Learn about the latest R&D underway at HealthUnity
Developing the Next Generation of Easy Access Health Information through Artificial Intelligence
The AI Health Navigator: AI models trained on multimodal domain knowledge and health data for medical information that is accessible, credible, and easy to understand for all
In a quest to unlock the potential of large-scale AI in healthcare via open knowledge and an open data community, HealthUnity asked: How can we enable people to navigate and manage their health with a better understanding of their personal conditions through credible domain expertise and real-world data?
The result: Access to trustworthy medical information from all or specific knowledge sources on any health topic in an easily understood language that demystifies individual health conditions and assists users in navigating expert healthcare knowledge.
The AI Health Navigator is the first digital health platform to combine domain-specific knowledge, open data knowledge, and privacy-protected personal health information using foundation models to navigate enormous amounts of data for instantaneous health information that would not otherwise be humanly possible. Advancements in multimodal AI, using a combination of text, voice, video, medical images, and other types of data sources, play a critical role in the platform development.
Powered by generative AI, the Health Navigator communicates like a human in a conversational manner, translating medical terms and domain expert literature into easy-to-understand text in any language. Individuals can ask questions and be empowered to advocate for their health through information that is credible and personalized.
Think of the AI Health Navigator as a next-gen conversation agent, beyond Alexa and Siri, with the knowledge of WebMD or a Wikipedia for Healthcare that is more interactive, more understandable, more empathetic, and more personalized. With a mission to transform the way we understand our own health, the platform makes healthcare knowledge accessible and understandable for everyone in their own context – bridging key stakeholders in a patient’s health journey, including clinical experts, patient advocates, and caregivers, together with the individual patient themselves gaining a better understanding about their own personal health.
Health equity has been the driving force in the ethical design of the AI Health Navigator with models based on inclusive world data from diverse groups for more trustworthy and equitable information.
“Imagine being able to go to one source for personalized, accurate information on all health and well-being needs. Expert knowledge from research papers, medical books, and healthcare authorities is translated into a simple, easy-to-understand language through machines trained on domain expertise and real-world data.”
The platform enables users to interact with the trained model for individualized support. Ask any questions about symptoms or health conditions and engage in an interactive conversation. Information gathered from domain expert sources addresses everything from chronic diseases, oncology, and mental health to symptoms-based navigation about side effects from medication, COVID-19, or cold and flu inquiries. The AI Health Navigator is a one-stop source for everything related to health and well-being, including support for basic health questions, nutrition and recipes for a healthy lifestyle, recommendations for exercising and improving sleep patterns, rehabilitation programs, and preventative care such as the top cancer-fighting foods and insights addressing unhealthy behavioral habits.
By syncing personal data from wearables, health tech devices, apps, and health history records, personalized information provides a more thorough look at how behaviors may impact an individual’s current health state.
“Receive trustworthy recommendations and in-depth clinical information from credible expert health resources.”
HealthUnity’s goal is for everyone to have access to trusted health information and recommended care for a better understanding and management of their personal health trajectory and a better quality of life.
Bridging Data Silos in Healthcare
The reason HealthUnity is able to develop a more robust knowledge generator is due to large-scale connected data enabled through renowned interdisciplinary team expertise and leading research institutes. Large language models based on domain-specific knowledge and multimodal heterogeneous health datasets will be combined with personal health information, all in one place, to provide recommended resources for users with a 360° holistic and longitudinal view of each person’s individual health.
Current medical sites essentially have credible domain knowledge but the terminology is highly clinical and can often be difficult to understand or even pronounce. On the other hand, generative AI that is learned can be understandable but may not have the domain knowledge or credibility and can be misleading. Neither of these provides a full holistic view of the patient. This has been an ongoing issue in health systems.
- We may know what a person’s sleep pattern is but we don’t know what they’re eating or other behavioral patterns
- We may know that they have a cough but we don’t know their environmental circumstances, such as air quality
- We currently only have a snapshot of the patient when they visit a medical site or are admitted to the ER, not the full picture of how they are living 365 days outside of the hospital
Harnessing the Power of AI in Healthcare
Generative AI offers powerful tools helping to generate content in a conversational style through natural language processing (NLP), among many other applications. It currently has its limitations though, as we have seen in its infancy with answers that are often generic with no assurance of accuracy. This language is typically learned from online news, social media, and large-scale web content that lacks domain knowledge.
“The critical missing piece is being able to combine this powerful generative AI capability with domain-specific knowledge and real-world evidence to provide relevant medical information translated into layman’s terms that are easier to understand.”
The generative transformer model, pre-trained in medical and biomedical domains, mines health information and transforms in-depth clinical literature into general public terms. Clear and simple language generation based on a variety of discriminative downstream NLP tasks enables instantaneous access to terabytes of data and domain-specific information through the AI Health Navigator large language model. Continued exponential growth in computational capabilities is anticipated, as with all AI/ML advancements.
The AI Health Navigator Roadmap
Phase I: The Training Phase – Training the dialog format will be based on the most advanced generative AI as the core backbone with knowledge injected into it by linking open data knowledge graphs combined with knowledge training and domain knowledge used to fine-tune the model. The model will be trained with heterogeneous data, domain-specific literature, and patient volunteers recruited for personal information data.
Phase II: Launching the navigator based on a combination of public general knowledge and domain knowledge.
Phase III: Launching the navigator based on public general knowledge, domain knowledge, and personal health conditions. Use of personal multimodal data about health and lifestyle will be the primary source of personal data used in context of the individual while answering questions.
Generative AI is only as good as its Data
HealthUnity, a non-profit open data consortium, was founded by leading data scientists, AI/ML experts, and healthcare visionaries with the intention of generating meaningful research through the collection of large-scale multimodal heterogeneous data and the development of useful products for personal and societal health. The team of renowned leaders in artificial intelligence and domain expertise is currently collaborating with top research partners for world-class data collection.
Generative transformer language models will be trained on near-limitless digitally published domain content including in-depth medical and biomedical information from medical books, journals, publications, research papers, and medical websites, in combination with the real-world evidence from data collections and the user’s personal health information. Multimodal data about an individual’s health and lifestyle from synced devices and EHR/EMRs will be the primary source of personal data used in context of the individual while answering questions.
The goal of HealthUnity extends beyond patients and practitioners with plans to become the go-to platform for scientists, researchers, and industry as a large-scale data source for clinical trials and drug development.
Predictive models within the platform are expected to play a critical role in scientific advancements, helping treatments reach patients faster and managing patient health for a better quality of life. “As an example, leveraging heterogeneous longitudinal information, will help us gain extensive knowledge with a better understanding of patient outcomes from large groups of patients who share similar behaviors and diseases,” said Jia Li, HealthUnity Co-founder and Chair.
HealthUnity’s novel approach is connecting the dots for the first comprehensive combined knowledge platform in healthcare, inclusive of domain-specific knowledge, open data knowledge, and privacy-protected personal health information, developed to enable accessible knowledge 24/7 for better information and better health.
The ambitious project brings together valuable medical information, data science, advanced generative AI, and inclusive real-world evidence covering expansive topics on health and well-being to help people live happier, healthier, more enjoyable lives.
The vision for the AI Health Navigator is to leverage AI, data, and domain knowledge to provide continuous care throughout a patient’s personal health journey and transform lives through improved health management.
The AI Health Navigator is currently in the early stages of R&D. We welcome collaborators in AI/ML, science, technology and healthcare to join us on our mission to improve access to health knowledge for personal and societal health and well-being.
To get involved, please contact: firstname.lastname@example.org