Online Program Home
My Program

Abstract Details

Activity Number: 618 - Machine Learning for Big Data
Type: Contributed
Date/Time: Thursday, August 1, 2019 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistical Learning and Data Science
Abstract #305056
Title: Patient Factors at Diagnosis and Overall Risk of Mortality in US Population-Based Pediatric Oncology: An Evaluation Using SEER Data
Author(s): Fatima Boukari* and Md Jobayer Hossain
Companies: Delaware State University and Nemours children Healthcare Systems
Keywords: unmeasured heterogeneity; mortality risk; variation; aggregated risk

Patient factors at diagnosis, including age, race-ethnicity, sex, and regions, have consistently demonstrated varying risk of mortality across site-specific pediatric cancers in the US. While these results have shown great utility in improving cancer-specific survival, public health policy often focuses on promoting well-being at the population level. Aggregated risk associated with patient factors across all cancers can better inform public health planning, surveillance, and resource allocation in childhood cancer. SEER data provide a rich, high-quality information pool to estimate aggregated mortality risk and SEER reports combined mortality statistics of all cancer sites by patient factors and for overall, but this aggregation doesn’t account for the unmeasured heterogeneity caused by possible interference of extraneous factors including wide variation in mortality of cancer types. This study has determined the effect of patient factors on overall pediatric cancer survival using frailty model, accounting for heterogeneity in mortality risk across cancer types and diagnosis years, on a SEER dataset of 89,730 pediatric cancer patients diagnosed and followed between 1973 and 2013.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program