Improving care for older adults: a model to segment the senior population

Area of Science:

• Gerontology
• Health Services Research
• Data Science in Healthcare

Background:

• Effective population-level care management for older adults requires robust risk stratification.
• Identifying distinct patient needs, from prevention to end-of-life care, is crucial for tailored interventions.

Purpose of the Study:

• To validate the Senior Segmentation Algorithm for classifying individuals aged 65+ into four distinct care groups.
• The algorithm utilizes electronic health record data to predict patient needs and healthcare utilization.

Main Methods:

• Validation involved comparing algorithm output with physician judgment for 1615 Medicare recipients.
• Assessed segmentation stability over one year, including transitions between care groups.
• Evaluated algorithm's predictive performance against mortality, hospitalization, and readmission rates.

Main Results:

• Algorithm demonstrated 85% concordance with physician assessments.
• Over one year, 85% of seniors remained in their assigned care group, with 11% transitioning to higher-need groups.
• Mortality and hospitalization prediction performance was comparable to existing scores.

Conclusions:

• The Senior Segmentation Algorithm reliably identifies older adults into care groups with similar needs and utilization patterns.
• Implementation across Kaiser Permanente aims to define group-specific care elements.
• Future work will focus on mitigating progression to higher-need groups and optimizing care delivery.