Imagine having instant access to your MRI from 10 years ago, the doctor’s notes from your ER visit last fall, and every test result you’ve ever received—all in one place. With AI, this scenario becomes more real thanks to its automation and processing capabilities. Your doctor could securely view your entire medical history and devise a treatment plan that’s uniquely suited for you. Sounds great, right?
Well, we might not be as far off from that reality as you think. Advances in technology have made healthcare more efficient and personalized in the last decade. The American Medical Association (AMA) reports that the adoption of digital healthcare tools to provide remote care, like telehealth and remote monitoring on wearable devices, has risen significantly since 2016, as overworked doctors try to reduce stress and burnout while providing the best possible patient care. In 2022, 93 percent of physicians cited digital health tools as an advantage for patient care.
AI offers the healthcare industry the opportunity to take an even more personalized approach to healthcare. Pharmaceutical and healthcare companies of all sizes—from the world’s largest biotech companies, like Roche, to data-driven healthcare platform start-ups, like Omada Health—are using AI to better analyze patient healthcare records, develop precision medicines, and create more customized patient support.
Improving the Speed and Accuracy of Diagnoses
The scenario about an AI-streamlined healthcare system is what Mustaqhusain Kazi, the global head of Roche Informatics strategy and digital innovation and chair of Alliance for AI in Healthcare, calls the holy grail.
“From doctors’ notes to test results, imaging scans, and a patient’s reported data from an iPhone app, a doctor can triangulate these datasets to see your whole medical picture,” Kazi says. Over time, doctors can use this data to track the progression of a disease and make more accurate and specific recommendations for your health. With your consent, and in a secure way, your doctor can also look at your data and compare it to aggregate information on patients with similar medical histories to better understand the disease progression, make earlier diagnoses, and create better treatment plans.
Collating all this data, however, is a daunting task that requires integrating unstructured data, like hand-written doctors’ notes and imaging scans, with structured data, like test results or data from a wearable app. If it were to be done manually, it would require an untenable number of human-powered hours and would present myriad opportunities for manual data-entry errors.
To solve this problem, Roche has created Apollo, an end-to-end platform built on AWS that leverages AWS’ advanced analytics tools. The Apollo platform’s AI can read your doctor’s hand-written notes, analyze them alongside your test results and imaging scans, and store them all securely. It can also anonymize datasets and allow researchers and data scientists access to a range of analytics tools with which to generate insights, share findings, and collaborate. As a result, Roche is improving the efficiency of clinical trials, developing new diagnostics, and better matching patients to therapies.
“We’re using the data from the patient to innovate on their behalf,” Kazi says. “And working towards our mission to provide every patient with the best treatment possible in the fastest possible time.”
Increasing Patient Trust
Built on the premise that better personal relationships with care providers drive better health outcomes, virtual-first healthcare provider Omada Health provides each of its members with a care team to support them in making lasting health changes. The care team is composed of diabetes specialists, physical therapists, behavioral health specialists, health coaches, and others who specialize in behavior change and clinical care.
Omada’s care team helps members by, for example, reminding them to take their medications, suggesting healthy eating choices, and recommending ways to lower their stress levels based on individual, real-time health data from Omada’s data-driven platform.
“Working through a health issue is so intensely personal. There’s probably no other place where trust is so important,” says Omada co-founder and CEO Sean Duffy. “How you’re interacting with someone on a daily basis, over time, is what builds that trust—and that’s something you can architect.
“We keep track of your blood sugar alongside more qualitative data—like your preferred method of communication or that you feel happiest when you’re at the ocean. It’s this combination of quantitative and qualitative information that builds a personal relationship, and it’s the personal relationship that drives behavioral health changes.”
Built on AWS, Omada’s platform uses AI to collect and analyze billions of health data points, integrate qualitative data and behavioral science, and give data-driven recommendations with a personal touch. When the platform surfaces a data anomaly—say, a member’s unusually low blood sugar—it automatically alerts the member’s health coach to reach out for a check-in. The platform also suggests an intervention for the coach based on what behavioral science predicts will be most effective for that particular member.
Omada's data-based approach is working. For example, with an average of 15 logins and 31 weekly points of engagement per member, Omada members managing diabetes are 11 percent more likely to take their medication than non-members—and realize better blood sugar control and lower total cholesterol. For those managing hypertension, nearly 50 percent of Omada members reduced their blood pressure levels to a lower stage. And across cohorts, members who message their coach experience two times more weight loss.
How Generative AI Creates Big Opportunities
Healthcare organizations are already using AI to synthesize data, find patterns, and make our healthcare more personal—but generative AI presents many possibilities for taking personalization to the next level.
Generative AI differs from other forms of AI and data analytics, primarily because it can create new content from the data it’s been trained on. Using generative AI, healthcare organizations can create chatbots to improve call center experiences or automate a doctor’s administrative tasks to allow them more time with their patients.
Like most AI, generative AI is powered by machine learning models—very large models, commonly referred to as foundation models, that are pre-trained on vast amounts of data. Healthcare organizations can customize and fine-tune these models on proprietary data for more complex uses, such as supporting doctors in making clinical diagnoses and personalizing patient care. Fine-tuning foundation models with proprietary data can also help ensure that the models are relevant to the organizations’ patient demographic and the models can be built with responsible AI in mind.
“Generative AI promises to enhance the patient experience, deepen the patient-doctor relationship, accelerate the development of new therapies, and improve the effectiveness of treatments and diagnosis—resulting in better health,” says Tehsin Syed, general manager of health AI at AWS. “Early adopters aren’t just talking about the possibilities, they’re moving ahead quickly to identify and explore areas where generative AI can reduce cost or complexity, increase capacity, and improve patient outcomes."
With careful development and thoughtful approaches, Kazi believes the changes AI will bring to healthcare will be some of the most exciting technological advances of our lifetime.
Learn more about building a modern data foundation to deliver the next generation of personalized healthcare with AI, ML, and generative AI.
This story was produced by AWS and edited by WIRED Brand Lab.
