Related Experiment Video
Updated: May 22, 2025
07:41
Performing Data Mining And Integrative Analysis Of Biomarker in Breast Cancer Using Multiple Publicly Accessible Databases
Published on: May 17, 2019
8.8K
Evaluating PREDICT and developing outcome prediction models in early-onset breast cancer using data from Alberta,
Robert B Basmadjian1, Yuan Xu1,2,3, May Lynn Quan1,2,3
1Department of Community Health Sciences, Cumming School of Medicine, University of Calgary, Calgary, AB, Canada.
Breast Cancer Research and Treatment
|March 12, 2025
Summary
NHS PREDICT v2.1 overestimates mortality in early-onset breast cancer (EoBC) patients. New models showed better discrimination but poor calibration, highlighting the need for updated data and variables for accurate EoBC outcome prediction.
Area of Science:
- Oncology
- Biostatistics
- Epidemiology
Background:
- Limited research exists on outcome prediction for early-onset breast cancer (EoBC).
- Accurate prediction models are crucial for tailoring treatment and improving survival rates in younger breast cancer patients.
Purpose of the Study:
- To evaluate the predictive performance of NHS PREDICT v2.1 for early-onset breast cancer (EoBC).
- To develop and validate novel prediction models for 5-year and 10-year all-cause mortality in EoBC patients.
Main Methods:
- A cohort of 1827 adults under 40 diagnosed with invasive breast cancer in Alberta, Canada (2004-2020) was analyzed.
- NHS PREDICT v2.1 was used to extract mortality estimates; LASSO Cox and Random Survival Forests (RSF) models were developed.
- Internal validation employed nested tenfold cross-validation, with performance assessed via ROC and calibration curves.
Main Results:
- NHS PREDICT v2.1 overestimated 5-year mortality by 2.4% and showed no significant difference at 10 years.
- LASSO Cox models demonstrated superior discrimination compared to RSF models at both 5 and 10 years.
- Both developed models exhibited poor calibration, underestimating mortality in the EoBC cohort.
Conclusions:
- NHS PREDICT v2.1 overestimates mortality in EoBC patients with higher predicted risks.
- Machine learning approaches did not yield clinically useful models without incorporating additional data.
- Future research requires extended follow-up and inclusion of systemic treatment variables for improved EoBC outcome prediction.
