By Futurist Thomas Frey
The American healthcare system didn’t become incomprehensible by accident. Over decades, complexity became a feature, not a bug—a way to obscure costs, justify denials, and make true price comparison virtually impossible. But AI doesn’t get confused by complexity. It finds patterns in chaos. And what it’s revealing about healthcare and health insurance is going to fundamentally reshape an industry that represents nearly 20% of the U.S. economy.
More importantly, it’s going to save lives that the current system’s opacity has been quietly claiming for years.
The Price Opacity Problem
Here’s an experiment that reveals the absurdity: call ten hospitals in your city and ask what they charge for a knee replacement. You’ll get ten different answers—if you get answers at all. The prices might vary by 300-400% for the identical procedure, performed by similarly credentialed surgeons, with similar outcomes.
This wasn’t always true. In the 1960s, healthcare pricing was relatively straightforward. But as insurance became the primary payer, prices decoupled from anything resembling a market. Hospitals began charging based on what insurance would pay, insurance companies negotiated “discounts” from inflated starting points, and the whole system evolved into an elaborate kabuki theater where nobody pays the sticker price and nobody knows the real price.
AI is now analyzing millions of insurance claims, hospital bills, and payment records. What it’s revealing is stunning: the same procedure at the same hospital can be billed at wildly different rates depending on subtle coding variations, insurance types, and even the day of the week. In one analysis, researchers found that identical surgical procedures showed price variations of up to 800% within the same hospital system—not because costs varied, but because billing optimization algorithms found ways to maximize reimbursement.
Even more troubling: AI analysis reveals that hospitals in less affluent areas consistently charge higher prices for the same procedures than hospitals in wealthy areas, the exact opposite of what economic logic would predict. The pattern suggests that pricing is based not on cost or competition, but on extraction—charging what the system can extract from patients with fewer alternatives.
The Prior Authorization Nightmare
Prior authorization—the requirement that insurance companies approve treatments before they happen—was supposed to prevent unnecessary procedures. In practice, it’s become a bureaucratic weapon that delays care, denies treatment, and generates billions in administrative costs.
AI analysis of prior authorization patterns is revealing systematic denial strategies that have nothing to do with medical necessity. Insurance companies are shown to automatically deny certain categories of claims on first submission, knowing that approximately 60-70% of providers won’t appeal. Those that do appeal face a second automatic denial. Only on third appeal does a claim get actual human review—and by that point, maybe 15-20% of original claims remain in the system.
This isn’t hypothetical. AI has analyzed tens of millions of prior authorization requests and revealed that initial denial rates vary dramatically based not on the medical procedure but on the provider type, patient demographics, and even time of year. Small practices see higher denial rates than large hospital systems for identical requests. Medicaid patients face longer approval times than privately insured patients for the same treatments. Claims submitted near fiscal quarter-ends show different approval patterns than identical claims submitted mid-quarter.
The human cost is staggering. AI analysis of cancer treatment timing shows that prior authorization delays increase the average time from diagnosis to treatment start by 15-25 days. For aggressive cancers, those weeks can be the difference between curable and terminal. The system is literally killing people with paperwork—and it’s been invisible until now because no human could analyze enough individual cases to prove the pattern.
The Surprise Billing Ecosystem
You go to an in-network hospital for a planned surgery. You verify that your surgeon is in-network. Everything should be covered, right? Then you get a bill for $7,000 from the anesthesiologist—who works at the in-network hospital but personally isn’t in your network.
This practice, known as surprise billing, seemed random and unfortunate. AI analysis reveals it’s neither. It’s systematic and highly profitable.
By analyzing millions of medical bills, researchers have discovered that certain specialties—anesthesiologists, radiologists, pathologists, emergency physicians—are far more likely to be out-of-network than others, even when working at in-network facilities. These are specialties that patients can’t choose in advance and can’t refuse during treatment. The pattern is too consistent to be coincidental.
Further analysis reveals that private equity firms have been systematically buying up practices in these specialties specifically because they generate high rates of surprise bills. One analysis found that after PE acquisition, out-of-network billing by these practices increased by 60-90% within two years, with no corresponding change in services provided. The business model is explicitly based on exploiting patients’ lack of choice during medical emergencies.
AI has also revealed that hospital emergency departments in certain regions have systematically shifted to staffing models using out-of-network physicians, even when the hospitals themselves are in-network. The result: patients in medical emergencies have no choice but to incur out-of-network charges that can run to tens of thousands of dollars.
The Pharmacy Benefit Manager Shell Game
Pharmacy Benefit Managers (PBMs) were supposed to negotiate lower drug prices for insurance companies and patients. Instead, they’ve created one of healthcare’s most profitable and opaque business models.
