Please upgrade your browser

We built this website using the latest browser technologies to deliver the very best experience.

This makes the site run faster and easier to use. Unfortunately, your browser is out of date and will not support some of these technologies.

We recommend that you use a modern browser such as Google Chrome or Microsoft Edge to view this website.

Download ChromeDownload Microsoft Edge
Case study

Sensyne Health aids National Health Service in the COVID-19 struggle with Microsoft HPC and AI technologies

November 2, 2021

When the United Kingdom’s National Health Service envisioned an app that could read COVID-19 tests at scale with an accuracy and speed not possible by human readers, Sensyne Health answered the call.

The app connects users to a web service that uses AI to deliver an Azure high-performance computing (HPC) solution that’s adeptly orchestrated with NVIDIA Triton Inference Server on graphics processing units. The result: the incredibly fast and accurate MagnifEye solution.

"Our initial benchmark of 500 tests per second is impressive. But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure."

Alan Payne, Chief Information and Product Officer, Sensyne Health

Suppose you could know with a very high degree of accuracy and in just seconds if you have COVID-19. How might that information affect the spread of the virus? The National Health Service (NHS), the United Kingdom’s publicly funded healthcare system, asked that question as it dug into the extraordinary struggle to control the COVID-19 outbreak. It wanted to overcome the issues of response time and the fallibility of reading Lateral Flow Tests (LFTs) with the human eye.

Sensyne Health came up with an answer: its MagnifEye solution, an app for mobile devices that use a device’s camera to capture the LFT stick image and analyze it in tenths of seconds with extraordinary accuracy. It was a matter of combining the right technologies: high-performance computing (HPC) on Microsoft Azure and Azure Cognitive Services, along with NVIDIA graphics processing unit (GPU) acceleration and NVIDIA Triton Inference Server.

Exercising a novel approach to the thorniest health problems

Until recently, analyzing enough data to solve some of the biggest issues in healthcare was a dream—timely analysis was an insurmountable barrier. But current AI and machine learning technology is changing that.

Sensyne Health was built on a unique premise: the ethical use of data with AI technology to combat threats to human health. “We’re an ethically driven safe haven and docking station between healthcare data and pharmaceutical research,” says Alan Payne, Chief Information and Product Officer at Sensyne Health. “We work only with fully anonymized data that has a clear and definitive purpose.” The company uses anonymized data from the United Kingdom’s NHS and United States’ healthcare systems to build rich datasets—more than 22.5 million records as of this writing. That data holds incredible potential to combat disease.

Also unique is Sensyne Health’s partnership with the NHS, which is not only a Sensyne Health stakeholder but a shareholder, too, owning equity in Sensyne and receiving a share of revenues to reinvest back into patient care. Sensyne Health was an obvious candidate when the NHS needed proposals to develop a mobile web service that could improve on LFT readings and make COVID-19 testing more accurate and convenient.

As one of three finalists from an initially crowded field, the company received a dataset to use as a prototype. The objective for the finalists: provide a way to read LFTs at 95 percent specificity and 95 percent sensitivity (false positives and false negatives). The stakes for such a tool are high. “If you’re doing a million tests and 5 percent are false positives, you’re going to create a huge compression on the health system,” explains Payne. Likewise, false negatives allow the virus to spread.

The difficulty stems from precisely reading the horizontal lines on the LFT stick, which resemble the thin lines on a pregnancy test. No two people see in necessarily the same way; vision acuity differs, and human brains receive and interpret images in varying ways. Sensyne Health’s plan was to train its algorithm on hundreds of thousands—even millions—of images. It just needed the right technology to analyze that vast sea of data.

Putting the chemistry between Microsoft and NVIDIA to work

A Microsoft Partner Network member, Sensyne Health has extensive experience with the Azure platform. It used Azure services to run data cost-effectively and in a highly secure manner for MagnifEye, supplying compute power with Azure HPC. “Microsoft has invested heavily in the cybersecurity aspect of the Azure platform,” says Payne. Sensyne needed multiple certifications for the product to comply with the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom, the General Data Protection Regulation (GDPR) in the European Union, and the Health Insurance Portability and Accountability Act of 1996 (HIPAA) in the United States. “The regulated products we build must follow strict medical device and software guidelines,” he explains. “With the better modeling and traceability through Azure logging mechanisms, we can identify all the source data that makes up the end product, which is key to MHRA requirements—we need to prove that auditability. For that reason alone, using Azure saved a lot of product development time.”

