A deeper dive on the private healthcare AI ecosystem
Healthcare is one of the first major industries to incorporate AI into its everyday applications.
This statement in itself often leaves people surprised.
After decades of cementing its infamous track record of slow tech adoption, healthcare is now advantageously positioned in the world of AI. For one, there is minimal reliance on old software and associated sunk cost biases, allowing AI to have a dramatic leapfrog effect. While others grapple with the hundreds of billions of dollars spent on enterprise software and complicated incumbent systems that need to be overridden, healthcare can harness the technology with a fresh slate.
As venture investors, we have the opportunity to see these solutions emerging in real-time. In fact, over the past year, more than half of the companies that have come across our desk mentioned AI in their story, albeit to varying extents. This may be an underestimation; a recent Deloitte survey indicated ~75% of healthcare business executives are testing or planning to implement AI in their operations. [1]
With our unique perspective comes many frequently-asked-questions on the topic. This piece aims to provide answers to the top five:
Can healthcare leverage existing solutions, or does it require a specialist AI?
How would you categorize young healthcare companies harnessing AI?
Who are the private players today and how has the market evolved?
Across the entire universe, who wins?
What excites you most in this space?
Can healthcare leverage existing solutions, or does it require a specialist AI?
Our take is two-fold:
For administrative applications, including insurance workflows, manufacturing, hospital / physician operations, healthcare can leverage existing large language models and horizontal infrastructure, leaving only the application-layer to be specific to the industry.
For more clinical use cases, the story is different. For example, OpenAI’s GPT-4 was trained on 1.75 trillion parameters. Achieving a similar scale in healthcare would require data from 20 billion patients—an impossible feat given the global population less than 8 billion. These applications are also dominated by vast amounts of idiosyncratic data – whether it be unique patient genetic markers, personal medical histories or individual responses to specific treatments. To add, strict regulatory considerations, such as HIPAA compliance laws, need to be made along the way.
With all of this in mind, successful execution of AI adoption is most likely to rely on a foundation of tailored and specialized solutions. Entrepreneurs and investors with deep industry experience recognize this opportunity: According to a recent report published by SVB, the pace of private AI deals in healthcare is even growing ~2x faster than in tech. [2]
How would you categorize the young healthcare companies harnessing AI?
As our team focuses on life science companies, and services and health tech companies selling into life sciences, most, if not all, solutions that we come across can be slotted into three categories:
Medical devices & diagnostics –? from MRIs to robotics, to smart wearable technologies
Perhaps surprisingly, the FDA has been an early believer in the power of AI embedded in medical devices. A structured approach was developed years ago to evaluate and clear products considered as software-as-a-medical device (SaMD). Just over the past five years, >500 AI-enabled medical devices have been approved for use in patients.
For example, models can be trained to recognize anatomies and anomalies in an MRI image and sensors implanted in your knee or hip post replacement can detect motion and assess recovery progress. In fact, by the age of 60, the average person has 1.8 implants; we don’t see a world where AI isn’t integrated to continuously monitor and share live data with physicians.
The Centers for Disease Control and Prevention (CDC) also estimates that training the next generation of doctors with tools that effectively monitor and manage patients’ chronic illnesses can save up to 850,000 lives in the U.S. each year, with an annual reduction in healthcare spending up to $270B. (Not to mention, by the time matriculating medical students finish their fellowship programs in 10 years, generative AI is expected to be 1,000x more powerful.) [3]
Drug discovery & development – from protein mapping to clinical trial design
Biotech and pharma companies are quickly realizing that the abundance of data at their fingertips can finally be used in highly productive ways, such as to unravel chemical and biological pathways underlying disease states, identify dosing regiments to best reduce drug side effects, or optimize patient selection for clinical trial enrollment.
There have been a few recent milestones worth noting in this space:
Given the billions of dollars (and years) it takes to successfully develop drugs, the entire end-to-end cycle, from drug discovery to approval, to commercialization, is highly vulnerable to disruption. Said differently, AI can allow biotech companies to dramatically increase their “shots on goal,” saving money and time.
Data & language-based services – from clinical-note taking to telemedicine
A striking one-third of the world’s total volume of data is generated in healthcare. ~96% of hospitals use electronic health records (EHRs) to store patient, yet a mere ~3% can be used due to fragmentation and siloed systems. [5] To give you a sense, this volume of information is equivalent to all published works in every language in recorded history. AI’s ability to make sense of vast amounts of relatively unstructured patient and logistical data will finally unlock its potential.
It’s also estimated that the industry wastes roughly $800 billion annually, or 25% of total spend, largely due to human administrative overhead. [6] Simply relieving providers of their IT burden can cut down on costs and save critical hours each week.
In the realm of language-based health AI solutions, there are also companies that allow patients to virtually chat with an LLM “physician.” One example is Hippocratic AI, powered by Polaris, the first safety-focused LLM for real-time patient-AI healthcare conversations. The one-trillion parameter constellation system is focused on driving safe and engaging conversations. Upon testing, Polaris performed significantly better than larger general-purpose LLMs, such as GPT-4. [7]
Who are the private players today and how has the market evolved?
The below bubble chart shows a “market map” of private healthcare companies in the U.S. tagged as using AI by Pitchbook.
Each bubble represents one company (date of its last fundraising round vs. post-money valuation).
