Healthcare’s "Missing" Industrial Revolution
Why the revolution is finally coming (and it's not all about AI)
Healthcare AI is in the spotlight right now, but will it live up to the hype? Traditionally one of the slowest to adopt digital technology, US healthcare has been at the forefront of AI adoption with over 70% of large healthcare payers and providers having already piloted AI.1 Billions have been invested in new healthcare AI startups in just the past year, and there are no signs of the excitement slowing down. AI appears to be sparking a new industrial revolution in healthcare.
What’s interesting is that US healthcare never experienced the third (digital) industrial revolution that so dramatically transformed other sectors of the economy. Over the last two decades, labor productivity in the US almost doubled as “the invention of the Internet, web browsing, search engines, and e-commerce produced a pervasive change in every aspect of business practice,” writes professor Robert Gordon, one of the world’s experts on economic growth.2 While there were enormous investments in electronic health records and IT systems, labor productivity in healthcare actually suffered over the last two decades – 64% lower growth than the national average, and one of the worst across industries.3
AI will no doubt play a critical role in helping healthcare finally catch-up, but it will not be the sole driving factor. Misaligned incentives, complex regulations, and adversarial relationships have historically plagued the health system and contributed to the slow adoption of digital technology. Fortunately, a confluence of five factors – two involving AI – has created a unique opportunity for the third industrial revolution to finally come for US healthcare.
First, data interoperability regulations such as the 21st Century Cures Act and CMS’s Interoperability and Patient Access rule are reducing the fragmentation of health data. These regulations are driving the adoption of new standards like FHIR and mandating the use of APIs, enabling electronic data to be more easily shared. While we’re still in the early days – adoption of these standards has been slow, the underlying data is often still messy – the ability to easily access a patient’s data across multiple providers or payers will enable better care coordination, care delivery, and cost savings.
Second, new value-based care (VBC) and insurance models are becoming mainstream and addressing the misaligned incentives in traditional healthcare. Nearly 100 million Americans are now receiving care under value-based arrangements that reward providers with incentive payments for the quality and efficiency of care delivered, as opposed to fee-for-service payments that only reward outputs.4 Medicare Advantage, a value-based model where private insurance companies cover health benefits for members and receive a fixed amount for each member from Medicare, now accounts for 54% of seniors.5 Under these models, organizations now have aligned incentives to become more efficient, including utilizing digital technologies for greater productivity. The shift to VBC has also meant more focus on data – in fact, data and analytics are “musts” for organizations to set up VBC contracts, improve operations, and show cost savings. Some go further though, and utilize data and AI as key differentiators in how they actually deliver care. For example, VBC startups like Strive Health and Waymark rely heavily on data and AI to prioritize patient outreach and preventive care, ultimately driving cost reductions.
Third, there has been a notable cultural shift towards cost containment as stakeholders face unsustainable, rising costs. Employers are reacting to 10-25% annual increases in healthcare premiums, far outpacing inflation, by either cutting benefits or exploring alternatives such as individual coverage health reimbursement arrangements (ICHRA). Half of US hospitals are operating on negative margins and looking for ways to reduce costs. For the first time ever, the federal government has negotiated drug prices for 10 medications to reduce pharmaceutical spending, finally joining other OECD countries in negotiating prices. Economists have been sounding the alarm on our unsustainable healthcare spending since the 1970s, but today at ~18% of US GDP, there appears to finally be a broad, systemic shift towards financial sustainability.
Fourth, AI has created a new culture around analytics/ML/AI experimentation. In 2023, fewer than 10% of healthcare organizations had integrated AI technologies into their business processes.6 But with the launch of ChatGPT and new AI innovations, there’s been renewed interest from payers and providers to adopt AI of all types – not just AI agents and large-language models (LLMs), but also supervised machine learning, computer vision, and AI automation. Today the hype of AI transforming healthcare still far exceeds the actual capabilities of most AI products, but the emergent behavior of experimenting with and piloting AI solutions is opening up opportunities across notetaking, workflow automation, claims and billing, clinical decision support, care coordination, and more.
Fifth, AI will also play a fundamental role in improving data usage and quality. An estimated 80% of healthcare data is unstructured – clinical notes, emails, PDFs, medical imaging, etc – which was previously inaccessible, required manual review, or human labeling. Fortunately, LLMs are remarkably good at working with unstructured data, which will unlock insights currently buried in a clinical note or across diverse data modalities. New AI models are also becoming more adept at integrating different data sources and schemas – whereas before it might take days or weeks to map two different data schemas with each other, now the same task might take minutes. As a result, systems can increasingly interact with each other, and data is less siloed.
The US healthcare system is too complex, regulated, and local to be transformed overnight, but as the saying goes “even the tallest mountain is climbed one step at a time." Together these factors are creating exciting, new opportunities for a long overdue revolution in US healthcare. Whether we end up calling it the third or the fourth industrial revolution for healthcare doesn’t matter – it will hopefully be a revolution where data, software, hardware, and AI finally result in meaningful improvements in healthcare productivity and quality. It’s time for healthcare to catch up – one step, one data record, one token at a time – to bend the cost curve on our $4.8 trillion annual healthcare spend.
Ethan Yeh, PhD is a Partner at Twine Ventures, investing in digital health and data/AI startups, and an Adjunct Professor in Electrical Computer Engineering at Rice University. He was previously a health economist at the World Bank and Head of Data Science at Stripe.
Many thanks to Chethan Bachireddy, Evonne Johnson, William Kang, Simon Lu, Vinod Mitta, and Harsha Thirumurthy for reviewing earlier drafts and feedback.
Lamb et al. 2024. “Generative AI in healthcare: Adoption trends and what’s next,” McKinsey & Company.
Gordon 2016. The Rise and Fall of American Growth. Princeton University Press. The first industrial revolution was based on the steam engine and took place from 1750-1840. The second industrial revolution was built on electricity and the internal combustion engine and lasted from 1870-1914. The third industrial revolution, based on digital technology, lasted from 1960-2004. There are many who argue that the fourth industrial revolution will be based on AI.
Sahni et al. 2017. “The IT Transformation Health Care Needs,” Harvard Business Review.
Atme et al. 2022. “Investing in the new era of value-based care,” McKinsey & Company.
Freed et al. 2024. “Medicare Advantage in 2024: Enrollment Update and Key Trends,” KFF.
Sahni et al. 2023. “The Potential Impact of Artificial Intelligence on Healthcare Spending,” NBER Working Paper 30857.