Sensyne Health used Azure Kubernetes Service (AKS) with NVIDIA V100 GPUs to provide the cost-effective scalability and speed it needed to deliver lightning-fast results. “We can automate spinning up a Kubernetes node literally in seconds,” explains Payne. “That’s pivotal because most requests for results occur between 7:30 and 8:30 in the morning for people who need to present their results to enter their workplaces or schools. Other spikes occur in the evening and at midday.” The company further speeds that process by using Azure Machine Learning to predictively spin up that infrastructure. Azure continuous integration and continuous delivery (CI/CD) expedites automated model updates. The millisecond response time and instant scalability of Azure Cosmos DB made it the perfect vehicle for test requests and results.

Figure 1. Managing LFT testing spikes at different times of the day using Azure Machine Learning within AKS combined with Azure continuous integration and continuous delivery for expediting automated model updates. Azure Cosmos DB response and scalability capabilities made it the perfect vehicle for LFT test requests and results.

The solution would also have to deploy at scale to multiple concurrent end users. NVIDIA supplied the other pieces of that puzzle, Triton Inference Server, which runs on NVIDIA V100 GPUs in AKS clusters to infer results for multiple test images in parallel—initially meeting the NHS standard of 500 tests per second. “Our initial benchmark of 500 tests per second is impressive,” asserts Payne. “But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure.” The machine response time to deliver the result for the user was a tenth of a second—effectively in real time for the user. “The performance we achieved in terms of the sensitivity of the algorithm is fantastic by any academic standard,” he adds. “With the Microsoft technology stack, we have incredible performance that we couldn’t have gotten otherwise.”

Figure 2. Sensyne Health achieved 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs and on AKS, providing the cost-effective scalability and speed needed to deliver lightning-fast results.

As an added bonus, Triton Inference Server was four times as fast as another comparably priced solution. “NVIDIA provided a price/performance ratio sea change,” says Payne. “It was remarkable throughput.” Triton Inference Server continues to perform impressively amidst rising data volumes. “We have three servers that service hundreds of thousands of tests a day, and we can easily scale to 30 to 40 million a week,” he adds. “This technology gives governments an enormous opportunity to get a handle on how the virus is performing and adapting and in which locations.”

“Seeing” with AI

Much of the project focused on anomaly detection to help ensure that the software accurately perceived the control and test lines of the LFT stick window and correctly interpreted the results. The Sensyne Health team trained the anomaly framework for several weeks, eventually creating nine anomaly checks that run in parallel. Those anomaly checks tell the user that the image isn’t in focus or is too dark or too light.

The complicating factor? That same training needs to be repeated for every test that a different manufacturer creates. “We’ve now trained the models on tests from more than six manufacturers, but we still retain a 99.6 percent false positive rate,” says Payne. “That’s astounding considering the variety of pictures we get from real-world use and the volume of images.”

Sensyne Health runs 10 models for each processed test—one to complete the LFT reading and the other nine for anomaly detection. That intensive processing brought up the question of whether to use central processing units (CPUs) or graphics processing units (GPUs) for inference support. Sensyne Health decided to use NVIDIA NC6 v3 CPUs with the NVIDIA V100 GPU. “The hybrid approach turned out to be far and away the most cost-effective, and it’s very straightforward to operate,” reports Payne. “We effectively run multiple models simultaneously to deliver an accurate result and identify the test manufacturer, giving us ‘one algorithm to rule them all.’“

The company creates rich data graphics with Power BI that NHS officials and elected officials can use for fast policy decisions. MagnifEye soon made a difference in the spread of COVID-19 in the United Kingdom thanks to the fast, accurate readings it provides and the invaluable information it makes available to policy makers.

Saving lives, enhancing healthcare resources for everyone

Andrew Beggs, Consultant Surgeon and Professor of Surgery and Cancer Genetics at the University of Birmingham and University Hospitals Birmingham and advisor to the UK Department of Health and Social Care, has been leading COVID-19 testing for NHS staff in his region. He emphasizes that end user interpretation was a major issue—until MagnifEye. “MagnifEye breaks the chain of infection by reducing the spread of COVID-19,” he says. “We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS.”

