Mayo Clinic & Bayesian Health Co-Develop AI-Powered Solution to Expand Palliative Care Access – Image for illustrative purposes only (Image credits: Pixabay)
Patients with advanced conditions often cycle through hospitals multiple times because supportive care arrives too late in their stay. This pattern leaves many without the pain management or planning that could ease their course and reduce returns. Mayo Clinic and Bayesian Health have now built an AI system meant to flag those needs automatically while patients are still in the acute phase.
The Everyday Reality Behind Repeated Admissions
One in three hospital readmissions involves people living with serious, life-limiting illness. Yet fewer than half of those who could benefit from palliative support ever receive a consultation during their inpatient time. The result is a loop of aggressive treatments that may not align with what patients actually want once their goals become clearer.
Clinicians in busy wards already know the signs of mounting distress, but spotting them across every record in real time exceeds what manual review can reliably achieve. The new platform works inside the existing electronic health record to surface patterns that might otherwise stay buried until a crisis forces action.
What the Trial Data Actually Showed
An earlier version of the tool underwent randomized testing at Mayo Clinic’s Department of Medicine. Results linked its use to a 44 percent rise in timely palliative referrals. The same deployment tracked a 25 percent drop in readmissions within 60 days and a 28 percent drop within 90 days.
These figures come from a controlled setting rather than broad rollout, so questions remain about how well the gains hold when the system expands to other hospitals or patient mixes. Still, the measured reductions in returns point to a practical shift in how care teams can intervene before discharge plans solidify.
How the System Fits Into Daily Work
The AI scans the full longitudinal record for each admitted patient instead of checking isolated vital signs or lab values. It then translates those signals into two separate views: one hospital-wide dashboard for the palliative team and one set of plain-language prompts that appear directly in the bedside clinician’s workflow.
Because the models keep learning from local feedback, accuracy can improve the longer the tool runs inside a given health system. This continuous adjustment aims to reduce the alert fatigue that often accompanies new technology in already crowded clinical environments.
Who Stands to Gain and What Comes Next
The partnership grew out of Mayo Clinic’s Practice Transformation Ventures program, with the Rochester Department of Medicine serving as the main clinical validator. Early focus stays on inpatient adults, though the same logic could later extend to other settings where serious illness trajectories are harder to track.
- Specialist teams receive a single, updated list of patients showing unmet needs.
- Frontline staff see clear next steps without leaving their usual screens.
- Health systems gain measurable data on referral timing and downstream readmission rates.
Jacob J. Strand, chair of palliative care at Mayo Clinic, noted that the core challenge has always been recognizing the right moment to act rather than deciding what the care plan should contain once that moment arrives.
Remaining Questions as Adoption Grows
Broader use will test whether the same performance appears outside the original trial site and across different electronic record platforms. Questions also persist about how the system handles rapidly changing patient preferences or incomplete documentation common in complex cases.
For now, the collaboration supplies one concrete route toward catching palliative needs before another preventable admission occurs. The longer-term effect on patient experience and overall hospital capacity will depend on how widely and carefully the approach is refined.
