Vital.io

Boarding is usually framed as a bed shortage. A new study in JAMA reframes it as a handoff problem. Researchers analyzed boarding across 56 hospitals in 17 health systems through the RESQUE-NET network. Among patients already accepted for general hospital admission, roughly 23% waited at least four hours before actually transitioning to inpatient-level care. Some waited 12 to 24 hours. The variation between hospitals was wide, and it was not random: larger hospitals, teaching institutions, and those serving more Medicaid patients had the longest delays. The distinction matters. These are not patients waiting for a decision. The decision is made, the destination is known, and the patient is still sitting in a department that is no longer really responsible for them and not yet handed to the team that will be. Lead author Alex Janke put it directly: "Risk concentrates at these moments where one clinical team hands a patient to another." Emergency and inpatient teams run on different clocks and different constraints. The interval between them is one almost nobody measures, which is part of why it stretches. Vital doesn’t change who owns the patient during that window. But we can help keep the patient and family informed inside it, so the hours when accountability is ambiguous aren’t also hours of silence. Read the study here: https://lnkd.in/gtjUjzzP

In October 2020, Ryan Randall sat in a hospital parking lot waiting to hear about his 15-month-old son. He was left with no updates and no way to know what was happening inside. That night is why he's now Vital's VP of Commercial. Ryan's story leads the first edition of The Vital Voice, our new monthly newsletter. Also inside: what roughly 140,000 emergency department visits told us about why patients leave before completing care, new Vital product updates, and what three healthcare leaders had to say about AI safety at HMPS26.

93% of health systems have deployed third-party AI, but only 4% say they have an advanced strategy for governing it. The Center for Connected Medicine at UPMC and KLAS Research surveyed C-suite, clinical informatics, and IT executives across 27 health systems for a new report on AI governance and testing, published earlier this month. Their findings: →  93% have deployed third-party AI solutions →  4% describe it as advanced The gap between those last two numbers is the story. Nearly everyone tests. Fewer than half have a place to test properly. The executives named the bottlenecks plainly: manual spreadsheets, inconsistent data definitions across teams, and not enough people. Procurement has outrun validation. A health system evaluating its eighth AI vendor with a spreadsheet and no sandbox isn't necessarily making a bad decision, but it's making an unverifiable one, and unverifiable decisions are what erode clinician and patient trust in the tools that actually work. The reasonable response is to expect more from vendors. Validation evidence, clear data provenance, security posture, and a deployment path that does not consume months of IT capacity should arrive with the product instead of being reconstructed afterward by a stretched informatics team. That is the standard Vital holds itself to: FHIR-first integration in 8 to 10 weeks with minimal IT lift, HITRUST CSF certified, SOC 2, HIPAA compliant. If only 4% of health systems have an advanced AI strategy, the burden of proof belongs with the teams building the AI tools that go in front of patients. Read the report coverage here: https://lnkd.in/gjXZn4ZE

Patients aren’t rejecting AI in their care. They’re rejecting AI that no one is accountable for. Salesforce surveyed adults across eight countries for its Connected Health Consumer Report, fielded earlier this year. The headline number: patients are 3x more likely to trust AI built into their provider’s secure portal than a public chatbot. Same underlying technology. Different answer, because the question patients are actually asking is not "is this AI accurate" but "who is responsible if it is wrong." Patients will take AI, at scale, at any hour, provided there is a named institution behind it and a person they can reach. The cost of getting it wrong shows up in the same survey: 46% have delayed care because a digital process was confusing, and 58% have postponed care because scheduling was too hard. Vital was built on the provider side of that 3x. Its explanations draw on a patient’s own real-time record inside their health system, not the open internet, and every interaction stays connected to the care team, so escalation to a human is the design rather than the fallback. Trust in healthcare AI is turning out to be less a question of model quality than of provenance and recourse. Patients want to know whose AI it is. Read the full report here: https://lnkd.in/eXQNFm2Z

