You are not logged in and only seeing 7 days of articles. Please sign up or login to view more
Friday 11 June 2021

Fujitsu helps develop DHSC's Cough-in-a-Box

Fujitsu logoIn September last year, researchers at Massachusetts Institute of Technology (MIT) published a paper in the IEEE Journal of Engineering in Medicine and Biology discussing the potential for AI-powered diagnosis of COVID-19 using audio recordings of a forced-cough. From a dataset of 5,320 subjects, the model achieved sensitivity (proportion of true positives) of 98.5%, with a specificity (proportion of true negatives) of 94.2%. This included correctly identifying 100% of asymptomatic subjects with the virus.

The Department of Health and Social Care (DHSC) is currently running a programme—Cough-in-a-Box—to investigate whether a similar approach could be utilised as part of the COVID-19 testing regime in the UK. In December 2020, DHSC awarded the Fujitsu Novo Innovation Service a contract via G-Cloud to supply its Smartforms low-code form automation technology. The company is using the platform to capture forced-cough audio recordings for subsequent processing. Fujitsu turned around the design of the form in 48 hours and has now converted the application to be cloud native, enabling it to utilise a scalable, multi-cloud model. It is using a combination of Microsoft Azure and Amazon Web Services (AWS) technology as part of the implementation, including Azure hosting and AWS Lambda serverless compute service. It is also providing restful APIs to facilitate the creation of a project dashboard.  

In May 2021, Fujitsu secured a new contract with DHSC to build upon the initial implementation through an Alpha extension phase, which is scheduled to run for a three month period. This is designed to enable DHSC to target a larger pool of up to 40,000 respondents.

The Joint Biosecurity Centre is currently working with the Alan Turing Institute to assess the feasibility of the programme. If the efficacy of using vocal biomarkers is proven, its role in the fight against the pandemic, and potentially other diseases, is huge. The MIT researchers who conducted the original study highlighted its use as a daily screening tool, for test pooling, and for extending screening to areas where access to other tests is restricted.

Although the monetary value of Fujitsu's contracts are relatively low (just £118k), if the programme does prove to be effective, the scale and importance of this work could be enormous. It also provides another example of how innovative technology can be used to improve humanitarian action.

Posted by: Dale Peters at 09:04

Tags: nhs   automation   innovation   AI   low-code   healthcare   covid-19  

Twitter   Facebook   LinkedIn   Email article link

© TechMarketView LLP 2007-2021: Unauthorised reproduction prohibited see full Terms and Conditions.