He’s optimistic about the impact that the technologies behind MagnifEye can bring to healthcare. “Machine learning and AI opens up a world of potential for clinicians in the NHS, including rapid screening of chest X-rays for COVID-19, increased accuracy of staging of cancer scans, and earlier detection of patient deterioration in a ward-based environment,” he explains. “We also hope to reduce delays in cancer diagnosis by deploying the technology to a triage-based environment where patients with cancer symptoms can be automatically and rapidly referred into the right investigation.”

Payne is adamant that without exceptional teamwork, the high-performance app couldn’t have been developed in such a tight time frame. “The cultural fit of the teams from Microsoft, NVIDIA, and Sensyne Health made MagnifEye possible,” says Payne. “We were in the middle of a pandemic, but everyone pulled together to rise to a great purpose. We’re grateful to Microsoft teams both in the United Kingdom and the United States, working with NVIDIA to bring their technologies together to help us create the solution we’re so proud of. We couldn’t have done it without them.”

"We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS."

Andrew Beggs: Consultant Surgeon and Professor of Surgery and Cancer Genetics
University of Birmingham and University Hospitals Birmingham, and advisor to the UK Department of Health and Social Care

View the case study on the Microsoft customer stories page: https://customers.microsoft.com/en-us/story/1430377058968477645-sensyne-health-partner-professional-services-azure-hpc

Case study

Sensyne Health aids National Health Service in the COVID-19 struggle with Microsoft HPC and AI technologies

November 2, 2021

When the United Kingdom’s National Health Service envisioned an app that could read COVID-19 tests at scale with an accuracy and speed not possible by human readers, Sensyne Health answered the call.

The app connects users to a web service that uses AI to deliver an Azure high-performance computing (HPC) solution that’s adeptly orchestrated with NVIDIA Triton Inference Server on graphics processing units. The result: the incredibly fast and accurate MagnifEye solution.

"Our initial benchmark of 500 tests per second is impressive. But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure."

Alan Payne, Chief Information and Product Officer, Sensyne Health

Suppose you could know with a very high degree of accuracy and in just seconds if you have COVID-19. How might that information affect the spread of the virus? The National Health Service (NHS), the United Kingdom’s publicly funded healthcare system, asked that question as it dug into the extraordinary struggle to control the COVID-19 outbreak. It wanted to overcome the issues of response time and the fallibility of reading Lateral Flow Tests (LFTs) with the human eye.

Sensyne Health came up with an answer: its MagnifEye solution, an app for mobile devices that use a device’s camera to capture the LFT stick image and analyze it in tenths of seconds with extraordinary accuracy. It was a matter of combining the right technologies: high-performance computing (HPC) on Microsoft Azure and Azure Cognitive Services, along with NVIDIA graphics processing unit (GPU) acceleration and NVIDIA Triton Inference Server.

Exercising a novel approach to the thorniest health problems

Until recently, analyzing enough data to solve some of the biggest issues in healthcare was a dream—timely analysis was an insurmountable barrier. But current AI and machine learning technology is changing that.

Sensyne Health was built on a unique premise: the ethical use of data with AI technology to combat threats to human health. “We’re an ethically driven safe haven and docking station between healthcare data and pharmaceutical research,” says Alan Payne, Chief Information and Product Officer at Sensyne Health. “We work only with fully anonymized data that has a clear and definitive purpose.” The company uses anonymized data from the United Kingdom’s NHS and United States’ healthcare systems to build rich datasets—more than 22.5 million records as of this writing. That data holds incredible potential to combat disease.

Also unique is Sensyne Health’s partnership with the NHS, which is not only a Sensyne Health stakeholder but a shareholder, too, owning equity in Sensyne and receiving a share of revenues to reinvest back into patient care. Sensyne Health was an obvious candidate when the NHS needed proposals to develop a mobile web service that could improve on LFT readings and make COVID-19 testing more accurate and convenient.

As one of three finalists from an initially crowded field, the company received a dataset to use as a prototype. The objective for the finalists: provide a way to read LFTs at 95 percent specificity and 95 percent sensitivity (false positives and false negatives). The stakes for such a tool are high. “If you’re doing a million tests and 5 percent are false positives, you’re going to create a huge compression on the health system,” explains Payne. Likewise, false negatives allow the virus to spread.

