Ethics of using artificial intelligence to target serious illness conversations for patients with life-limiting illness
Saturday, October 14, 2023
7:30 AM – 8:45 AM ET
Location: Essex C (Fourth Floor)
For patients with life-limiting illness, receiving serious illness conversations (SIC) can improve provision of goal-concordant care and affect patient outcomes. However, SICs are often limited by clinician time constraints and uncertainty regarding who would benefit from these conversations. Disparities in SICs exist by patient race, ethnicity and socioeconomic status suggesting clinician biases influence who they choose to talk about serious illness. Predictive analytics, using tools such as machine learning algorithms which use autonomous systems to analyze, test, and learn from patient data, are being increasingly used in an attempt to target SICs to patients at highest risk of death. While predictive analytics has the potential to reduce disparities by objectively assessing who is targeted for SICs, they may worsen disparities by underestimating risk of death for patients with limited access to healthcare and minority groups, thereby decreasing targeted access to SICs for these groups. Underestimation of patient risk may occur due to inaccurate and missing data or limited inclusion of socioeconomic demographics associated with mortality into predictive analytics models, disproportionately impacting patients with low access to healthcare and minority groups We will discuss our findings using prediction analytics to target SICs and evaluate how this affects equity in SIC access for patients across racial, ethnic and socioeconomic groups. We will also discuss ways to improve equity in SIC access by assessing how inclusion of disadvantage indices may affect equitable targeting of SICs and evaluating how clinician directed behavioral nudges can be used alongside predictive analytics results to improve SIC completion.
Juan Rojas – Pulmonary Critical Care – Rush University Medical Center; Robert Arnold – Palliative Care and Medical Ethics – University of Pittsburgh