Improving End-of-Life Care through AI-Based Clinical Decision Support.

Many patients with serious illness prioritize comfort over the prolongation of life in the final days and weeks prior to death. Goals-of-care discussions (GOCDs) can provide patients with the opportunity to express their preferred end-of-life experience, prevent aggressive and often futile interventions, improve patient satisfaction, and reduce unnecessary costs. Clinicians may feel uncomfortable initiating these conversations, however, given the lack of relevant training and difficulty identifying patients at elevated risk of mortality. The authors developed an intervention to promote GOCDs that combines mortality risk estimation using AI, clinician training, person-to-person alert messages, streamlined clinical workflows, and enhanced palliative care capacity. The program was implemented across 8 of 10 adult hospitals in the BJC HealthCare system, and data were collected between December 22, 2020 (at launch in the first hospital sites) and December 31, 2024. During this time, more than 300 clinicians were trained through the program, and they identified 13,976 patients as candidates for GOCDs. Clinicians exhibited a high response rate (93%) to the patient eligibility alert. Among patients without a prior documented GOCD at the time of messaging, 54% of responding physicians opted to initiate a GOCD with the patient, and another 24% requested palliative care to initiate the GOCD. Systemwide improvements were observed across several metrics, including a fivefold increase in GOCDs, from 1.2% of 146,257 encounters in 2021 to 6.7% of 167,681 encounters in 2024, and a 63% increase in the proportion of encounters with palliative care consults, from 2.2% of 146,257 encounters in 2021 to 3.6% of 167,681 encounters in 2024. The Vizient Mortality Index (a ratio of observed to expected mortality) also decreased by 32% during this time frame (where a lower score indicates that fewer patients died than would have been expected), from 0.92 in 2021 to 0.62 in 2024. Keys to implementation included identifying a staff member at each site to orchestrate alert messaging and ensuring adequate palliative care staffing at each site. This case study demonstrates the importance of pairing accurate mortality predictions with systems and resources that enable clinicians to act on these predictions, including engaging with alerts and improving through comprehensive clinical training.