Mayo Clinic boosts palliative referrals 44%

Mayo Clinic and Bayesian Health said on May 19 that they had co-developed an AI system to identify hospitalized patients who may benefit from palliative care earlier in their stay. The tool is embedded in the electronic health record and is designed to surface patient-specific signals inside clinicians’ existing workflow, rather than through a separate dashboard. Mayo Clinic said the program was associated in a randomized clinical trial with a 44% increase in timely palliative care referrals, a 25% reduction in 60-day readmissions and a 28% reduction in 90-day readmissions. ### What exactly changed inside the hospital workflow? Bayesian Health said its role was to integrate the model into the electronic health record so care teams could see the information within existing clinical workflows. Mayo Clinic said the system is meant to identify patients with unmet palliative care needs earlier and give clinicians context for the conversations and care coordination that follow. (newsnetwork.mayoclinic.org) Mayo Clinic described the problem in operational terms. Roughly one-third of readmissions involve patients with serious illness, the clinic said, while fewer than half of those patients receive palliative care consultations. The tool was built to narrow that gap by flagging patients earlier in the admission. (newsnetwork.mayoclinic.org) ### Why does a 44% increase in referrals matter here? The 44% figure refers to timely palliative care referrals in a randomized clinical trial conducted at Mayo Clinic, according to the organizations’ May 19 announcement. The same announcement said the earlier version of the program was also associated with lower readmissions at both 60 and 90 days, as well as improved patient quality of life. (newsnetwork.mayoclinic.org) Jacob J. Strand, chair of Palliative Care at Mayo Clinic, said the issue was not only identifying unmet needs, but identifying them early enough “to change the course of care.” Strand said the difference came from tailoring workflows to local culture, patient acuity and coordination across central and bedside teams. (newsnetwork.mayoclinic.org) ### Was this a research project or a live deployment? ClinicalTrials.gov lists a Mayo Clinic study titled “Impact of Predictive Modeling on Time to Palliative Care in an Outpatient Primary Care Population” as completed, with actual enrollment of 127,070 and primary completion on May 31, 2021. The registry says the study used a machine learning algorithm to identify patients who may benefit from palliative care consults and present them to a palliative care specialist for review. (newsnetwork.mayoclinic.org) The May 19 announcement describes the current effort as a deployed hospital workflow tool built from that earlier work. Mayo Clinic said its Department of Medicine led the clinical development and validation, while Bayesian Health handled EHR integration. ### What is the narrower lesson for clinical AI? (clinicaltrials.gov) Mayo Clinic said this was its first collaboration of this kind to use AI across the care process in a complex hospital setting. The organizations framed the system as one that supports clinicians at the point of care, with patient information kept coordinated and confidential inside the existing record. (newsnetwork.mayoclinic.org) Bayesian Health has previously built EHR-embedded clinical intelligence tools for hospital use, including sepsis detection programs described on its research page. In this case, the palliative care program adds another example of a targeted model tied to a specific workflow, referral decision and downstream utilization measure. That is an inference from the announced design and reported outcomes, not a claim either organization made in those words. (newsnetwork.mayoclinic.org) ### What happens next, and where can readers check the record? May 19 is the date of the public announcement from Mayo Clinic and Bayesian Health, and ClinicalTrials.gov continues to host the related Mayo study record under NCT04604457. Mayo Clinic’s news release says the tool is now part of care-team workflow, while the trial registry preserves the earlier study design, dates and enrollment details. (newsnetwork.mayoclinic.org) (bayesianhealth.com)