On 13 November 2024, Dr. Khaled El Emam shared his top dos and don'ts of publishing studies utilizing machine learning models.
Training machine learning models on health data requires the analyst to pay attention to many details, and they are all important. The analysis and modeling details pertain to how the data is prepared, the model is trained and evaluated, the results interpreted, and the methods and results reported. As an author and an editor of the JMIR AI journal, Khaled has been involved in and reviewed many such studies. This presentation is an attempt to distill key learnings about how to train useful prognostic models and avoid common pitfalls that reduce the value of the model (even if the results may seem good at first glance), and make the resultant manuscript more difficult to publish because of subtle methodology problems. The objective is to highlight to analysts and authors key considerations, some obvious and some may not be, that will improve the utility, validity, and usefulness of the models that are developed.
Bio
In addition to being co-editor-in-chief of the JMIR AI journal, Dr. Khaled El Emam is the Canada Research Chair (Tier 1) in Medical AI at the University of Ottawa, where he is a Professor in the School of Epidemiology and Public Health. He is also a Senior Scientist at the Children’s Hospital of Eastern Ontario Research Institute and Director of the multi-disciplinary Electronic Health Information Laboratory, conducting research on privacy enhancing technologies to enable the sharing of health data for secondary purposes, including synthetic data generation and de-identification methods. Khaled has also recently taken up the post of Scholar-in-Residence at the Office of the Information and Privacy Commissioner of Ontario (IPC).