Mr. AI Goes to Washington
Authors: Manish Shah, text-davinci-003, ChatGPT Dec 15 Version
Like many, I’ve become enamored with the latest developments of AGI technologies such as Stable Diffusion, DALL-E, Midjourney, and ChatGPT. These technologies unlock a new way of “producing information goods”. Much like how machines (and machine tooling) unlocked a new way of producing industrial goods (cars, planes, tractors, etc) in the last 75 years, these new assistive AI tools make it possible for more people to translate thoughts into information-based goods. As a case study to demonstrate this potential, I’m using ChatGPT to be a co-author on a position paper for an immigration policy. All research was performed by the AI and much of the framework for the logic was developed by the AI. I supplied the initial ideas (ie. the prompts) and asked additional questions (ie. more prompts) to enhance the discussion and add more substance. I’m also constructing paper and adding the emotional touches. Originally I thought writing this position paper would take me several months to perform the research (on Google) and crafting the main argument. However, with AI, I was able to go from idea to paper in a couple days (after excluding my own procrastination of course).
Can we fill a major hole in our labor supply and reach a rare bipartisan agreement on a contentious issue at a time when congress is diametrically opposed more than ever? Call me crazy or maybe it’s new year optimism but I think there is an opportunity within immigration policy that is staring at us in plain view.
The Physician Shortage Problem
It is becoming common knowledge that the US is in the midst of a severe physician shortage. As is the case with climate change, this problem is going to cause significant pain and erode quality of life over generations. And it isn’t just the future that is in danger, we are suffering now with poor health outcomes and spiraling healthcare costs.
We’ve known about labor shortages for a while, long before the pandemic. However, in my experience as a digital health founder, I believe we (myself included) have been avoiding tackling this issue head on. We’ve been burying our heads in the sand by focusing on technology as the solution.?
Or, “with remote monitoring, we can avoid having patients come to a hospital or clinic altogether.”?
And on and on the narratives go.
These aren’t wrong, they do help reduce the impact of a short labor supply. But not enough and often these solutions come with negative externalities that exacerbate the issue of staffing shortages. Even though we’ve invested in lots of new technologies, they don’t integrate with each other and no one technology does everything. So physicians are dealing with multiple systems which are infuriating and increase their risk of burnout.
Furthermore, the companies behind these solutions are realizing they need to become care providers themselves which increases the need for more medical professionals with appropriate licensure to render care to patients.
By the Numbers
Technology will not fill the hole in our healthcare labor shortages and I believe we need to approach this solution head on. According to the Association of American Medical Colleges, there is a projected physician shortage of up to 122,000 physicians by 2032 and according to the American Association of Colleges of Nursing, there is a projected shortage of up to 200,000 nurses in the US by 2025.
The shortage is based on a few macro trends. First, with the aging population in the US,? more people are entering the stage of life where medical need is highest. The current system cannot handle a rising tide of elderly individuals with more chronic physical and mental health issues that require treatment or management over several years. We’ve designed our health systems over the last 100 years to handle acute health care needs. Other factors are contributing to the shortage is how much we are constraining or limiting our supply of medically trained professionals. According to the American Medical Association, approximately 4,500 physicians leave the profession each year due to burnout or retirement.? According to the American Association of Colleges of Nursing, approximately 8,000 nurses leave the profession each year due to burnout or retirement. And we are limiting the number of new professionals can enter the workforce: According to the American Medical Association, approximately 28,000 new physicians enter the profession each year and According to the American Association of Colleges of Nursing, approximately 70,000 new nurses enter the profession each year.
Additionally, to constrain supply further, we are making it more and more difficult for students to make the decision to go into the medical field with low levels of job satisfaction, high levels or burnout, and high levels of medical school debt - according to data from the AAMC, it is likely that the average debt for medical school graduates will be in the range of $200,000 to $225,000, depending on the individual's circumstances.
It is also worth noting that the distribution of physicians is not even across the US. Rural locations have far fewer physicians than areas with high population density.
Potential Solutions
1. Immigration Policy (Federal)
My AI colleague and I believe a small, focused immigration reform policy could help fill some (not all) of the hole in the labor force around healthcare services. The simple idea is that we make it easier for individuals trained outside the US in the medical profession to obtain visas and licenses to practice here in the US.?
Immigration is one of the most contentious issues in the country. Democrats and Republicans would all agree that we need broad immigration reform, but the details of that broad reform is where it is difficult to reach consensus and any such reform is DOA. This proposal is meant to be a small, focused immigration reform option that both parties could find common ground, the strength of our labor supply in healthcare and the ability to take care of our aging population.
There are two paths to consider, immigrant (path to citizenship) and nonimmigrant (path to productivity) methods for gaining the ability to live and work inside the US for a longer period of time.
According to the US Department of State, in 2019 there were only 8,814 immigrant visas issued to applicants who listed a medical profession as their occupation.??
Among the nonimmigrant visa pathways, the main method is the US H1-B program, visas given to employees as long as they are “sponsored” by a US corporation. One of the main problems with the H1-B visa is the US government sets an annual limit on the number of H1-B visas available - in 2021, the limit was 85,000…total. This pool is split between all industries regardless of what professionals would serve a higher function in the current (or future) economy. And there are many organizations that hoard H1-B visas for their contract labor employees. Other nonimmigrant visa types (J-1, O-1, and TN) may stand in for H1-B visas, but those also have time restrictions and work restrictions.
