Automated Identification of Patients with Advanced Illness.

Area of Science:

• Healthcare Management
• Clinical Informatics
• Machine Learning in Medicine

Background:

• Early identification of advanced illness is vital for aligning patient care with their wishes.
• Goals of care conversations empower patients, preventing unwanted default treatment plans.
• Current methods for identifying these patients need improvement.

Purpose of the Study:

• To evaluate the performance of the LACE score versus a novel Hospital Impairment Score (HIS) for identifying advanced illness patients.
• To introduce and describe the implementation of the HIS model in an inpatient setting.

Main Methods:

• The study compared the LACE (Length of stay, Acuity of Admission, Co-morbidities, Emergency room visits) score with the HIS.
• The HIS model integrates rule-based insights with a machine learning algorithm.
• The HIS model was piloted at a single hospital before production launch.

Main Results:

• The Hospital Impairment Score (HIS) demonstrated significantly superior performance compared to the LACE score.
• The HIS model successfully identified advanced illness patients within the inpatient population.
• The HIS model is currently in production and in use by clinicians.

Conclusions:

• The HIS model offers a more effective approach to identifying advanced illness patients than the LACE score.
• Implementing the HIS model can improve the alignment of care with patient values and wishes.
• Machine learning-driven tools can enhance clinical decision-making for patient care planning.