Perceived Risks and Benefits of Using a Survival and Functional Outcome Machine Learning Model for Glioblastoma

<h3>Project Overview</h3>
<p>This project aims to better understand stakeholder perspectives on the use of an advanced machine learning (ML) artificial intelligence (AI) algorithm to predict prognosis and inform surgical decision-making. Stakeholders include patients (PAT), caregivers (CRG), and neurosurgeons (NSG). Patients are adults diagnosed with a malignant glioblastoma (GBM) brain tumor. Caregivers are adults who provide care for individuals with GBM and include spouses, partners, parents, and adult children. Neurosurgeons are clinicians who specialize in the treatment of GBM and practice in the United States. The ML AI algorithm, developed by scientists at Washington University in St. Louis, is designed to support prognosis prediction, such as life expectancy and quality of life, and to inform surgical decision-making for individuals with malignant GBM. Using data from resting-state functional MRI (rs-fMRI) brain scans, the algorithm predicts how patients are likely to respond to tumor resection surgery. It can estimate the likelihood of short-, mid-, and long-term survival (<1 year, 1–2 years, and >2 years, respectively) with approximately 90% accuracy. In addition, the algorithm classifies functional outcomes as positive or negative based on a Karnofsky Performance Status (KPS) cutoff score of 70, achieving 94% accuracy. Although the algorithm has not yet been implemented in clinical practice, plans for widespread future use are underway. Prior to large-scale implementation, it is essential to understand the perspectives of key stakeholders who will be directly affected by its use. Gaining insight into perceived risks, benefits, and ethical and practical challenges including new skills required by neurosurgeons and the unique informational needs of patients and caregivers is critical to responsible and effective adoption.</p>