领英推荐
Removing the seven outliers to get a closer look:
Altogether, here’s what we noticed: [8]
By count, there are 442 VC-backed healthcare companies in the U.S. tagged with AI. Health tech was the most plentiful (55%), followed by medical devices (23%).
A record 1-in-4 healthcare dollars are going to AI. Since 2016, over $20B of VC capital has been deployed into healthcare-AI companies.
The average biotech-AI deal is at least 2x more expensive than any other subsector. (Though we have to call out Xaira Therapeutics, which just closed a $1B Series A this year.)
Medical devices so far proved to have the strongest average return-on-invested-capital (ROIC). This could be attributed to relatively shorter product development cycles, established FDA pathways, higher margins or the inherent scalability of the business.
Since 2016, 147 private healthcare-AI companies have exited (~75% via M&A, ~25% IPO). We believe this is just the beginning.
Across the entire universe, who wins?
While competition may be crowded for young healthcare-AI companies, much like any new tech entering the market, it’s inevitable that a large share will fail and many others consolidate. The “winners” will likely:
What excites you most in this space?
Seeing tangible change in healthcare may take time, much like it did for other major tech disruptors such as the internet, cloud computing, and mobile technology. However, the potential benefits of AI in healthcare are already generating immense excitement. At the very top of our list: Patients are poised to benefit tremendously from this imminent technological revolution.
Hospital operating rooms are on the brink of transformation with the integration of smaller and more advanced robotics. Physicians will soon have the capability to predict patient ailments before symptoms even appear, enabling proactive and preventive care. Personalized treatment plans will be crafted with pinpoint accuracy, revolutionizing patient outcomes. And the days of the doctors’ office fax machine will finally be behind us. These advancements merely scratch the surface of what AI can achieve, and it’s all happening now.
For more of our team’s take, check out our Managing Partner Anya Schiess’ podcast appearance here!
[1] At the nexus of healthcare and Generative AI, Deloitte AI Institute, 2023.
[2] 2024 shaping up to be a big year for healthcare AI companies. But some investors remain cautious, Fierce Biotech, June 2024.
[3] “Medical Education Needs Radical Reform”, Forbes, July 2024.
[4] AI in Drug Development: From Hype to the Clinic and Beyond, Cowen Research, November 2023.
[5] World Economic Forum
[6] Journal of American Medical Association, 2019
[7] Polaris: A Safety-focused LLM Constellation Architecture for Healthcare, arXiv, March 2024.
[8] Pitchbook, August 2024.
NOT FOR RETAIL DISTRIBUTION: This communication has been prepared exclusively for institutional, wholesale, professional clients and qualified investors only, as defined by local laws and regulations.
The views contained herein are not to be taken as advice or a recommendation to buy or sell any investment in any jurisdiction, nor is it a commitment from J.P. Morgan Asset Management or any of its subsidiaries to participate in any of the transactions mentioned herein. Any forecasts, figures, opinions or investment techniques and strategies set out are for information purposes only, based on certain assumptions and current market conditions and are subject to change without prior notice. All information presented herein is considered to be accurate at the time of production. This material does not contain sufficient information to support an investment decision and it should not be relied upon by you in evaluating the merits of investing in any securities or products.?In addition, users should make an independent assessment of the legal, regulatory, tax, credit and accounting implications and determine, together with their own financial professional, if any investment mentioned herein is believed to be appropriate to their personal goals. Investors should ensure that they obtain all available relevant information before making any investment. It should be noted that investment involves risks, the value of investments and the income from them may fluctuate in accordance with market conditions and taxation agreements and investors may not get back the full amount invested. Both past performance and yields are not reliable indicators of current and future results.
J.P. Morgan Asset Management is the brand for the asset management business of JPMorganChase. and its affiliates worldwide.
Copyright 2024 JPMorganChase. All rights reserved.
Material ID: 095r242008174152 – For Global Use
Healthcare Investor | MPH
6 个月Great read that does a good job of covering a very broad subject, succinctly. Although I'm not sure I agree that there's minimal sunk cost in the Healthcare tech stack. On the provider and payer side of things at least, there's significant tech debt in EHR and Core Admin systems respectively. Certainly could see fast adoption of point solutions in specific areas, but a lot of great new tech waves have crashed and died against 10 year old instances of Epic (to everyone's detriment). The newcomers to the sector e.g. One Medical, were able to truly build from scratch and have delivered a much better digital patient experience as a result. I'm hoping that the incumbents will get on the bandwagon soon.
Founder & President
6 个月Thank you, Olivia, for your superbly researched and presented deep dive into this very important topic. As a side note, I found the FDA's SaMD policy guidance document to be very helpful, but in the midst of developing an AI-powered app/device that we hope will be first-in-class, we needed some additional clarification. I reached out to the FDA's Digital Health Center of Excellence ([email protected]), and I must say, the exchange of communication was very productive. Side note #2: I'm not ready to call the recommendations in this JAMA Viewpoint article a cause for concern, but something to keep an eye on... https://jamanetwork.com/journals/jama/fullarticle/2822176?guestAccessKey=707a7cdd-19e4-44aa-9c12-2998c907109e&utm
Co-Founder of Altrosyn and DIrector at CDTECH | Inventor | Manufacturer
6 个月It's true that healthcare has often been slow to adopt new technologies. The shift towards AI is exciting, but it also presents unique challenges. How are you navigating the balance between innovation and patient safety in your work?