For decades, the medical industry has operated on a “hunch and react” model. You feel a symptom, a doctor makes a guess, and you wait to see if the treatment sticks. We’ve spent the last fifty years looking for a “replacement” for the human physician—a Silicon Valley savior to automate the exam room.
But as we move deeper into 2026, the real revolution isn’t a replacement; it’s a partnership found in the most mundane place imaginable: your billing codes.
In a recent conversation, Joshua Resnikoff, CEO of Sunstone Health and a former Harvard researcher, revealed that the key to solving the “Diagnostic Odyssey”—the grueling seven-year average wait for a rare disease diagnosis—isn’t hidden in your DNA alone. It’s buried in the paper trail of every doctor’s visit, lab test, and insurance claim you’ve ever filed.
The Hidden Language of the “Odyssey”
Most people view insurance claims as financial nuisances. To an AI, however, they are a longitudinal map of human suffering. While a doctor might only have 11 minutes to review your case, a machine learning model can look back across years of “anonymous” billing codes to find patterns of expression that a human would miss.
-
The Problem: The healthcare system is built to react to patients it can already “box” into a diagnosis.
-
The Consequence: Families spend years bouncing between specialists, only to be told their child’s condition is “just a milestone delay” or “not terminal”.
-
The Solution: Sunstone Health uses AI to scan these claims, identifying “high-risk” patterns for rare genetic disorders like epilepsy or autism.
The “11-Minute Wall” and the AI Bridge
We often blame doctors for missing the “rare” stuff, but Resnikoff points out that the system is the bottleneck. A physician in a Minute Clinic doesn’t have the time to read six years of notes from five different specialists.
AI acts as a bridge, not a replacement. It doesn’t diagnose; it flags. It says to the doctor: “Based on this patient’s history, they look like they have a genetic basis for their symptoms. Let’s move to a diagnostic test now, rather than in 2031”. This process collapses the diagnostic timeline from seven years to just 12 weeks.
The Privacy Divorce: Protecting the Patient
A common fear in 2026 is that AI will allow employers to “weed out” expensive employees. Resnikoff’s model introduces a critical “divorce” between the patient and the health plan.
-
Tokenized Data: Data is processed as strings of numbers, keeping the patient’s identity hidden from the AI until the family opts in.
-
The Employer Firewall: Employers pay for the service as a benefit, but they are legally “protected from knowing” who the high-risk patients are.
-
Data Sovereignty: Unlike many university systems, Sunstone ensures the patient owns their genetic data. You can download it, delete it, or take it with you—the “right to be forgotten” is a core tenet.
The Unexplored Angle: The Shift to “Bio-Intelligence” Benefits
While the podcast focuses on rare diseases, there is a larger trend emerging: the shift from Health Insurance to Bio-Intelligence.
In the past, an employer provided a safety net for when you got sick. Today, we are seeing the rise of the “Precision Medicine Benefit”. This isn’t just about finding rare diseases; it’s about Pharmacogenomics (PGX)—understanding how your body responds to caffeine, antidepressants, or heart medication before you ever take a pill. We are moving toward a world where your employer doesn’t just pay for your “guinea pig” phase of trial-and-error medicine but funds the intelligence required to get the treatment right the first time.
“The waiting part is the most expensive part in healthcare. By the time a problem actually presents, you’re trying to correct years worth of disease exacerbation.” — Joshua Resnikoff
Final Thought: The Moral Compass of the Machine
We often worry if AI will be used to deny claims or “do evil”. And while that “dark side” exists, the antidote is mission-driven technology that prioritizes clinical outcomes over actuarial spreadsheets. If we can use fire to warm a home or burn it down, we can use AI to alienate patients or finally give them the answers they’ve been seeking for seven years.
The choice isn’t between humans and machines. It’s between a system that waits for you to break and one that sees you coming.
