EUVIMEDOpdag EUVIMED
ValidierungsstudieOpen access

Transfer Learning and Machine Learning for Training Five-Year Survival Prognostic Models in Early Breast Cancer: Development and Validation Study

Lisa Pilgram iD, Kai Yang iD, Ana-Alicia Beltran-Bless iD, Gregory R Pond iD, Lisa Vandermeer iD, John Hilton iD, Marie-France Savard iD, Andreanne LeBlanc iD, Lois Shepherd iD, Bingshu Chen iD, John Ms Bartlett iD, Karen J Taylor iD and 13 more

Journal of Medical Internet Research · 2026-04-14 · 28 · e88665

DOI: 10.2196/88665 ↗

At a glance

Validierungsstudie in Journal of Medical Internet Research, published online on 2026-04-14. Written by Lisa Pilgram and 24 co-authors. Institution: School of Epidemiology and Public Health, Faculty of Medicine, University of Ottawa, Ottawa, ON, Canada. Major MEDLINE topics: Breast Neoplasms, Machine Learning. Cited 1 times so far (Crossref); 65 references. The full text is freely available in PubMed Central.

AbstractSubjectsBibliografiske oplysningerCiteRelated studies

Abstract (original language)

Background

Prognostic information is essential for decision-making in breast cancer management. In recent years, trials and clinical practice have emphasized genomic prognostication tools, despite clinicopathological methods being more affordable and accessible. PREDICT v3 is one such tool with promising results across cohorts. Advances in machine learning (ML), transfer learning, and ensemble methods provide opportunities to enhance these approaches, especially where missing data and model assumptions differ across diverse populations.

Objective

This study evaluates the potential to improve survival prognostication in breast cancer. More precisely, we compare de novo ML, transfer learning from the pretrained prognostication model PREDICT v3, and a stacked ensemble approach.

Methods

Data from the MA.27 trial (NCT00066573) were used for model training, with external validation on data from the Tamoxifen Exemestane Adjuvant Multinational trial (NCT00279448 and NCT00032136) and a US Surveillance, Epidemiology, and End Results cohort. Transfer learning was applied by re-estimating the parameters of the pretrained prognostic tool PREDICT v3. De novo ML included random survival forests and extreme gradient boosting, and the ensemble was implemented using weighted linear stacking of model predictions. Internal and external validation was assessed in terms of the integrated calibration index and discrimination. Shapley Additive Explanations values were used to explain model predictions and decision-curve analysis to facilitate the interpretation of performance differences.

Results

Transfer learning, de novo random survival forest, and the stacked ensemble improved calibration in MA.27 over the pretrained model (integrated calibration index reduced from 0.042 in PREDICT v3 to ≤0.007) while discrimination remained comparable (AUROC increased from 0.738 in PREDICT v3 to 0.744-0.799). In decision-curve analysis, these approaches demonstrated consistently positive net benefit across clinically relevant thresholds, while PREDICT v3 lost net benefit beyond 7.5% predicted risk. Invalid PREDICT v3 predictions were observed in 23.8% to 25.8% of MA.27 individuals due to missing information. In contrast, ML models and the stacked ensemble predicted survival despite missing data. Across all models, patient age, nodal status, pathological grading, and tumor size had the highest Shapley Additive Explanations values, indicating their importance for survival prognostication. External validation in the US Surveillance, Epidemiology, and End Results cohort confirmed the benefits of transfer learning, RSF, and ensemble in terms of calibration while maintaining discrimination at comparable levels. In contrast, generalizability was limited in the Tamoxifen Exemestane Adjuvant Multinational trial, a cohort with a substantially different distribution of clinicopathological characteristics.