The difficulty stems from precisely reading the horizontal lines on the LFT stick, which resemble the thin lines on a pregnancy test. No two people see in necessarily the same way; vision acuity differs, and human brains receive and interpret images in varying ways. Sensyne Health’s plan was to train its algorithm on hundreds of thousands—even millions—of images. It just needed the right technology to analyze that vast sea of data.

Putting the chemistry between Microsoft and NVIDIA to work

A Microsoft Partner Network member, Sensyne Health has extensive experience with the Azure platform. It used Azure services to run data cost-effectively and in a highly secure manner for MagnifEye, supplying compute power with Azure HPC. “Microsoft has invested heavily in the cybersecurity aspect of the Azure platform,” says Payne. Sensyne needed multiple certifications for the product to comply with the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom, the General Data Protection Regulation (GDPR) in the European Union, and the Health Insurance Portability and Accountability Act of 1996 (HIPAA) in the United States. “The regulated products we build must follow strict medical device and software guidelines,” he explains. “With the better modeling and traceability through Azure logging mechanisms, we can identify all the source data that makes up the end product, which is key to MHRA requirements—we need to prove that auditability. For that reason alone, using Azure saved a lot of product development time.”

Sensyne Health used Azure Kubernetes Service (AKS) with NVIDIA V100 GPUs to provide the cost-effective scalability and speed it needed to deliver lightning-fast results. “We can automate spinning up a Kubernetes node literally in seconds,” explains Payne. “That’s pivotal because most requests for results occur between 7:30 and 8:30 in the morning for people who need to present their results to enter their workplaces or schools. Other spikes occur in the evening and at midday.” The company further speeds that process by using Azure Machine Learning to predictively spin up that infrastructure. Azure continuous integration and continuous delivery (CI/CD) expedites automated model updates. The millisecond response time and instant scalability of Azure Cosmos DB made it the perfect vehicle for test requests and results.

Figure 1. Managing LFT testing spikes at different times of the day using Azure Machine Learning within AKS combined with Azure continuous integration and continuous delivery for expediting automated model updates. Azure Cosmos DB response and scalability capabilities made it the perfect vehicle for LFT test requests and results.

The solution would also have to deploy at scale to multiple concurrent end users. NVIDIA supplied the other pieces of that puzzle, Triton Inference Server, which runs on NVIDIA V100 GPUs in AKS clusters to infer results for multiple test images in parallel—initially meeting the NHS standard of 500 tests per second. “Our initial benchmark of 500 tests per second is impressive,” asserts Payne. “But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure.” The machine response time to deliver the result for the user was a tenth of a second—effectively in real time for the user. “The performance we achieved in terms of the sensitivity of the algorithm is fantastic by any academic standard,” he adds. “With the Microsoft technology stack, we have incredible performance that we couldn’t have gotten otherwise.”

Figure 2. Sensyne Health achieved 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs and on AKS, providing the cost-effective scalability and speed needed to deliver lightning-fast results.

As an added bonus, Triton Inference Server was four times as fast as another comparably priced solution. “NVIDIA provided a price/performance ratio sea change,” says Payne. “It was remarkable throughput.” Triton Inference Server continues to perform impressively amidst rising data volumes. “We have three servers that service hundreds of thousands of tests a day, and we can easily scale to 30 to 40 million a week,” he adds. “This technology gives governments an enormous opportunity to get a handle on how the virus is performing and adapting and in which locations.”

“Seeing” with AI

Much of the project focused on anomaly detection to help ensure that the software accurately perceived the control and test lines of the LFT stick window and correctly interpreted the results. The Sensyne Health team trained the anomaly framework for several weeks, eventually creating nine anomaly checks that run in parallel. Those anomaly checks tell the user that the image isn’t in focus or is too dark or too light.

The complicating factor? That same training needs to be repeated for every test that a different manufacturer creates. “We’ve now trained the models on tests from more than six manufacturers, but we still retain a 99.6 percent false positive rate,” says Payne. “That’s astounding considering the variety of pictures we get from real-world use and the volume of images.”