领英推荐
With an annual real deficit of 95,000 physicians in 2023 (after accounting for burnout, retirement, and new licensed physicians exiting and entering the workforce), there is a lot we can improve by adjusting our immigration policy to attract health care talent abroad to serve our population.
For both immigrant and nonimmigrant visa paths, we need to reduce the time it takes to apply for a work-permitting visa and the time it takes to process those applications. This process can take between several weeks to several months - a horribly slow process given much of it is work that can be automated. We should be able to bring this down to HOURS, not months and weeks.
Additionally, we should consider increasing the total number of H1-B visas given out per year and add prioritization for skills that fit the needs of our economy. We can also increase the number of EB-1 visas (currently 40,000 per year) and consider adding new designations for medical professionals.?
This is a much simpler policy to propose and should be easier to pass than a broad-based immigration policy. This should especially be championed by states with large rural and aging populations where the pain of physicians shortages are felt more.
2. Labor Policy (State)
Unfortunately, the supply issue can’t be solved solely with a stroke of the pen at the federal level. We need to make other improvements as well. Thankfully, these improvements will apply to domestically trained medical professionals as well.
We need to adjust the licensing and credentialing processes at the state level. Medical boards in each state set the guidelines for obtaining licensure to practice medicine within a state’s borders.??
In order for a doctor trained in India to practice medicine in California, they must obtain a license from the Medical Board of California and pass the United States Medical Licensing Examination (USMLE). Additionally, they may be required to complete a postgraduate medical training program, such as a residency program, in order to gain certification to practice in the state.?
A residency program (if there is availability) usually takes 3-7 years to complete. This is after the person has already completed training in a foreign program and already has much of the training necessary to begin serving patients with low acuity needs.?
States with greater populations of aging individuals and large rural areas (where medical options are limited) could make it easier for immigrant professionals to obtain licenses and credentials necessary to treat patients. Additionally, states can make it easier for medical professionals with different license types to serve part-time or “incident to” a physician where there isn’t a direct risk to the patient. This is often the case in low acuity settings and chronic disease management.
3. Funding for Medical Schools
Finally, in order to truly address the physician shortages in the US, we need to address the labor pipeline. We simply do not train enough new medical professionals in the US at all skill levels.?
According to the Association of American Medical Colleges, as of 2020 there were approximately 1,700 accredited academic medical institutions (AMI) in the US offering medical residency, fellowship, and advanced training programs. These institutions accept 33,000 students into medical residency programs each year, roughly 18-19 pupils per AMI per year. Only 28,000 enter the medical profession after completing a post-graduate residency program.
Just to do some basic math, we estimate a physician deficit of -95,000 physicians per year and we are only producing 23.5K new each year or ~25% of the deficit. You can see that the hole is just getting bigger each year.?
?(This is just the physician deficit, it does not include the deficit of other medical professionals such as nurse practitioners, physician assistants, and more.)
We need to double or triple our number of academic medical institutions otherwise we will continue to sink into this hole.?
It is imperative that we grow our medical training infrastructure in the US and update the curriculum to teach modern technology as a complement to existing protocols used to diagnose and treat patients. This will not only improve the total number of trained medical professionals, it will arm each with modern tools necessary to deliver high quality care to more people. We would never train a soldier with muskets and gunpowder and then send them into a war in 2022. Yet this is what we are ok with doing in our medical education system.
Conclusion
As a technology entrepreneur, the last thing I imagined myself advocating for was a new policy to solve a problem. Generally, I think entrepreneurs build solutions and markets distribute solutions to problems. Unless the market’s objective function is improved by exploiting the problem itself. This is what is happening today and by many players. We have surging costs from travel nursing that exploit the limited supply of nurse staffing (and poor wages for nurses). We have hospital systems (with tax exempt statuses) raising fees due to high labor costs and change management costs when implementing new technologies that somehow take DECADES to implement - the very same technology that is supposed to reduce the burden for physicians. We have health insurers raising premiums due to higher medical costs. We have employers depressing real wages because of high medical insurance benefit costs and passing more costs to employees via high deductible plans.
Across the board everyone (except the patients and the providers) is finding a way to exploit this environment for their own gain.?
As an entrepreneur, this is when you decide if it’s a game worth playing. In my opinion, this game is not conducive to innovation and delivering high quality care at lower costs. Instead, I’m looking for alternative ways to leverage technology to change the game. With the breakthroughs in AI that are reaching the public now (and will be coming), there is a way to do just that. Technology is often the catalyst to change the game, here I think there is an opportunity to speed up the process by which we change the rules for all.?
Appendix
There are also other options worth exploring, mainly around making healthcare attractive once again for medical professionals. Putting the altruistic reasons aside, it is not just as attractive financially for people to become medical professionals as it used to be. There are yearly rate cuts, difficult to navigate contracting with payers, and all around SLOW and PAINFUL payment systems for medical professionals. We should consider faster payment rails for medical professionals (same day payment for hourly shifts), factoring on submitted claims, and attractive lines of credit for medical professionals with high MIPS scores.