inventory management workloads—and even relies on AWS to power its automated production facility.
Because AWS brings the expertise of providing scale and efficiency on demand, Moderna is free to focus on cutting-edge therapies, spending less time overseeing the usage of servers or searching for critical information. “We extract all the data provided by the instruments on site and send them to the AWS cloud,” Damiani says. “We have this wealth of integrated data in one place that we can leverage for insights.”
That reliable, flexible infrastructure proved critical when Moderna faced one of its biggest challenges: developing a COVID-19 vaccine candidate. By combining its mRNA capabilities with an AWS-powered research engine, Moderna was able to provide the first clinical batch of its vaccine candidate to the National Institutes of Health for a Phase 1 trial just 42 days after the initial sequencing of the virus. “It took lots of energy and time from people,” Damiani says. “But if we didn’t have all this infrastructure in place before, it would have been very difficult to meet such a short cycle.”
Bringing Scalability and Flexibility to Healthcare Organizations
As they continue to address expanding needs driven by everything from the coronavirus pandemic to an aging population, healthcare organizations are also facing continued stresses on their technology resources. According to IDC, overall IT spending by healthcare organizations is expected to decline less than 1% this year, down from pre-pandemic forecasts of as much as 10% growth.
“CIOs in healthcare are being challenged. Medical teams and clinicians are trying to experiment with new ways of improving quality, reducing costs, and generally improving the health of the population,” says Shez Partovi, director of healthcare, life sciences, and genomics at AWS.
“That problem is one AWS solves. When the clinical teams want to experiment, it gives the CIO the lever to say, ‘OK, let’s try.’ “They quickly spin up the required cloud infrastructure and run the experiment,” he says. “With the elasticity provided by the cloud infrastructure, if the experiment works, they can just continue scaling.”
The end result, Kass-Hout says, is a healthcare ecosystem no longer treating people like risks or costs, but providing an experience truly individualized for every patient. “We talk about value-based care, reduction of costs, reduction of errors, and improving outcomes. That requires data about each individual. It’s a pattern or prediction—and extracting knowledge from it. Human beings, no matter how smart we are, can use a tool like AI to help us focus, prioritize, and summarize information.”