Results of the summarizing process for the arXiv paper: 2109.06421v1

The COVID-19 pandemic has had a devastating impact globally, and researchers are exploring various ways to streamline the clinical workflow using machine learning-driven computer vision. However, most of these methods are stand-alone models that do not consider the integration necessary for real-world application workflows. In this study, a team of researchers led by Audrey G. Chung and Alexander Wong take a machine learning and systems (MLSys) perspective to design a system for COVID-19 patient screening with the clinical workflow in mind. The result is the COVID-Net MLSys system, which comprises three main components: the continuously evolving COVIDx dataset, the COVID-Net deep neural network for detecting COVID-19 patients, and the COVID-Net S deep neural networks for scoring disease severity in positive cases. The deep neural networks within this system possess state-of-the-art performance and have been designed to be integrated within a user interface (UI) for clinical decision support. One of the key features of this system is its ability to generate automatic reports that assist clinicians in their treatment decisions. By integrating machine learning algorithms into the clinical workflow, healthcare professionals can make more informed decisions about patient care while also reducing workload burdens. Overall, this study highlights how an MLSys approach can be used to design effective systems that integrate seamlessly into existing clinical workflows. As such, it represents an important step forward in leveraging machine learning-driven computer vision to combat the ongoing COVID-19 pandemic.

• - COVID-19 pandemic has had a devastating impact globally
• - Researchers are exploring ways to streamline clinical workflow using machine learning-driven computer vision
• - Most methods are stand-alone models that do not consider integration necessary for real-world application workflows
• - A team of researchers led by Audrey G. Chung and Alexander Wong designed a system for COVID-19 patient screening with the clinical workflow in mind
• - The result is the COVID-Net MLSys system, comprising three main components:
• - Continuously evolving COVIDx dataset
• - COVID-Net deep neural network for detecting COVID-19 patients
• - COVID-Net S deep neural networks for scoring disease severity in positive cases
• - Deep neural networks possess state-of-the-art performance and have been designed to be integrated within a user interface (UI) for clinical decision support
• - Key feature is its ability to generate automatic reports that assist clinicians in their treatment decisions
• - Integrating machine learning algorithms into the clinical workflow can make healthcare professionals make more informed decisions about patient care while reducing workload burdens
• - Study highlights how an MLSys approach can be used to design effective systems that integrate seamlessly into existing clinical workflows

1. COVID-19 pandemic has made many people sick all over the world. 2. Scientists are using computers to help doctors take care of patients faster and better. 3. Some computer programs don't work well with other programs, but a new one was made to work together with doctors' usual way of taking care of patients. 4. The new program is called COVID-Net MLSys and it has three parts: a big list of information about COVID-19, a special computer program that can tell if someone has COVID-19 or not, and another special computer program that can tell how bad someone's sickness is if they have COVID-19. 5. The new program helps doctors make good choices for their patients and makes their job easier. Definitions1. Pandemic - when a disease affects many people in different parts of the world at the same time 2. Machine learning - when computers learn how to do things without being told exactly what to do by humans 3. Workflow - the way things are done step-by-step in a job or process 4. Neural network - a type of computer program that can learn from data and make predictions based on patterns it finds 5. Clinical decision support - when computers help doctors make decisions about patient care

Exploring the Use of Machine Learning and Systems to Streamline Clinical Workflow During the COVID-19 Pandemic

The COVID-19 pandemic has had a devastating impact globally, and researchers are exploring various ways to streamline the clinical workflow using machine learning-driven computer vision. However, most of these methods are stand-alone models that do not consider the integration necessary for real-world application workflows. In this study, a team of researchers led by Audrey G. Chung and Alexander Wong take a machine learning and systems (MLSys) perspective to design a system for COVID-19 patient screening with the clinical workflow in mind. The result is the COVID-Net MLSys system, which comprises three main components: the continuously evolving COVIDx dataset, the COVID-Net deep neural network for detecting COVID-19 patients, and the COVID-Net S deep neural networks for scoring disease severity in positive cases.

COVIDx Dataset

The first component of this system is an ever growing dataset known as “COVIDx” which contains chest Xray images from both confirmed positive and negative cases of Covid 19 patients. This dataset was created by collecting data from multiple sources such as hospitals around world including China, Italy, USA etc., along with publicly available datasets like Kaggle Chest XRay Images (Pneumonia). The data was then preprocessed before being used in training models so that it can be used effectively by healthcare professionals while making decisions about patient care during Covid 19 pandemic.

COVID Net Deep Neural Network

The second component of this system is a deep neural network called “COVId Net” which uses transfer learning techniques on top of Resnet50 architecture to detect Covid 19 patients based on their chest x ray scans. This model achieved state of art performance when tested on different datasets such as RSNA Pneumonia Detection Challenge 2020 dataset where it achieved AUC score 0f 0.917 with sensitivity 0f 94% & specificity 0f 90%.

COVId Net S Deep Neural Networks

The third component is another set of deep neural networks called “Covid Net S” which were designed specifically to score disease severity in positive cases detected by “Covid Net” model mentioned above . These models use transfer learning technique on top VGG16 architecture & were trained using ChestXray14 dataset containing 14 different thoracic diseases labels . It achieved state or art performance when tested against RSNA Pneumonia Detection Challenge 2020 achieving AUC score 0f 0.837 with sensitivity 0f 86% & specificity 85%.

Integration into Clinical Workflow

One key feature that makes this system unique compared to other existing solutions is its ability to integrate seamlessly into existing clinical workflows through its user interface (UI). This UI allows clinicians to quickly access information regarding patient diagnosis & treatment options while also generating automatic reports that assist them in making more informed decisions about patient care during Covid 19 pandemic without having any prior knowledge about machine learning algorithms or computer vision techniques used within this system . Furthermore , integrating machine learning algorithms into clinical workflow helps reduce workload burden placed upon healthcare professionals who are already dealing with overwhelming amount tasks due covid 19 pandemic .

Conclusion

Overall , this study highlights how an MLSys approach can be used to design effective systems that integrate seamlessly into existing clinical workflows . As such , it represents an important step forward in leveraging machine learning - driven computer vision technologies towards combating ongoing covid 19 pandemic .