AI in cardiology: big market, real hurdles
The North American AI cardiology market is forecast to top $4.5 billion by 2033 as predictive tools and early‑detection algorithms gain traction, but experts caution that interoperability and governance lag behind adoption. New trends in healthcare data mining—FHIR interoperability and real‑time AI insights—promise to break silos, though oversight and risk management remain critical obstacles ( )
Analyst reports diverge on scale: one North America–focused forecast pegs the market at USD 4.52 billion by 2033 with a reported CAGR of about 21.4% (March 24, 2026 market report). (marksparksolutions.com) A separate industry estimate placed North American AI cardiology revenue at roughly USD 792.5 million in 2025 while projecting a global AI‑in‑cardiology market of about USD 14.83 billion by 2033. (grandviewresearch.com) Regulators are moving on adaptive software: the U.S. FDA issued final guidance on Predetermined Change Control Plans for AI/ML‑enabled device software functions on December 4, 2024, signaling a formal pathway for iterative algorithm updates. (akingump.com) Interoperability mandates are changing data access rules: the ONC’s Cures Act Final Rule and its API certification requirement obligate applicable health IT developers to support standardized FHIR APIs for patient data exchange. (federalregister.gov) EHRs and standards already support near‑real‑time decisioning via HL7 CDS Hooks and SMART on FHIR, with major vendors like Epic documenting CDS Hooks integrations for in‑workflow, synchronous clinical decision support. (build.fhir.org) Clinical products are advancing alongside standards: AliveCor received dual U.S. FDA clearances for its KAI 12L AI and Kardia 12L handheld ECG system in June 2024, reporting training on more than 1.75 million ECGs for the model. (alivecor.com) Peer‑review and policy literature flag governance shortfalls: a JAMA analysis describes health AI as occupying a “liminal regulatory space” across multiple agencies, and a recent npj Digital Medicine review (2026) synthesized 35 governance frameworks and identified seven critical domains that remain unevenly implemented. (jamanetwork.com) Technical integration still stalls clinical reuse: systematic reviews note persistent FHIR data‑modeling, semantic mapping, and scalability challenges that limit the ability to assemble consistent, multi‑source real‑time datasets for cardiac AI training and monitoring. (medinform.jmir.org)
