The Potential Impact of Generative AI in Healthcare
The US Healthcare System
The US healthcare system is unique and unlike any other healthcare system in the world predominantly due to its capitalistic structure whereby the main pillars that make up the system are for-profit companies and entities. In theory, applying the same capitalistic concepts that have paved the way for the American economy, from its commerce machine to the banking system core to the global financial structures, should yield similar positive outcomes. However, over the last 40–50 years the US healthcare system has emerged as one of America’s weakest industries. The US healthcare system lags far behind its peers while spending materially more than all of them. Since 1980, the United States has separated itself from the other OECD countries in terms of % spend of GDP, growing from 8% in 1980 to almost 20% today. This equates to roughly $4.3 trillion of the US economy, yet all this over expenditure has led to worse outcomes than all its peers. Across all major indicators, including life expectancy and avoidable mortalities, the US healthcare system materially trails its peers.
The US healthcare system is highly complex, and this paper is not meant to provide a historical overview as to the various elements that led to the current state of affairs. Instead, I’d like to explore the role of technology within healthcare, and the potential impact generative AI can have across the industry.
Emergence of Digital Health
The digital health ecosystem is still relatively young, with only over a decade of yearly VC funding exceeding $1B a year. Given the heavy spotlight on the overall healthcare industry, the COVID-19 pandemic significantly accelerated funding within Digital Health, reaching a yearly record of $30B in 2021. Excitement over the potential to disrupt several pain points across the industry grew including the adoption of real-world evidence and decentralized trials. Investments in digital products supporting disease treatment grew 2.6x YoY as the coverage pathway for prescription digital therapeutics widened. Healthcare marketplaces also experienced a 3.2x YoY funding growth, driven by upticks in D2C marketplaces, caregiver marketplaces and clinical job boards. With all this investment and excitement, came a certain degree of expectation of materialized positive change throughout the industry. However, time after time, innovation within healthcare was met with structural hurdles that prevented a full realization of its potential. Unlike in other industries that had recently been disrupted by technology such as the financial system with fintech and commerce with software and tech enabled services, healthcare startups continuously ran up against large structural hurdles that prevented large-scale adoption by the largest healthcare institutions, specifically on the software side.
The Promise of Generative AI
Traditional enterprise software solutions that have emerged over the last 10 years have struggled to penetrate healthcare enterprises particularly on the payer and provider side. Even though, as stated previously, healthcare represents 20% of the American economy, only one of the 100 largest public software companies is a healthcare company. Daisy Wolf and Vijay Pande, two healthcare partners at Andreesen Horowitz, explore why this has been the case and why it presents a tremendous opportunity for generative AI in the next wave of healthcare software innovation. Within health systems, there has been a reluctance to train burned-out staff on new systems. The opportunity lies in presenting these workers with a solution that is 10x better than any previous software tool that will take more of the admin work off their hands and give these workers time back.
Framework for the Gen AI Opportunity within Healthcare
The long term potential across both non-clinical and clinical applications could be transformative in Healthcare, however, I believe it is important to follow a structured framework to see where the current opportunity lies that can begin to make an impact presently across the three key healthcare pillars: Providers, Payors and Lifescience Companies. A tenure investor that has seen success in the intersection of B2B software and Healthcare, emergence capital, has developed a clear framework that showcases where they see the most opportunity right now. The three most important factors that consistently came up in their research were:
1. Degree of AI Unlock — What will be the magnitude of the impact this solution will have on day to day practical routine tasks done by staff working in health systems and insurers? Can this impact be quantified right away?
2. Urgency of customer pain — As we noted above, the healthcare industry is plagued with pain points, however some are more much relevant and painful than others. Hospital systems operate on low single digit operating margins (some are break even, and others even negative), and face increasingly higher variable costs coupled with higher clinical staff turnover.
3. Perceived Risk — There is broadly lower tolerance for applications that could be perceived as high risk and might slow adoption of some solutions. However, the higher risk solutions could end up having the biggest impact.
With this framework, and in-line with how other sophisticated healthcare investors are thinking, they were able to identify a series of non-clinical admin based tasks as the most attractive immediate opportunity that presented the highest unlock tackling the biggest pain points. It is no surprise that manual, labor intensive tasks within providers like note taking, billing & claims management and prior patient authorization all present attractive opportunities given the fairly archaic and manual operations many traditional healthcare systems and providers have.
Life science companies have invested more in technology, software and innovation and as a result I believe they face less urgent pain points which translates to a lesser AI unlock in the short term when compared with legacy providers. Nevertheless, pharma companies will continue to heavily invest in the continuous improvement of the clinical trial process that in itself continues to face several headwinds both from a regulatory and cost perspective. Trial recruitment and reporting specifically will yield plenty of opportunities for gen AI first builders.