Sensyne Health runs 10 models for each processed test—one to complete the LFT reading and the other nine for anomaly detection. That intensive processing brought up the question of whether to use central processing units (CPUs) or graphics processing units (GPUs) for inference support. Sensyne Health decided to use NVIDIA NC6 v3 CPUs with the NVIDIA V100 GPU. “The hybrid approach turned out to be far and away the most cost-effective, and it’s very straightforward to operate,” reports Payne. “We effectively run multiple models simultaneously to deliver an accurate result and identify the test manufacturer, giving us ‘one algorithm to rule them all.’“

The company creates rich data graphics with Power BI that NHS officials and elected officials can use for fast policy decisions. MagnifEye soon made a difference in the spread of COVID-19 in the United Kingdom thanks to the fast, accurate readings it provides and the invaluable information it makes available to policy makers.

Saving lives, enhancing healthcare resources for everyone

Andrew Beggs, Consultant Surgeon and Professor of Surgery and Cancer Genetics at the University of Birmingham and University Hospitals Birmingham and advisor to the UK Department of Health and Social Care, has been leading COVID-19 testing for NHS staff in his region. He emphasizes that end user interpretation was a major issue—until MagnifEye. “MagnifEye breaks the chain of infection by reducing the spread of COVID-19,” he says. “We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS.”

He’s optimistic about the impact that the technologies behind MagnifEye can bring to healthcare. “Machine learning and AI opens up a world of potential for clinicians in the NHS, including rapid screening of chest X-rays for COVID-19, increased accuracy of staging of cancer scans, and earlier detection of patient deterioration in a ward-based environment,” he explains. “We also hope to reduce delays in cancer diagnosis by deploying the technology to a triage-based environment where patients with cancer symptoms can be automatically and rapidly referred into the right investigation.”

Payne is adamant that without exceptional teamwork, the high-performance app couldn’t have been developed in such a tight time frame. “The cultural fit of the teams from Microsoft, NVIDIA, and Sensyne Health made MagnifEye possible,” says Payne. “We were in the middle of a pandemic, but everyone pulled together to rise to a great purpose. We’re grateful to Microsoft teams both in the United Kingdom and the United States, working with NVIDIA to bring their technologies together to help us create the solution we’re so proud of. We couldn’t have done it without them.”

"We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS."

Andrew Beggs: Consultant Surgeon and Professor of Surgery and Cancer Genetics
University of Birmingham and University Hospitals Birmingham, and advisor to the UK Department of Health and Social Care

View the case study on the Microsoft customer stories page: https://customers.microsoft.com/en-us/story/1430377058968477645-sensyne-health-partner-professional-services-azure-hpc

Case study

Sensyne Health aids National Health Service in the COVID-19 struggle with Microsoft HPC and AI technologies

Sensyne Health aids National Health Service in the COVID-19 struggle with Microsoft HPC and AI technologies

November 2, 2021

When the United Kingdom’s National Health Service envisioned an app that could read COVID-19 tests at scale with an accuracy and speed not possible by human readers, Sensyne Health answered the call.

The app connects users to a web service that uses AI to deliver an Azure high-performance computing (HPC) solution that’s adeptly orchestrated with NVIDIA Triton Inference Server on graphics processing units. The result: the incredibly fast and accurate MagnifEye solution.

"Our initial benchmark of 500 tests per second is impressive. But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure."

Alan Payne, Chief Information and Product Officer, Sensyne Health

Suppose you could know with a very high degree of accuracy and in just seconds if you have COVID-19. How might that information affect the spread of the virus? The National Health Service (NHS), the United Kingdom’s publicly funded healthcare system, asked that question as it dug into the extraordinary struggle to control the COVID-19 outbreak. It wanted to overcome the issues of response time and the fallibility of reading Lateral Flow Tests (LFTs) with the human eye.

Sensyne Health came up with an answer: its MagnifEye solution, an app for mobile devices that use a device’s camera to capture the LFT stick image and analyze it in tenths of seconds with extraordinary accuracy. It was a matter of combining the right technologies: high-performance computing (HPC) on Microsoft Azure and Azure Cognitive Services, along with NVIDIA graphics processing unit (GPU) acceleration and NVIDIA Triton Inference Server.

