What You Should Know

• Mayo Clinic and real-time clinical intelligence vendor Bayesian Health have co-developed an AI solution to identify hospitalized patients who require palliative care early in their stay.
• Validated randomized clinical trial findings showed an earlier version of the tool was associated with a 44% increase in timely palliative referrals, a 25% reduction in 60-day readmissions, and a 28% reduction in 90-day readmissions.
• The platform targets a severe gap in serious illness management, addressing industry data showing that while one-third of hospital readmissions involve patients with serious illness, fewer than half ever receive a palliative consultation.
• Integrated directly into the Electronic Health Record (EHR) workflow, the solution provides an automated, hospital-wide view for palliative teams and clear, interpretable guidance for bedside clinicians.
• The strategic collaboration was executed under Mayo Clinic’s Practice Transformation Ventures (PTV) framework, with the Department of Medicine in Rochester serving as the core practice owner and validator.

---

Slashing Readmissions by 25%: How Mayo Clinic and Bayesian Health Developed a Proactive Palliative AI

The management of serious illness within complex inpatient settings represents one of the most significant clinical and financial challenges facing the modern healthcare economy. Approximately one-third of all hospital readmissions involve patients struggling with advanced, life-limiting conditions, many of whom experience a frustrating cycle of repeated, preventable hospitalizations. Despite clear evidence that early supportive interventions dramatically stabilize these trajectories, fewer than half of hospitalized patients who qualify for palliative care ever receive a formal consultation during their stay.

The obstacle is rarely a lack of clinical intent, but rather a structural timing failure. In dense, fast-moving acute care environments, identifying subtle, multi-system declines, complex symptom burdens, or shifting goals of care requires a level of continuous record synthesis that exhausts traditional human bandwidth. By the time a manual consult is finally triggered, it is frequently too late in the inpatient stay to prevent non-beneficial, aggressive treatments or alter the trajectory of a post-discharge readmission.

To shift this paradigm from reactive crisis management to proactive clinical orchestration, Mayo Clinic has finalized a co-development partnership with Bayesian Health. Engineered under Mayo Clinic’s Practice Transformation Ventures (PTV) framework, the strategic alliance has delivered an EHR-integrated, real-time clinical intelligence platform designed to catch unmet palliative care needs at the earliest stages of admission.

Validated Outcomes: Hard Data from a Randomized Trial

While much of the healthcare artificial intelligence market is flooded with uncalibrated pilots and unverified marketing models, the Mayo Clinic-Bayesian Health platform enters the industry backed by rigorous clinical trial data.

In a randomized clinical trial conducted internally by Mayo Clinic’s Department of Medicine, validated findings from an earlier iteration of the software program demonstrated exceptional clinical efficacy. The deployment of the point-of-care tool was directly associated with:

• A 44% increase in timely palliative care referrals, allowing clinicians to capture patients before their trajectories destabilized.
• A 25% reduction in 60-day hospital readmissions, significantly preserving enterprise bed capacity.
• A 28% reduction in 90-day readmissions, alongside a measurable, auditable improvement in documented patient quality of life.

Jacob J. Strand, M.D., chair of Palliative Care at Mayo Clinic, emphasized that the true barrier to effective palliative medicine has never been defining a care plan, but discovering the underlying patient need early enough to alter the course of clinical care. By delivering tailored, patient-specific signals to bedside and central teams alike, the technology successfully cuts through inpatient complexity to hardwire consistent decision-making.

Turning Data Noise into Interpretable Action

The underlying infrastructure of the Bayesian Health platform functions by continuously reading the entire longitudinal electronic record of a hospitalized patient. Rather than evaluating vital signs or laboratory values in isolation, the software applies complex clinical reasoning to map subtle, compounding changes against the patient’s individual baseline.

The new palliative module introduces a dual-sided operational dashboard that eliminates traditional notification fatigue. Specialized palliative consultation teams are provided with a real-time, hospital-wide view of vulnerable patients showing signs of unmet pain management or caregiver distress. Simultaneously, frontline bedside clinicians receive clear, interpretable guidance directly inside their daily EHR workflow, offering a streamlined, automated path to execute a consultation request.

Furthermore, the clinical AI is built on a continuous reinforcement framework. The underlying models constantly learn from localized clinician feedback and shifting patient population trends, systematically improving its predictive identification accuracy the longer it runs within the health system.