Everyone’s health and wellness journey is unique, and AI can help create more personalized health experiences to match this individuality. We're researching how fine-tuning Gemini models for personal health and wellness can unlock novel insights and interactions for consumers. Imagine an AI that not only understands your health data from wearables but also gives you personalized insights and recommendations, like a virtual health coach. Our two new research papers explore this in depth: Towards a Personal Health Large Language Model (PH-LLM): We've fine-tuned Gemini models to create PH-LLM, which generates personalized sleep and fitness recommendations comparable to human experts. Transforming Wearable Data into Personal Health Insights using Large Language Model Agents: This agent system leverages code generation and information retrieval to analyze your wearable data, providing detailed insights into your health patterns. These advances demonstrate how AI can help us better understand our health data and make more informed decisions about our well-being. We look forward to continuing research into how generative AI can enable novel health experiences. Read more: https://goo.gle/3RnwHbl
Dennis Esser
2w
Great 💪🏻! Tho could even be extended with blood work, DNA data, lab results, doctors letters, lifestyle audits etc. etc. Would love to learn more about the development and roadmap 👀
It should be useful to create an AI empowered platform to enable people with different health conditions, needs, and other factors such as their budget to get reccommendations for monitoring programs. Budget doesn't need to exclude what could be better for them than what is suggested based on it. Besides, even for those who can't afford something but maybe at a higher risk, there is all justification for government to step in to reduce its future healthcare burden. Moreover, linking health programs to some incentives or maybe penalties could promote overall health gains for a country. Individuals who don't follow recommendations even when they could and despite repeated reminders may be penalized in advance to urge them to act.
Google Health Karen DeSalvo - Regarding your "We're researching how fine-tuning Gemini models for personal health and wellness can unlock novel insights and interactions for consumers." - I am a consumer/patient and here's what I'd like to see - - "Future Healthcare Today: How Technology is Revolutionizing Holistic Wellness” - https://books2read.com/u/3nBMDo
How will our privacy and data be protected? Just curious. Advancements like these are both exciting and inevitable, yet without certain protections and safeguards..... I'm looking forward to reviewing your terms of service & privacy policy 😁
It should also be useful to screen people based on conventional criteria of risk and then monitor their health for those risks, e.g. overweight and sedentary lifestyle, and x years of age could together be significant factors, so if groups with such characteristics are monitored, that could generate insights for benefit of others. The benefit could be in terms of alarms in advance for preventive or corrective action
Hana Sediva
2w
Brilliant news. 👏 My recent PhD thesis was on designing personalised multimodal digital health behaviour change intervention to improve fitness, diet (and sleep), while using ML to identify the most relevant techniques that drive such health behaviour change.
Anvil D. Nel
2w
Https://www.lifelinetech.co.za
How do I get involved in the UK??!! I use #geminiai every day to write, analyse, summarise and teach. Adding in personal data would be amazing and transformative. Please please get in touch!!
Best if we can distill what is most critical, and the best minimum, for each individual to monitor and what is easiest to monitor at a certain frequency and/or in easy and standardised tests(maybe at home, self testing) combined with such essence e.g. the pattern of changes in one's heart beat in resting or some other position or during some movement, blood glucose change during sleep or some activity, body temperature changes, SpO2 and its changes, (maybe) rate of respiration, etc. What should be that set of parameters would also depend on what health conditions should one needs to worry most about. Of course, for apparently healthy individuals, the goal could be a little different, say to improve performance - strength, stamina, maintaining weight, etc. So, essentially, such plans should help everyone in some way. Moreover, and I wrote this earlier too on a similar post, that for some people, it may also be possible to virtualize the personalised insights generated by a thorough monitoring. That could be stored as a model in an app with the original data at the back end (at the server), and the model and data can be upgraded by a monitoring plan every 3 or 6 months.