Exercising a novel approach to the thorniest health problems

Until recently, analyzing enough data to solve some of the biggest issues in healthcare was a dream—timely analysis was an insurmountable barrier. But current AI and machine learning technology is changing that.

Sensyne Health was built on a unique premise: the ethical use of data with AI technology to combat threats to human health. “We’re an ethically driven safe haven and docking station between healthcare data and pharmaceutical research,” says Alan Payne, Chief Information and Product Officer at Sensyne Health. “We work only with fully anonymized data that has a clear and definitive purpose.” The company uses anonymized data from the United Kingdom’s NHS and United States’ healthcare systems to build rich datasets—more than 22.5 million records as of this writing. That data holds incredible potential to combat disease.

Also unique is Sensyne Health’s partnership with the NHS, which is not only a Sensyne Health stakeholder but a shareholder, too, owning equity in Sensyne and receiving a share of revenues to reinvest back into patient care. Sensyne Health was an obvious candidate when the NHS needed proposals to develop a mobile web service that could improve on LFT readings and make COVID-19 testing more accurate and convenient.

As one of three finalists from an initially crowded field, the company received a dataset to use as a prototype. The objective for the finalists: provide a way to read LFTs at 95 percent specificity and 95 percent sensitivity (false positives and false negatives). The stakes for such a tool are high. “If you’re doing a million tests and 5 percent are false positives, you’re going to create a huge compression on the health system,” explains Payne. Likewise, false negatives allow the virus to spread.

The difficulty stems from precisely reading the horizontal lines on the LFT stick, which resemble the thin lines on a pregnancy test. No two people see in necessarily the same way; vision acuity differs, and human brains receive and interpret images in varying ways. Sensyne Health’s plan was to train its algorithm on hundreds of thousands—even millions—of images. It just needed the right technology to analyze that vast sea of data.

Putting the chemistry between Microsoft and NVIDIA to work

A Microsoft Partner Network member, Sensyne Health has extensive experience with the Azure platform. It used Azure services to run data cost-effectively and in a highly secure manner for MagnifEye, supplying compute power with Azure HPC. “Microsoft has invested heavily in the cybersecurity aspect of the Azure platform,” says Payne. Sensyne needed multiple certifications for the product to comply with the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom, the General Data Protection Regulation (GDPR) in the European Union, and the Health Insurance Portability and Accountability Act of 1996 (HIPAA) in the United States. “The regulated products we build must follow strict medical device and software guidelines,” he explains. “With the better modeling and traceability through Azure logging mechanisms, we can identify all the source data that makes up the end product, which is key to MHRA requirements—we need to prove that auditability. For that reason alone, using Azure saved a lot of product development time.”

Sensyne Health used Azure Kubernetes Service (AKS) with NVIDIA V100 GPUs to provide the cost-effective scalability and speed it needed to deliver lightning-fast results. “We can automate spinning up a Kubernetes node literally in seconds,” explains Payne. “That’s pivotal because most requests for results occur between 7:30 and 8:30 in the morning for people who need to present their results to enter their workplaces or schools. Other spikes occur in the evening and at midday.” The company further speeds that process by using Azure Machine Learning to predictively spin up that infrastructure. Azure continuous integration and continuous delivery (CI/CD) expedites automated model updates. The millisecond response time and instant scalability of Azure Cosmos DB made it the perfect vehicle for test requests and results.

Figure 1. Managing LFT testing spikes at different times of the day using Azure Machine Learning within AKS combined with Azure continuous integration and continuous delivery for expediting automated model updates. Azure Cosmos DB response and scalability capabilities made it the perfect vehicle for LFT test requests and results.

The solution would also have to deploy at scale to multiple concurrent end users. NVIDIA supplied the other pieces of that puzzle, Triton Inference Server, which runs on NVIDIA V100 GPUs in AKS clusters to infer results for multiple test images in parallel—initially meeting the NHS standard of 500 tests per second. “Our initial benchmark of 500 tests per second is impressive,” asserts Payne. “But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure.” The machine response time to deliver the result for the user was a tenth of a second—effectively in real time for the user. “The performance we achieved in terms of the sensitivity of the algorithm is fantastic by any academic standard,” he adds. “With the Microsoft technology stack, we have incredible performance that we couldn’t have gotten otherwise.”