Two out of three patients now read their test results before anyone explains them. New research in JAMA Network Open, led by Jemar Bather at the NYU School of Global Public Health, analyzed the 2024 Health Information National Trends Survey. 68.6% viewed immediately released test results before hearing from their clinician. Following the 21st Century Cures Act, results reach patients the moment they exist. What this study measures is what happens next. How well patients understood what they saw: → 34.9% understood very well → 31.7% understood well → 26.8% understood only fairly well → 6.6% understood poorly Roughly a third of patients opening their own results are not confident they know what those results mean. And the pattern isn’t random. Digital literacy was the strongest predictor of viewing results early, so the patients most likely to encounter a raw number with no explanation are also the most digitally fluent ones. Patient-centered communication was associated with roughly half the prevalence of poor comprehension. The variable that moved understanding was not the portal. It was whether someone explained. The authors point directly at the fix, including AI that summarizes results in plain language. That is what Vital does. Reading real-time EHR data, Vital translates results, tests, and next steps into language matched to how each patient actually reads, with no app and no account, at the moment the result lands rather than days later at a follow-up. Read the full study here: https://lnkd.in/gqmyAata

A diagnosis that is correct but never clearly explained is not a finished diagnosis. That idea sits inside the Centers for Disease Control and Prevention's Core Elements of Hospital Diagnostic Excellence, developed with AHRQ and CMS. The framework lays out six elements for hospital diagnostic programs, spanning leadership accountability, diagnostic stewardship, stronger systems and processes, and learning from diagnostic safety events. What stands out is how it treats communication with patients, families, and caregivers as part of the diagnostic process itself rather than a courtesy that follows it. Ordering a test, interpreting it, communicating it, and acting on it are named as one chain. The stakes are documented. An AHRQ-funded analysis in BMJ Quality & Safety estimates that 795,000 Americans die or are permanently disabled each year as a result of diagnostic error, including 371,000 deaths. Most diagnostic safety work targets clinician reasoning. Far less of it targets the last link, whether the patient understood what was found and what to do next. That link is where Vital works, turning results and next steps into language patients can act on while they are still in the building. Read the framework here: https://lnkd.in/etNeM76J

Emergency department boarding is usually counted in minutes. A new study counts it in patients who got sicker while they waited. Researchers analyzed 173,168 adult encounters across a five-hospital academic health system from 2018 to 2024, looking at patients admitted to a floor bed who boarded between 4 and 48 hours. Published in Annals of Emergency Medicine, the analysis found 3.6% deteriorated, escalating to intermediate or intensive care within 48 hours of the admission order. Nearly half of those escalations, 45%, happened while the patient was still in the ED. After adjusting for comorbidities, vital signs, labs, census, and socioeconomic status, boarding duration was independently associated with higher odds of deterioration. The accompanying editorial calls it a dose of delay. Boarding is not a queue patients passively sit in. It is exposure, and the dose climbs with every hour. Length of stay is a safety measure before it is a satisfaction measure or a financial one. Vital doesn’t create more beds, but it can help make the boarding interval visible, giving patients and families real-time status instead of silence. It also helps care teams identify who has been waiting longest. Hours nobody is tracking are the hardest ones to shorten. Read the full study here: https://lnkd.in/gkD7YFZM

AI is giving clinicians back the equivalent of more than 16 working days a year. What happens to that time is the part worth arguing about. Philips published its Future Health Index 2026 in June. The index finds among clinicians using AI tools their organization provided, 58% report better workflow efficiency, 54% report faster diagnostic decision-making, and 49% report less work-related stress. Half say they now have capacity for roughly eight more patients per week. Coverage of numbers like these tends to stop at throughput. More patients seen, more revenue captured. But the stress finding is the more useful one, because it says something about what the time is for. A clinician who isn't rushed can explain a result, answer the question behind the question, and actually be present in the room. That only holds if the routine explaining gets handled somewhere. Vital takes the predictable questions, what a test means and what happens next, so the time AI gives back goes to the conversations that need a person. Read the full report here: https://lnkd.in/eBePW-nk

The clinician's job is quietly shifting from source of information to interpreter of it. Maneesh Goyal, COO of Mayo Clinic Platform, described the change in Becker's Healthcare. As patients arrive having already used generative AI to research their condition and prepare questions, the clinician becomes the trusted expert who helps them interpret that information against their own history and circumstances. That's a real change in the work, and it's worth naming what it costs. Interpretation takes time, and it lands in visits nobody lengthened to accommodate it. A patient who arrives with five AI-generated questions is more engaged and also more expensive in minutes. Systems can respond two ways. Absorb the time, or move the routine part of interpretation earlier so the visit starts further along. The second is what Vital is built for. Plain-language explanations of tests and results drawn from the patient's own record, before and during the visit, so the clinician's time goes to the judgment only they can provide rather than the translation. Read the full article here: https://lnkd.in/gfhap5-q

Thank you for leading the conversation and providing support for this extremely important topic where patient safety and doing no harm is critical! #HMPS26 #patientexperience #patientsafety