AI in Healthcare: Navigating Market Forces and End-User Benefits

The US healthcare system is often criticized for being one of the most expensive in the world, yet it fails to deliver proportionally better outcomes.

How best do we utilize the immense potential of Artificial intelligence (AI) in healthcare to tackle this. Is it by providing more care keeping healthcare spend constant? Or can we reduce the national spend on healthcare? In this post, we consider market forces that hinder cost reduction and emphasize the need to pass benefits to end-users (patients).

The High Cost of US Healthcare: A Look at Outcomes and AI’s Potential Solution

The US spends a significant portion of its GDP on healthcare, yet it ranks poorly in terms of outcomes compared to other developed countries. This paradox is often attributed to the complex interplay between various stakeholders – including payors, providers, and pharmaceutical companies. Each player has its own set of incentives, leading to a system that prioritizes volume over value.

AI has the potential to transform healthcare by streamlining clinical workflows, improving diagnosis accuracy, and enhancing patient outcomes. By leveraging AI-driven efficiency, we can potentially reduce treatment costs without compromising quality. Another aspect to be highlighted is the emerging utility of AI agents that can do many routine tasks very effectively reducing the costs of AI utilization. But these opportunities come with significant challenges.

There are numerous articles on this topic that get into specifics of how to leverage AI some even promising the moon (or mars), so I focus on enabling successful adoption and maximizing the value. I had also written a few invited articles at the end of 2024 on AI and digital adoption in enterprises in general, and also specifically in healthcare which I refer to at the end of this article***. You can also see how the company I lead, ReSurfX (https://resurfx.com) caters to these goals in innovative ways.

In one of those invited articles ‘Contrasting Needs Impacting Evolution and Adoption of Digital Transformation Driven Diagnostics and Health Solutions’ the first theme I brought up after highlighting the potential of AI was a call-to-action (CTA) to colleagues in the sector to pass on some of the cost benefits from AI to end-users (patients) as well. Passing benefits to end-users is crucial for success, because if patients will still have to pay the same amount for health services the AI implementations in healthcare is less likely be successful. The other topic I brought up is that AI could be a double-edged sword; while accelerating value it goes counter to many careful practices and ethics held high in the medical field.

While we are trying to create safeguards to cater to the specific needs of the medical field and the scientific culture in pharma and biotech, we are also discovering and highlighting many practices of the field that contribute to the high healthcare costs and poor outcomes that can be fixed.

Technological and other approaches that increases cost benefit to health system still do not offer that benefit to patients

Implementation of recent advances in tech and other care delivery innovation in the healthcare system do not always pass the benefit to the end-user (patients). Here are two examples from recent times:

When health systems started using Physician Assistants (PAs) and Nurse Practitioners (NPs) instead of physicians for patient care, reducing costs the patients do not benefit in most cases (e.g., they pay the same copays).

Another example is introduction of telehealth allowing patients to receive remote care. Patients still pay the same copay, whereas the health system gains time and effort such as taking vitals during visit.

Both these are simple examples where the effect of technology or another approach significantly reduced the cost of care, but in most cases the patient does not derive any cost benefit. Introduction of AI and other innovations to health system in ways that do not distribute value to population across the spectrum of demographic parameters is going to be less effective or cause other indirect problems.

Healthcare Market Construct: Understanding the Challenge

While the potential of AI in healthcare is exciting, we must not forget that market forces can be major obstacle to reducing costs. Changes that reduce top line revenue negatively impact the market value. Currently most business models for leveraging AI are built within this construct – which is sure to maintain status quo when it comes to overall healthcare spend and cost to the end-users.

To clarify, let us take a publicly traded company: when they announce their annual (or even quarterly) earnings, they will be penalized by investors (market) if they report we started using AI that reduced our costs and price, so our revenue is 80% this year and we expect to increase efficiency further next year but expect our revenue to become 60% though our margins are higher. This is also true for private entities or not-for-profits sharing annual statements with their board.

On the one hand we say in a country like the USA that we have the most expensive health system and spend more proportion of GDP and that is becoming untenable – but on the other our market system is not conducive to reducing cost to end users.

That does not make good sense if we want to reduce healthcare spend, does it? If that is the case, where do we go from here?

