This is complementary to our systematic review of ML in subtype definition and risk prediction for CVD, showing lack of validation and implementation studies. We suggested a framework to help address development, validation and impact of ML models. https:/
5,720 followers
5,720 followers
I agree-excellent thread. In CVD, we showed lack of validation and evaluation of machine learning for subtype classification and risk prediction. We need to address reproducibility but also generalisability. https://t.co/FUGyEi97Lq
5,720 followers
We used our framework for implementing ML-development, validation & impact https://t.co/FUGyEi97Lq across 2 independent,population-based datasets, using 4 ML methods for subtyping &risk prediction with 645 factors for HF. We identified & valid
68 followers
RT @BMCMedicine: A systematic review by @amibanerjee1 and colleagues investigating the clinical utility and validity of #MachineLearning in…