AI data‑readiness gap

Most healthcare systems still cannot feed their data directly into artificial intelligence tools: a new April 20 analysis said less than 20% of enterprise healthcare data is ready for AI without major cleanup. (hitconsultant.net) Zach Evans, chief technology officer at Xsolis, wrote that AI projects in health systems often stall because records sit in separate billing, clinical, lab, imaging, and claims systems that were never built to work as one dataset. He said the bottleneck is not usually the model itself but the fragmented data underneath it. (hitconsultant.net) “AI-ready” data means information is standardized, linked to the right patient and encounter, and governed so a model can use it without guessing what a field means or whether it is complete. Gartner said in February that organizations will abandon 60% of AI projects through 2026 if those projects are not backed by AI-ready data. (hitconsultant.net) (gartner.com) The timing matters because healthcare is spending heavily on AI while its data foundation remains uneven. Menlo Ventures said healthcare captured about $1.5 billion in vertical AI spending in 2025, up from $450 million in 2024 and ahead of every other industry segment. (menlovc.com) (hitconsultant.net) Federal data released the same day showed the gap is wider in behavioral health, where digital records are common but data sharing is not. The Office of the National Coordinator for Health Information Technology said 68% of substance-use and mental-health treatment facilities used an electronic health record in 2024, but only 19% participated in a Health Information Exchange. (healthit.gov) (hitconsultant.net) That means a patient can have a digital chart at a behavioral-health clinic and still have key information fail to follow them to an emergency department, primary-care office, or another treatment program. ONC said 67% of facilities that were not in a Health Information Exchange were unfamiliar with one or did not know whether one was available in their area. (healthit.gov) (hitconsultant.net) Behavioral-health providers have trailed other parts of medicine for years in certified record adoption and interoperability tools. ONC and the Substance Abuse and Mental Health Services Administration launched a three-year Behavioral Health Information Technology Initiative in February 2024 with more than $20 million in SAMHSA funding to address that gap. (healthit.gov) Hospitals are moving ahead with AI anyway. ONC reported that 71% of hospitals used predictive AI integrated with the electronic health record in 2024, up from 66% in 2023, which leaves health systems trying to scale algorithms on top of data infrastructure that is still patchy across care settings. (healthit.gov) The near-term work is less about buying another model than about making records usable across systems, sites, and specialties. Until that cleanup happens, more healthcare organizations will have AI pilots than AI at scale. (hitconsultant.net)