Image: © senengmotret/Stock.adobe.com
With time being of the essence and lives at risk, data science has been recruited to help find treatments for Covid-19 and solutions to problems posed by the pandemic.
In the same way that the data we generate online is seen as a highly valuable resource for tech companies, vast quantities of medical data is seen as potentially holding answers to a number of science’s greatest questions.
This can range from a start-up using AI to search for peptides that could benefit human health, to a multinational pharma giant looking to gain access to huge quantities of genetic data to create future drugs.
Now, however, data science and AI are being recruited to help solve the health crisis of a generation: Covid-19. While there is still much to learn about SARS-CoV2 (the novel coronavirus) and the disease it causes, the groundwork needed for scientists to find treatments and a potential vaccine is being laid down right now.
Just recently, Nvidia announced plans to build what it believes will be the UK’s “most power supercomputer”, which will be used to advance AI research in healthcare. The company said that the computer will be available to healthcare researchers in the UK who want to use AI to solve pressing medical challenges, including those presented by Covid-19.
Also, case numbers and medical data gathered across the world have been used to compile in-depth charts and maps tracking Covid-19, helping researchers and the general public to understand how it is spreading.
Data science has also helped us better trace and diagnose the novel coronavirus. In July, one such data science breakthrough was revealed by a team of University of Oxford scientists. In a pre-print paper, the team revealed Curial AI, a test using AI that can rapidly screen for Covid-19 in patients arriving in emergency departments.
Image: © Syda Productions/Stock.adobe.com
‘Unprecedented demand’ on hospital informatics
Currently, testing for Covid-19 is by a molecular analysis of a nose and throat swab, which typically has a turnaround time of 12 to 48 hours and requires specialist equipment and staff.
However, the AI developed in the Oxford study gives a near real-time prediction of a patient’s Covid-19 status by assessing data routinely collected during the first hour in emergency departments, such as blood tests and vital signs, to determine the chance of a patient testing positive for the coronavirus.
The numbers involved in the programme so far have been large, to say the least. As of July, its two early-detection models were able to identify Covid-19 from lab tests, blood gas and vital signs in 115,394 emergency presentations and 72,310 admissions to hospital.
The project is led by Dr Andrew Soltan, a UK National Institute of Health Research academic clinical fellow at the John Radcliffe Hospital in Oxford.
“During the first wave, there was unprecedented demand on hospital informatics departments for data relating to the Covid-19 pandemic,” Soltan told Siliconrepublic.com.
“Thanks to strong collaborative links between the hospital, the Institute for Biomedical Engineering and the Big Data Institute in Oxford, alongside the wonderful work of so many colleagues and patients, we were able to gain access to the necessary data for this work.
“As a collaboration between clinicians and experts in machine learning with healthcare data, we had access to a huge wealth of expertise to tackle the challenges of developing and validating the Curial AI rapidly.”
Really pleased to share our #AI powered screening-test for #COVID19 in patients presenting to hospital. Rapid test result within 1hr at front door of hospitals using routine bloods & vitalshttps://t.co/zMIVpNR9JG#NHS #MedTwitter #healthcare #digitalhealth #HealthTech
— Andrew Soltan (@andrewsoltan) July 9, 2020
Needing a little more time
Soltan said that AI’s ability to sift through huge quantities of data could play an integral part in our efforts to both control and better understand Covid-19. However, while broader steps to use the power of this tech can be made quickly, refinement of these tools may be a slower process.
“We are seeing developments that previously would have taken decades now occurring in months, but a little more time is needed before the prime time of AI in Covid-19,” he said.
“The pace of development is breathtaking, but the expectations for AI-driven innovation are also high. Passing verdict now would be too soon and I am sure that, with time, we will see the full benefits of AI in fighting the pandemic.”
Back in July, the Curial AI team said that its next hope was to take the tech out of the lab and deploy it in a clinical setting to bolster testing numbers at a time of rising case numbers in many countries.
According to Soltan, the team is now refining a web-based user interface for clinicians to use the platform.
“We’re working alongside key clinical stakeholders and leaders to determine how best Curial AI can fit in to the clinical pathway, safely and without biases,” he said. “A key part of the work is developing a safe clinical governance framework, and audit pathway to allow us to monitor performance of Curial AI in real time and ensure that its use benefits patient safety.”
