Timezone: »

Improving ECG-based COVID-19 diagnosis and mortality predictions using pre-pandemic medical records at population-scale
Weijie Sun · Sunil Vasu Kalmady · Nariman Sepehrvand · Luan Chu · Zihan Wang · Amir Salimi · Abram Hindle · Russell Greiner · Padma Kaul

Fri Dec 02 09:30 AM -- 10:30 AM (PST) @
Event URL: https://openreview.net/forum?id=OelsHY4vnn »

Pandemic outbreaks such as COVID-19 occur unexpectedly, and need immediate action due to their potential devastating consequences on global health. Point-of-care routine assessments such as electrocardiogram (ECG), can be used to develop prediction models for identifying individuals at risk. However, there is often too little clinically-annotated medical data, especially in early phases of a pandemic, to develop accurate prediction models. In such situations, historical pre-pandemic health records can be utilized to estimate a preliminary model, which can then be fine-tuned based on limited available pandemic data. This study shows this approach -- pre-train deep learning models with pre-pandemic data -- can work effectively, by demonstrating substantial performance improvement over three different COVID-19 related diagnostic and prognostic prediction tasks. Similar transfer learning strategies can be useful for developing timely artificial intelligence solutions in future pandemic outbreaks.

Author Information

Weijie Sun (Canadian VIGOUR Centre, University of Alberta)
Sunil Vasu Kalmady (University of Alberta)
Sunil Vasu Kalmady

Adjunct Professor, Faculty of Science - Computing Science, University of Alberta. Senior Machine Learning Specialist, Faculty of Medicine & Dentistry - Medicine Dept, University of Alberta. My professional goal is to advance the options for personalized treatment of complex medical disorders via application of machine learning and data science. To this end, I have undergone extensive post-doctoral training in Alberta Machine Intelligence Institute (AMII), one of Canada’s three artificial intelligence centers of excellence. My qualifications are complemented by over 5 years of research experience in developing, evaluating and deploying machine learning models using various structured and unstructured real-word healthcare datasets. In my current position as an adjunct professor of computing science and a senior machine learning specialist, I focus on developing learning tools to predict prognostic outcomes in cardiovascular diseases using electronic medical records, electrocardiograms and echocardiograms at the population scale. In the past, I have developed successful AI methods to identify and predict specific symptom clusters and treatment responses in psychiatric disorders such as Schizophrenia and OCD using multimodal imaging, which have been published in distinguished journals such as Nature Schizophrenia and featured in several news reports.

Nariman Sepehrvand
Luan Chu
Zihan Wang (University of Alberta)
Amir Salimi (University of Alberta)
Abram Hindle (University of Alberta)
Russell Greiner (University of Alberta)
Padma Kaul (University of Alberta)

More from the Same Authors