Google continues to build out its health-related initiatives leveraging artificial intelligence to help consumers find information on internet searches, advance language-generating tech and offer new tools for developers to build health apps.
The tech giant Tuesday also unveiled new AI partnerships focused on cancer screenings and maternal health and efforts to give Fitbit users easier access to health and wellness data from their devices.
"The future of health is consumer driven. People will expect a mobile-first experience with more personalized insights, services and care. That means enterprises including Google will need to evolve to meet consumers where they are," said Karen DeSalvo, M.D., Google’s chief health officer, during Google Health's annual The Check Up event at the company's new Pier 57 Manhattan office.
"Google Health is our companywide effort to do just that. Building health into the products and services people already use every day," she said.
More than a year ago, the tech giant announced a shift in its healthcare strategy to weave health efforts into its existing business lines like YouTube and search features. Google had previously taken a more siloed approach to healthcare but decided to unwind its unified health division in August 2021.
DeSalvo, a physician, now leads a clinical unit that counsels several Google divisions, such as Maps, hardware and cloud, according to reporting from Bloomberg.
Google is ramping up efforts to integrate AI into its search capabilities to expand the health information that it surfaces for consumers. Last fall, the Google Search team rolled out multiple features for its search engine that aim to make it easier for users to access key information about obtaining Medicaid and Medicare benefits. Building on those efforts, Google said it will make Medicaid re-enrollment information easier to find on Search.
Millions of people signed up for Medicaid during the pandemic. At that time, the requirement to re-enroll each year was paused, but that pause is about to expire March 31, which means that if people fail to re-enroll, they will lose their healthcare coverage.
"Information is a determinant of health and in fact, every day millions of people go to search and ask health-related questions," DeSalvo said during the event Tuesday. "In 2021 alone, health videos on YouTube were viewed more than 100 billion times. By delivering authoritative and authentic health information, we're helping people find agency to navigate their health journey."
The company also announced that it’s making it easier for people to find affordable healthcare centers near them. Consumers will be able to see providers that identify as community health centers offering free or low-cost care, according to Hema Budaraju, senior director of product, health and social impact at Google Search.
Google also leveraged its conversational AI technology, called Duplex, to verify providers' information on Google Search.
"We know how important it is when you're searching for healthcare providers that the information you find is accurate, like the clinic's phone number and address. Duplex called hundreds of thousands of healthcare providers in the U.S. to verify their information on Google Search. And we’ve used this technology to verify if providers accept certain Medicaid plans in their state," Budaraju said during the event.
The tech giant also is partnering with ThroughLine in New Zealand, which is a network of mental health and crisis helplines, to increase the number of crisis helplines that appear at the top of search results in additional languages and countries for searches related to suicide, domestic violence and other personal crisis topics.
In the area of health and wellness, Google also will make more of Fitbit’s health metrics dashboard feature available to users without a subscription to surface trends about users' health, such as breathing rate, skin temperature and blood oxygen.
Open source tool kit for health apps
Google launched Open Health Stack, a suite of open-source building blocks to help developers quickly build digital health apps in under-resourced areas.
Digital health solutions built to date have focused on solving for single, disease-focused conditions, leading to data silos and making it difficult for healthcare workers to access the patient information they need, according to Fred Hersch, senior product manager of Open Health Stack, in a blog post.
Over the past few years, Fast Healthcare Interoperability Standards (HL7 FHIR) has created an open standard to address these problems.
"At Google, we really want to accelerate the future of digital care in low-resource settings to lower the barrier to equitable healthcare. That's why we're announcing the launch of Open House stack, the suite of open-source tools to help local developers create next-generation healthcare apps," Katherine Chou, senior director of research and innovations at Google, said during the event Tuesday. "Open Health Stack uses FHIR resources as its underlying data standard that makes it easier to build patient-centered solutions that can connect within and across healthcare systems. For example, one component, the Android FHIR SDK, makes the data stored on the app secure and accessible offline so that it's safe and useful even in places without cell phone coverage or internet access."
AI research efforts
Google also shared updates health AI updates on its progress on medical large language model (LLM) research, partnerships that are bringing solutions into real-world settings, and new ways AI can help with disease detection.
The tech giant continues to invest in medical large language model research. LLMs are AI tools that demonstrate capabilities in language understanding and generation—has opened up new ways to use AI to solve real-world problems.
"Late last year, we took our first step towards rethinking conversational AI systems in medicine with Med-PaLM, a large language model designed to provide high quality and authoritative answers to medical questions," said Alan Karthikesalingam, M.D., Ph.D., a surgeon-scientist who leads the healthcare machine learning research group at Google Health.
Med-PaLM is a version of PaLM tuned for the medical domain. Med-PaLM was the first to obtain a “passing score” (>60%) on U.S. medical licensing-style questions. This model not only answered multiple choice and open-ended questions accurately, but also provided rationale and evaluated its own responses, according to Google executives.
The technology's next iteration, Med-PaLM 2, consistently performed at an “expert” doctor level on medical exam questions, scoring 85%. This is an 18% improvement from Med-PaLM’s previous performance and far surpasses similar AI models, Yossi Matias, vice president of engineering and research at Google, and Greg Corrado, Ph.D., head of health AI at Google, wrote in a blog post.
Google is competing with other tech rivals in the area of generative AI. In response to OpenAI Inc.’s surprise success with ChatGPT, Google’s senior management declared a “code red” at the company and directed that all of its most important products—those with more than a billion users—must incorporate generative AI within months, Bloomberg reported.
"There are many ways that an AI system like Med-PaLM can be a building block for advanced natural language processing and healthcare. And we'd like to work with researchers and experts to advance this work," Karthikesalingam said during the event. "The potential here is tremendous, but it's crucial that real-world applications are explored in a responsible and ethical manner."
Also in the AI space, Google said it's been working with Mayo Clinic to explore how AI can support the tedious, time-consuming process of planning for radiotherapy, a common cancer treatment used to treat more than half of cancers in the U.S. The most labor-intensive step in the planning process is a technique called “contouring,” where clinicians draw lines on CT scans to separate areas of cancer from nearby healthy tissues that can be damaged by radiation during treatment. This process can take up to seven hours for a single patient.
Google and Mayo Clinic have focused on developing a radiotherapy model to automatically outline or contour around organs on CT scans. "Three years ago, we partnered with Mayo Clinic to start exploring how AI could help clinicians spend less time on contouring. Much like Google Maps can detect billion buildings and roads using AI, we built technology that could detect organs in CT scans and quickly outline them for review by specialists," Corrado said Tuesday.
Google also is building AI models to help simplify acquiring and interpreting ultrasound images to identify important information like gestational age in expecting mothers and early detection of breast cancer. The company is partnering with Jacaranda Health, a Kenya-based nonprofit, to explore how new AI tools can support point-of-care ultrasound for pregnant women.
Google also is working with Chang Gung Memorial Hospital in Taiwan to research using AI models to help with the early detection of breast cancer using ultrasound.
Through another partnership, this one with not-for-profit Right to Care, Google is working to make AI-powered tuberculosis screenings widely available across sub-Saharan Africa.