Conclusions

This study demonstrates that transfer learning, de novo RSF, and a stacked ensemble can improve prognostication compared with the pretrained PREDICT v3, particularly in the presence of missing or uncertain inputs. Transportability may be limited in cohorts with different clinicopathological profiles, requiring local validation before clinical deployment. Ultimately, better survival estimation can provide more meaningful guidance in breast cancer care. Trial Registration ClinicalTrials.gov NCT00066573; https://clinicaltrials.gov/study/NCT00066573, NCT00279448; https://clinicaltrials.gov/study/NCT00279448, NCT00032136; https://clinicaltrials.gov/study/NCT00032136

Subjects

MeSH terms (MEDLINE)

Breast NeoplasmsMachine LearningFemaleHumansBoosting Machine Learning AlgorithmsPrediction AlgorithmsPredictive Learning ModelsPrognosisRandom ForestSurvival AnalysisClinical Trials, Phase III as TopicMulticenter Studies as TopicRandomized Controlled Trials as Topic

Author keywords

breast cancer survivalensemblesmachine learningprognostic modelstransfer learning

Authors and institutions

  • School of Epidemiology and Public Health, Faculty of Medicine, University of Ottawa, Ottawa, ON, Canada.
  • Research Institute, Children's Hospital of Eastern Ontario, Ottawa, ON, Canada.
  • Department of Nephrology and Medical Intensive Care, Charité - Universitaetsmedizin Berlin, Berlin, Germany.
  • Division of Medical Oncology, Faculty of Medicine, University of Ottawa, Ottawa, ON, Canada.
  • Department of Oncology, McMaster University, Hamilton, ON, Canada.
  • Cancer Therapeutics Program, Ottawa Hospital Research Institute, Ottawa, ON, Canada.

ORCID profiles: Lisa Pilgram, Kai Yang, Ana-Alicia Beltran-Bless, Gregory R Pond, Lisa Vandermeer, John Hilton, Marie-France Savard, Andreanne LeBlanc, Lois Shepherd, Bingshu Chen, John Ms Bartlett, Karen J Taylor, Jane Bayani, Sarah Barker, Melanie Spears, Cornelis Jh van der Velde, Elma Meershoek-Klein Kranenbarg, Luc Dirix, Elizabeth Mallon, Annette Hasenburg, Christos Markopoulos, Lamin Juwara, Fida K Dankar, Mark Clemons, Khaled El Emam

Cite

APA

Pilgram, L., Yang, K., Beltran-Bless, A. A., Pond, G. R., Vandermeer, L., Hilton, J., Savard, M. F., LeBlanc, A., Shepherd, L., Chen, B., Bartlett, J. M., Taylor, K. J., Bayani, J., Barker, S., Spears, M., van der Velde, C. J., Meershoek-Klein Kranenbarg, E., Dirix, L., Mallon, E., …, & El Emam, K. (2026). Transfer Learning and Machine Learning for Training Five-Year Survival Prognostic Models in Early Breast Cancer: Development and Validation Study. Journal of Medical Internet Research, 28, e88665. https://doi.org/10.2196/88665

Vancouver

Pilgram L, Yang K, Beltran-Bless AA, Pond GR, Vandermeer L, Hilton J, et al. Transfer Learning and Machine Learning for Training Five-Year Survival Prognostic Models in Early Breast Cancer: Development and Validation Study. Journal of Medical Internet Research. 2026;28:e88665. doi:10.2196/88665

BibTeX
@article{Pilgram2026,
  author = {Pilgram, Lisa and Yang, Kai and Beltran-Bless, Ana-Alicia and Pond, Gregory R and Vandermeer, Lisa and Hilton, John and Savard, Marie-France and LeBlanc, Andreanne and Shepherd, Lois and Chen, Bingshu and Bartlett, John Ms and Taylor, Karen J and Bayani, Jane and Barker, Sarah and Spears, Melanie and van der Velde, Cornelis Jh and Meershoek-Klein Kranenbarg, Elma and Dirix, Luc and Mallon, Elizabeth and Hasenburg, Annette and Markopoulos, Christos and Juwara, Lamin and Dankar, Fida K and Clemons, Mark and El Emam, Khaled},
  title = {Transfer Learning and Machine Learning for Training Five-Year Survival Prognostic Models in Early Breast Cancer: Development and Validation Study},
  journal = {Journal of Medical Internet Research},
  year = {2026},
  volume = {28},
  pages = {e88665},
  doi = {10.2196/88665}
}
Start for free →