Here’s how it works: PBMs negotiate “rebates” from pharmaceutical companies in exchange for favorable formulary placement. But these rebates often don’t flow to patients or even to insurance companies—they’re retained by the PBMs as revenue. Meanwhile, PBMs also own their own mail-order pharmacies, which they steer patients toward through coverage rules, allowing them to profit on both the negotiation and the dispensing.
AI analysis of drug pricing across the PBM ecosystem is revealing a system that often increases costs rather than reducing them. Researchers analyzing prescription data found that patients frequently pay more for medications through their insurance (via copays and deductibles) than they would if they paid cash at certain discount pharmacies. But patients are often explicitly prohibited by their insurance from using those cheaper options.
Even more damning: AI has identified numerous cases where PBMs classify drugs as “non-preferred” (requiring higher copays) not because cheaper alternatives exist, but because competing drugs offer higher rebates to the PBM. The financial incentives point toward higher-rebate drugs, not lower-cost drugs—directly contrary to what patients and insurance purchasers assume is happening.
In analyzing specialty pharmacy costs, AI revealed that PBMs often force patients to use PBM-owned specialty pharmacies that charge 40-60% more than independent specialty pharmacies for identical medications. Patients who try to use cheaper alternatives find their claims denied or their medications classified as “out of network” even when the pharmacy itself is in-network.
The Suppressed Treatment Problem
This is where The Awakening gets truly uncomfortable. AI analysis is beginning to reveal that certain effective treatments have been systematically deprioritized not because they don’t work, but because they’re unprofitable.
Consider off-patent drugs. Once a drug’s patent expires, generic versions become available at a fraction of the original cost. Standard economic theory says this is good—effective treatments become accessible. But AI analysis of treatment patterns reveals something darker: doctors prescribe newer, patented drugs even when older, cheaper drugs work equally well or better for certain conditions.
Why? AI has identified the pattern: pharmaceutical company representatives visit doctors’ offices, sponsor continuing education, fund research, and provide “consulting” opportunities. Doctors who interact more frequently with pharmaceutical representatives show measurably different prescribing patterns than those who don’t—they’re significantly more likely to prescribe newer, expensive drugs even when clinical guidelines don’t support the choice.
This isn’t bribery in the legal sense. It’s the systematic cultivation of prescribing habits through relationship and influence. And it’s been invisible until AI could analyze millions of prescriptions alongside data on pharmaceutical company payments to physicians.
More troubling still: AI analysis of clinical trial data has revealed numerous studies of promising treatments that simply stopped—not because the treatments failed, but because pharmaceutical companies declined to fund Phase 3 trials for drugs that couldn’t be patented or wouldn’t generate sufficient return. The treatments weren’t “suppressed” by conspiracy; they were simply abandoned by a system that only pursues profitable research.
The Diagnostic Code Inflation
Medical billing uses diagnostic codes to describe patient conditions and justify treatments. In theory, this creates consistency and accountability. In practice, it’s created a game of code inflation that increases costs without improving care.
AI analysis of diagnostic coding patterns reveals systematic “upcoding”—describing patient conditions in ways that justify higher reimbursement without changing actual treatment. A patient with slightly elevated blood pressure gets coded as having “hypertensive crisis.” A minor scratch becomes a “complex wound.” Normal aging becomes “failure to thrive.”
The pattern is measurable: AI has found that the average number of diagnostic codes per patient visit has increased approximately 40% over the past decade, while the actual treatments provided have remained largely static. Hospitals and practices aren’t seeing sicker patients—they’re describing patients in ways that generate higher payments.
Even more specifically, AI has identified that certain hospitals show dramatic increases in high-severity diagnoses on the last day of each fiscal quarter—precisely when hitting revenue targets matters most. The patients aren’t actually sicker; the coding is more aggressive.
This isn’t just about money. Upcoding creates false medical records that follow patients through the system. A patient coded as having “chronic heart failure” for billing purposes may later be denied insurance or charged higher premiums based on that record, even if the diagnosis was a billing optimization rather than a clinical reality.
The Cure Suppression Question
This is the most controversial aspect of The Awakening in healthcare, but AI analysis is forcing us to confront it: are there cures that have been sidelined because treating ongoing conditions is more profitable than curing them?
The business model of chronic disease management is straightforward: a patient with Type 2 diabetes might spend $10,000-15,000 per year on medications, testing supplies, and monitoring for decades. A cure would eliminate that revenue stream. The financial incentive clearly points toward management, not cure.
