تقرير
Instability in clinical risk stratification models using deep learning
| العنوان: | Instability in clinical risk stratification models using deep learning |
|---|---|
| المؤلفون: | Lopez-Martinez, Daniel, Yakubovich, Alex, Seneviratne, Martin, Lelkes, Adam D., Tyagi, Akshit, Kemp, Jonas, Steinberg, Ethan, Downing, N. Lance, Li, Ron C., Morse, Keith E., Shah, Nigam H., Chen, Ming-Jun |
| سنة النشر: | 2022 |
| المجموعة: | Computer Science |
| مصطلحات موضوعية: | Computer Science - Machine Learning, Computer Science - Artificial Intelligence |
| الوصف: | While it has been well known in the ML community that deep learning models suffer from instability, the consequences for healthcare deployments are under characterised. We study the stability of different model architectures trained on electronic health records, using a set of outpatient prediction tasks as a case study. We show that repeated training runs of the same deep learning model on the same training data can result in significantly different outcomes at a patient level even though global performance metrics remain stable. We propose two stability metrics for measuring the effect of randomness of model training, as well as mitigation strategies for improving model stability. Comment: Accepted for publication in Machine Learning for Health (ML4H) 2022 |
| نوع الوثيقة: | Working Paper |
| URL الوصول: | http://arxiv.org/abs/2211.10828 |
| رقم الأكسشن: | edsarx.2211.10828 |
| قاعدة البيانات: | arXiv |
| الوصف غير متاح. |