AI-assisted diagnosis and prognostication in COVID-19 [COVID-19]

  • Research type

    Research Study

  • Full title

    Artificial intelligence-assisted diagnosis and prognostication in COVID-19 using electrocardiograms and imaging (AI-COV-19)

  • IRAS ID

    283210

  • Contact name

    Fu Siong Ng

  • Contact email

    f.ng@imperial.ac.uk

  • Sponsor organisation

    Imperial College London

  • Duration of Study in the UK

    1 years, 11 months, 18 days

  • Research summary

    Research Summary

    Coronavirus Disease 2019 (COVID-19) has been widespread worldwide since December 2019. It is highly contagious, and severe cases can lead to acute respiratory distress or multiple organ failure. On 11 March 2020, the WHO made the assessment that COVID-19 can be characterised as a pandemic. \n\nWith the development of machine learning, deep learning based artificial intelligence (AI) technology has demonstrated tremendous success in the field of medical data analysis due to its capacity of extracting rich features from imaging and complex clinical datasets.\n\nIn this study, we aim to use clinical data collected as part of routine clinical care (heart tracings, X-rays and CT scans) to train artificial intelligence and machine learning algorithms, to accurately predict the course of disease in patients with Covid-19 infection, using these datasets.

    Summary of Results

    At start of the pandemic in March 2020, it was difficult to identify patients who were at higher risk of getting severe COVID19 infection. Before the pandemic, recordings of the heart beat (ECG) had been used to help identify people at higher risk of health complications. We therefore used the ECGs of patients admitted to hospital with COVID19 to try to identify individuals who would go on to require specialist or intensive care during their hospital stay. We used computer-based techniques, called machine learning, on patients’ ECGs to do this. We were unable to predict the patients that would go on to require intensive care or become seriously unwell from coronavirus from their ECGs. This was because we only had a small number of ECGs and patient information was limited. In summary we were unable to identify the patients with coronavirus infection who would require intensive care or have a more serious course infection during their hospital stay based on recordings of their heart beat.

  • REC name

    London - Hampstead Research Ethics Committee

  • REC reference

    20/HRA/2467

  • Date of REC Opinion

    18 May 2020

  • REC opinion

    Favourable Opinion