AI analysis of research funding patterns is revealing troubling trends. Studies focused on diabetes management receive far more pharmaceutical funding than studies focused on diabetes reversal through lifestyle intervention, even though clinical data shows that significant percentages of Type 2 diabetes cases can be reversed through intensive diet and exercise programs. The less profitable research simply attracts less funding.
Similarly, analysis of hepatitis C treatment reveals an instructive case study. For years, hepatitis C was managed with expensive, long-term treatments that controlled but didn’t eliminate the virus. When truly curative treatments finally emerged, pharmaceutical companies priced them at $84,000-94,000 per patient—not because the drugs cost that much to produce, but because the companies calculated lifetime treatment costs for the chronic management alternative and priced just below it.
AI analysis of patent strategies reveals another pattern: pharmaceutical companies often patent dozens of minor variations of successful drugs, creating “patent thickets” that can extend monopoly pricing for decades beyond the original patent. These variations rarely represent meaningful therapeutic improvements—they’re legal strategies to prevent generic competition and maintain pricing power.
The Medical Imaging Markup
Medical imaging—X-rays, MRIs, CT scans—shows some of the most dramatic pricing inconsistencies in healthcare. AI analysis of imaging costs reveals that identical scans can vary by 1,000% or more depending on where they’re performed.
An MRI at a hospital might cost $3,000-4,000. The identical scan at an independent imaging center costs $400-600. The equipment is the same, the technicians have equivalent training, the radiologist reading the scan might be the same person. The difference is pure markup.
AI has revealed that hospitals systematically price imaging higher because they can—insurance negotiated rates are based on hospital charge masters that bear no relationship to actual costs. Independent imaging centers, which must compete on price for self-pay patients, reveal what the scans actually cost to produce.
Even more specifically, AI analysis shows that hospital-owned imaging centers charge significantly more than independent centers, even when they’re physically separate from the hospital. The ownership connection alone drives pricing up 200-300%, with no difference in service, equipment, or outcomes.
The Administrative Bloat Problem
In 1970, the United States had approximately 3.2 healthcare administrators for every physician. Today, that ratio is approximately 10 to 1. AI analysis of healthcare employment patterns reveals where all these administrators are going: navigating the complexity that the system itself created.
Insurance companies employ tens of thousands of people to deny claims. Healthcare providers employ tens of thousands to appeal those denials. Both sides employ teams to negotiate prices that will never actually be paid. Hospitals employ specialists who do nothing but optimize billing codes. Physicians spend 2-3 hours on administrative tasks for every hour of patient care.
AI analysis estimates that administrative costs now represent approximately 25-30% of total healthcare spending—roughly $800 billion to $1 trillion per year. This isn’t the cost of care. It’s the cost of the complexity we’ve built around care.
Comparisons with other developed countries are stark. Nations with simpler payment systems show administrative costs of 12-16% of healthcare spending. The difference—$400-600 billion per year in the U.S. alone—funds zero additional care. It’s pure waste.
What The Awakening Means for Healthcare
The patterns AI is revealing aren’t secrets in the traditional sense. Healthcare insiders have known about many of these issues for years. But knowing and proving are different things. AI provides the proof at a scale that can’t be dismissed or explained away.
We’re approaching an inflection point. Once these patterns are quantifiable and undeniable, the system has three options: reform, collapse, or evolve into something that makes the current complexity look simple by comparison.
The early signs suggest we’ll see elements of all three. Some organizations are already embracing transparency, publishing real prices, and simplifying their processes. Others are fighting to preserve opacity through regulatory capture and legislative complexity. Still others are building AI systems to navigate and exploit the complexity even more efficiently than before.
But here’s what’s different this time: patients are getting access to the same AI tools. Apps that compare imaging costs, identify surprise billing before it happens, flag upcoding, and recommend cheaper treatment alternatives are already emerging. The information asymmetry that healthcare has relied on for decades is eroding.
The complexity was supposed to make healthcare too complicated for ordinary people to understand. Instead, it made healthcare too complicated for ordinary people—but perfectly suited for AI analysis.
The awakening in healthcare isn’t just about exposing problems. It’s about revealing that solutions have existed all along, buried beneath layers of profitable complexity. AI doesn’t just find the problems—it points directly at the simpler, cheaper, more effective alternatives that have been hiding in plain sight.
In our next column: Education and Credentialing—The Degree Illusion.
Related Articles:
Health Affairs – Price Variation in Health Care: Insights from Health Care Markets
JAMA – Costs of Health Care Administration in the United States and Canada
The New England Journal of Medicine – Prior Authorization—A Barrier to Patient Care
