Podcast: Play in new window | Download
The future of health will be shaped by consumer expectations for a mobile-centric experience with personalized insights and care services. Information is determinant of health, where people already search for health information on Google hundreds of millions of times a day. Additionally people view YouTube videos about health conditions 100 billion times globally in a year. As consumers seek information ubiquity in their online experience, health information will also become more personalized through wearables and other mobile devices.
Our future in health will also be enabled by AI. Artificial Intelligence has the potential to transform the health of people on a planetary scale akin to the discovery of penicillin. If developed boldly and responsibly, AI will be a powerful for health equity on a global scale. It will also bring the joy back to practicing medicine by reducing cognitive burden and giving providers more time to spend with patient.
In this week’s episode, we explore health tech consumerism and AI enablement with Dr. Karen DeSalvo, Chief Health Officer at Google. Dr. DeSalvo is an internist and health leader working at the intersection of medicine, public health, and information technology. She has dedicated her career to improving health outcomes for all with a focus on solutions that address all the determinants of health. Dr. DeSalvo continues to be a powerful voice and advocate for eliminating inequities and improving the public’s health. Under her watch, Google has optimized search and YouTube to better answer common health questions, updated its consumer health wearables to function more like medical devices and built artificial intelligence products to meet industry demands.
This episode covers various topics in the realm of healthcare technology innovation from consumerism, Generative AI and LLMs, health equity by design, and various initiatives underway at Google to connect and bring meaning to health information. In the interview, we also discuss the role of technology in mitigating the health impacts of climate change and addressing the epidemic of loneliness and isolation at a global level.
Episode Bookmarks:
01:30 Introduction to Karen DeSalvo, Chief Health Officer at Google.
03:30 How Google understands “information as a determinant of health.”
05:00 “We see heavy consumer orientation to the way we see our opportunity to improve the health of everyone everywhere.”
05:30 The evolution of healthcare businesses to meet people in an increasingly virtual world with ever-changing consumer expectations.
06:00 Informing health empowerment through high quality information and personalized insights.
06:30 Personal reflections from clinical practice when the flow of information was not enabled by technology automation.
07:30 Modern-day tools for patient education and personal health tracking and measurement.
08:00 The optimization of Google search results to convey trust in the provision of health information.
08:30 “The conveyance of information through trusted messengers is an important way we address information as a determinant of health.”
09:00 Patients showing up with more knowledge and power – a priority goal for Google Health.
09:30 How AI can improve health for everyone everywhere. (Karen’s recent blog on the future of AI as a transformational path forward in population health.)
10:00 Leveraging AI at Google Health to advance medical research, improve accuracy and efficiency of diagnostic processes, and improve health information quality.
11:00 A future world were everyone has access to the best quality care on their phone (e.g. AI-enabled health agents combined with the human care team).
12:00 Developing health technology for the entire world. (“A billion people on the planet don’t have access to primary care.”)
12:15 AI can address workforce challenges by reducing cognitive load to address burnout and filling capability and capacity gaps.
12:45 Using large language models in high risk health scenarios (e.g. suicide-related searches, AFib detection on watch products).
13:45 How ChatGPT elevated the interest in front-facing, consumer-focused large language models.
14:15 Search generative experience and the development of Bard (a conversational AI tool by Google).
14:45 Enhancing the Fitbit experience with generative AI and large language models (seearticle)
15:00 Med-PaLM harnesses the power of Google’s LLMs in the medical domain to accelerate scientific advancement.
16:00 “Do no harm” as the ultimate guiding philosophy in AI innovation.
16:30 Human oversight needed to mitigate the risk of AI hallucinations.
18:30 Google use cases for generative AI in clinical care — from helping radiologists to detect breast cancer, to supporting diabetic retinopathy screening.
19:00 Using AI to enhance radiology clinician workflows in mammogram screening.
20:00 Using AI to identify multi-drug resistant tuberculosis in Sub-Saharan Africa.
21:00 Developing an AI model that can perform many tasks at once (e.g. medical record summary, presenting research evidence and recommended clinical pathways)
22:30 Reduction of algorithmic bias to improve health equity and quality of care.
22:30 “The next generation of AI models are going to bring joy back to medicine. They will reduce cognitive burden and give providers more time to spend with patients.”
26:45 Reduction of implicit bias and the “health equity by design” approach at Google.
28:00 Solving for global health inequities observed with the 4M’s (i.e. metabolic disease, malignancy, maternal health, and mental health).
29:00 Developing cloud partnerships and research collaborations for population health in action.
30:00 Ensuring effective deployment of tools and technology for everyone everywhere.
30:45 Diversity, equity, and inclusion of Google Health’s teams.
31:45 Challenges and risk in the early stages of AI foundation models.
33:00 Additional background on Med-PaLM as a medically-tuned, domain specific large language model.
35:00 Use cases for Med-PaLM in care delivery, payer, and life sciences environments.
36:00 Multimodal application of Med-PaLM with wide-ranging capabilities to enhance clinician workflows.
36:45 “Large language models have many capabilities. Constraining them to do the appropriate thing is such an important priority.”
38:30 There is no one-size-fits-all approach to LLMs – it depends on the preferred use case (e.g. writing contracts, customer chatbots, distilling research insights, pop health enablement).
40:00 Climate change is the biggest threat to global public health.
41:00 “Health is more than health care.” (The impact of social determinants and the physical environment on health and wellbeing).
42:00 Applying data and analytics during the pandemic to improve health equity.
43:00 Novel signals from Google searches on symptoms — identifying trends to inform public health priorities.
44:00 The need for timely, granular, and actionable data in the public health setting.
44:30 AI predictions for future climate risks associated with health, fire, and floods.
45:00 Notifying patients with pulmonary disease of air quality risks in real-time.
46:00 The impact of social isolation on public health due to its association with a range of negative physical, mental, and emotional outcomes.
46:45 The health impact of loneliness is so far-reaching that one study compares it to smoking up to 15 cigarettes a day!
47:45 Dr. DeSalvo’s important work on the WHO Commission on Social Connection to better address the global loneliness challenge.
49:00 Using technology to address loneliness and isolation (e.g. AI and robotics to augment human support for seniors).
50:00 The detrimental impact of social media on the youth – striking a balance between screen time and human interaction.
51:30 Parting thoughts from Dr. DeSalvo pm the importance of health for all and value for all.