The Code Gets You Paid, but the Variant Gets You Treated: The AI Paradox in Healthcare

Not long ago, I read a vendor pitch for an AI documentation tool. The selling point was almost celebratory: the system listened to the visit, settled on a diagnosis, and produced a ready-to-submit ICD-10 code. Off it went. In the world of billing, that’s a finished transaction. But in the world of treating a patient, it’s nearly meaningless.

The gap between a code that satisfies a payer and the information a physician needs to treat a person is what nearly every healthcare AI tool is racing past. As we push these systems toward precision medicine and genomics, this gap stops being an annoyance and becomes a patient safety issue.

The Precision Paradox

ICD-10 and the coding systems around it were built to do one thing well: classify an encounter so a claim can be paid. The trouble starts when we ask them to stand in for clinical reality because a diagnostic category and a clinical diagnosis are not the same thing.

Consider how codes are chosen. An AI tool (or a hurried clinician) often selects whatever sits first in a list or whatever is specific enough to clear the claim. "Other chronic pulmonary disease" will get a COPD encounter paid, but it tells you nearly nothing about what the patient actually has. Similarly, "malignant neoplasm of brain, unspecified" will get the bill paid, but it doesn’t tell you the patient has a glioblastoma, which demands a completely different treatment path and prognosis than more favorable tumors sharing that same vague code.

For everyday internal medicine, this imprecision is already a problem. An unspecified code lands on the problem list and stays there, influencing decisions for years. But the stakes climb sharply when we move toward genomics and targeted therapy, where the treatment decision can hinge on a phenotype-the observable expression of a patient’s genetic makeup-and the precise variant underneath it.

Take Charcot-Marie-Tooth disease (CMT), for example. CMT is a hereditary neurological disorder with over 90 different genetic causes. Each variant can have vastly different clinical presentations and require distinct treatment approaches. If an AI tool or clinician selects a generic code like "hereditary motor and sensory neuropathy, unspecified," it could lead to misdiagnosis and inappropriate treatment.

Why It Matters

The implications of this precision gap are profound. In the realm of drug discovery and clinical trials, accurate genetic data is crucial. Clinical trials often require specific genetic markers to identify eligible participants. If the initial diagnosis is imprecise, patients may be excluded from potentially life-saving treatments or included inappropriately, skewing trial results.

The rise of precision medicine means that treatment options are increasingly tailored to individual genetic profiles. For instance, certain breast cancer subtypes respond better to targeted therapies than traditional chemotherapy. If a patient’s genetic variant is not accurately identified, they may receive less effective or even harmful treatments.

The human cost is also significant. Misdiagnoses and inappropriate treatments can lead to unnecessary suffering, prolonged illness, and higher healthcare costs. Patients may lose trust in the medical system, further complicating their care journey.

To bridge this gap, we need AI tools that go beyond generating billing codes and provide clinically meaningful insights. This requires a multidisciplinary approach involving clinicians, geneticists, data scientists, and policymakers. Collaboration is essential to develop standards for integrating genomic data into electronic health records (EHRs) and ensuring that AI systems are trained on comprehensive, high-quality datasets.

In the meantime, healthcare providers must remain vigilant. They should critically evaluate AI-generated codes and supplement them with detailed clinical notes and genetic testing when necessary. Patient advocacy groups can also play a crucial role by raising awareness and pushing for better data standards.

Ultimately, the goal is to ensure that the tools we use in healthcare enhance, rather than hinder, our ability to provide accurate and effective care. The code may get you paid, but it’s the variant that gets you treated-and sometimes, saved.