Rewiring Monetary Distribution in Healthcare: A Path Forward

At the outset this leaves us with no choice to reduce cost to patients from savings enabled by AI, other tech or innovation. Alternative is to frame it as, you are getting easier access to care or better outcome (however, better outcome is yet to be proven at this time). Another popular train of thought is to give end users access to more care for the same cost – so patients get more for their money.

To overcome the challenge, we propose a two-phase transition to distribute the benefits of AI and technology more equitably across stakeholders:

Phase-1: Initially we only introduce few key categories that are clear benefits such as tackling obesity, smoking and diabetes to everyone – benefitting both the patients and health system. To those in the lower economic rungs we increase care accessibility enabled by AI and other innovations with better outcomes which reduces cost to each of the stake holders. In this first wave we can add additional optional services that will be utilized only by economically well-off cohort – such as proactive continuous monitoring, warnings, suggesting in real-time wellness options etc. with guardrails. Even if each layer of prevention adds incremental benefits, many of them could in principle give an even higher chance for better and longer life than without those.

Phase-2: With time, we integrate more wellness into the system. which might even take a decade or more to integrate properly into the system. In that time the system by use of AI and other innovations will figure out the value of rolling out more categories of wellness benefits addressing critical factors like obesity and smoking rolled out in the first phase, as the system stands to gain.

Here I highlight two notes of caution from the past: (1) When I use the phrase system stands to gain – one should understand that payors standing to gain often leads to care providers at the losing end as it is today and vice-versa, hence it ends up being an endless cat-and-mouse game between them (and other major stakeholder such as pharmaceuticals as well). Thus, we need to factor in these aspects when we build into the rearrange monetary flow paradigm. (2) We should also be aware many tech foray into healthcare failed in the past – however, this time positive effect of AI seems more certain – but time will be the judge of it.

This second point above is in the way of fast adoption as the risk is still significant and business strategies and evaluations are still evolving. We know at this time that many front-end and administrative processes (i.e., operational aspects) can be efficiently handled by AI – but we haven’t fully worked out the value at enterprise levels yet, developed the guardrails specific for this sector or seen the economic benefits spreading across stakeholders.

Pharmaceutical Challenges and Opportunities

Till just over a decade ago the cost of drugs in overall healthcare spend was just over 10%. That changed with innovation in biologics class of drugs – currently we have many expensive drugs from this class. Another recent example is the glp-1 class of drugs that tackles two of the most important causes of many health problems, diabetes and obesity, which increase the number of people (patients) eligible for them. With such innovative medicines the pharmaceutical’s proportion of spend has significantly increased the payor portion of the costs. This problem needs unique set of solutions including novel pricing strategies, which are too deep for inclusion in this article.

Increasing Healthcare Access in Developing Countries with AI Advances

We should also discuss a scenario where increase in healthcare cost is not only advantageous but likely a necessity. In developing countries, a much larger proportion of the population is in poverty and has minimal access to healthcare. The inclusion of benefits from AI and other technologies leads to lower costs. This, in turn, increases access to healthcare for more people thus resulting in the overall health expenditure with outsized benefits. In that case, it is a very attractive solution to societal problems despite overall increase in that nation’s health spend, unlike in the health system of the US we discussed above. The impact of AI and other technologies that reduce costs significantly in this scenario may indirectly benefit healthcare stakeholders in developed countries who often are the primary source for innovative medicines and care options.

Summary: Leveraging AI to Maximize Societal Value in Healthcare

Emerging tech like AI has the potential to improve healthcare outcomes while reducing costs but it also comes with significant challenges. Here I proposed a two-phase transition to rewire monetary distribution across stakeholders in adopting AI-driven cost savings. This approach should help overcome the market construct challenge and distribute of benefits to end-user.

===================================================

***Related Articles and Expanded Versions:

AI for Enterprises. Part 1: Where Are We in Tackling the Popular Adage GIGO? [invited article in Massachusetts Technology Leadership Council (MTLC) discussing key challenges and solutions for AI and digital assets adoption and scaling]

Longer version of above article: AI for Enterprises. Part 1: Where are we in tackling the popular adage GIGO? [Extended version]

AI for Enterprises Part 2: Metrics for Matriculates? [invited article in Massachusetts Technology Leadership Council (MTLC) discussing key challenges and solutions for AI and digital assets adoption and scaling].

Longer version of above article: AI for Enterprises. Part 2: Metrics for Matriculates? [Extended version]