The Code Gets You Paid -- The Variant Gets You Treated

A key gap in healthcare AI is that billing-ready ICD-10 codes don’t necessarily match the clinical reality needed to treat a patient, the article argues. In coding systems, diagnostic categories are designed to classify encounters so claims get paid, but AI tools can default to vague or list-first entries that clear reimbursement without capturing the true condition. The piece says that imprecision can linger on problem lists for years, but becomes riskier as medicine moves toward genomics and targeted therapy, where treatment may depend on the exact genetic variant. It cites Charcot-Marie-Tooth disease: identifying the right variant is portrayed as critical now that targeted therapies are in development.