Background: Advance care planning (ACP) is an underutilized aspect of patient care due to clinician time constraints, inadequate care processes, and poor clinician prognostication. Consequently, hospitalized patients with a high risk of mortality and readmission often do not have clearly defined goals of care, and their families may not be prepared to discuss end-of-life decisions. This creates difficult scenarios for patients and additional burden for their families. Machine learning models can now be used to identify patients with a high risk of mortality and morbidity. These models can be utilized to increase ACP discussions and end-of-life care planning.

Purpose: To utilize a machine-learning model to identify high-risk patients for deployment of a targeted advance care planning intervention and multi-disciplinary care review.

Description: Our team utilized a machine-learning model to identify hospitalized general medicine patients appropriate for ACP. Once identified, an administrator notifies the patient’s attending of record as well as the pharmacy, coding and documentation, and case management personnel involved in the patient’s care. The attending physician determines if ACP is appropriate by reviewing the patient’s history and current clinical status. If appropriate, the attending has a goals of care discussion with the patient and completes documentation in a standardized template in the electronic health record. Pharmacy conducts a medication reconciliation at the time of discharge on the identified patient. The coding and documentation improvement team reviews the clinical chart to ensure appropriate documentation of the patient’s clinical status. Case management facilitates referrals, provides access to discharge resources, reinforces the importance of close outpatient follow up, offers palliative care resource education, and sends the ACP documentation to any post-acute referrals.

Conclusions: By utilizing a machine-learning model and an integrated team approach to complex care, we created an innovative multi-disciplinary program to improve care for patients at high risk for mortality and morbidity. This project helps direct limited resources to a high-risk patient population. The collaboration of front-line providers in conjunction with targeted utilization of ancillary services improves patient care and optimizes care delivery. Since implementation, there has been an increase in completion of ACP documentation as well as change in resuscitation preference to do-not-attempt-resuscitation in the electronic medical record of identified patients.