Figure 2. Sensyne Health achieved 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs and on AKS, providing the cost-effective scalability and speed needed to deliver lightning-fast results.

As an added bonus, Triton Inference Server was four times as fast as another comparably priced solution. “NVIDIA provided a price/performance ratio sea change,” says Payne. “It was remarkable throughput.” Triton Inference Server continues to perform impressively amidst rising data volumes. “We have three servers that service hundreds of thousands of tests a day, and we can easily scale to 30 to 40 million a week,” he adds. “This technology gives governments an enormous opportunity to get a handle on how the virus is performing and adapting and in which locations.”

“Seeing” with AI

Much of the project focused on anomaly detection to help ensure that the software accurately perceived the control and test lines of the LFT stick window and correctly interpreted the results. The Sensyne Health team trained the anomaly framework for several weeks, eventually creating nine anomaly checks that run in parallel. Those anomaly checks tell the user that the image isn’t in focus or is too dark or too light.

The complicating factor? That same training needs to be repeated for every test that a different manufacturer creates. “We’ve now trained the models on tests from more than six manufacturers, but we still retain a 99.6 percent false positive rate,” says Payne. “That’s astounding considering the variety of pictures we get from real-world use and the volume of images.”

Sensyne Health runs 10 models for each processed test—one to complete the LFT reading and the other nine for anomaly detection. That intensive processing brought up the question of whether to use central processing units (CPUs) or graphics processing units (GPUs) for inference support. Sensyne Health decided to use NVIDIA NC6 v3 CPUs with the NVIDIA V100 GPU. “The hybrid approach turned out to be far and away the most cost-effective, and it’s very straightforward to operate,” reports Payne. “We effectively run multiple models simultaneously to deliver an accurate result and identify the test manufacturer, giving us ‘one algorithm to rule them all.’“

The company creates rich data graphics with Power BI that NHS officials and elected officials can use for fast policy decisions. MagnifEye soon made a difference in the spread of COVID-19 in the United Kingdom thanks to the fast, accurate readings it provides and the invaluable information it makes available to policy makers.

Saving lives, enhancing healthcare resources for everyone

Andrew Beggs, Consultant Surgeon and Professor of Surgery and Cancer Genetics at the University of Birmingham and University Hospitals Birmingham and advisor to the UK Department of Health and Social Care, has been leading COVID-19 testing for NHS staff in his region. He emphasizes that end user interpretation was a major issue—until MagnifEye. “MagnifEye breaks the chain of infection by reducing the spread of COVID-19,” he says. “We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS.”

He’s optimistic about the impact that the technologies behind MagnifEye can bring to healthcare. “Machine learning and AI opens up a world of potential for clinicians in the NHS, including rapid screening of chest X-rays for COVID-19, increased accuracy of staging of cancer scans, and earlier detection of patient deterioration in a ward-based environment,” he explains. “We also hope to reduce delays in cancer diagnosis by deploying the technology to a triage-based environment where patients with cancer symptoms can be automatically and rapidly referred into the right investigation.”

Payne is adamant that without exceptional teamwork, the high-performance app couldn’t have been developed in such a tight time frame. “The cultural fit of the teams from Microsoft, NVIDIA, and Sensyne Health made MagnifEye possible,” says Payne. “We were in the middle of a pandemic, but everyone pulled together to rise to a great purpose. We’re grateful to Microsoft teams both in the United Kingdom and the United States, working with NVIDIA to bring their technologies together to help us create the solution we’re so proud of. We couldn’t have done it without them.”

"We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS."

Andrew Beggs: Consultant Surgeon and Professor of Surgery and Cancer Genetics
University of Birmingham and University Hospitals Birmingham, and advisor to the UK Department of Health and Social Care

View the case study on the Microsoft customer stories page: https://customers.microsoft.com/en-us/story/1430377058968477645-sensyne-health-partner-professional-services-azure-hpc

Case study

Sensyne Health aids National Health Service in the COVID-19 struggle with Microsoft HPC and AI technologies

Sensyne Health aids National Health Service in the COVID-19 struggle with Microsoft HPC and AI technologies

When the United Kingdom’s National Health Service envisioned an app that could read COVID-19 tests at scale with an accuracy and speed not possible by human readers, Sensyne Health answered the call.

The app connects users to a web service that uses AI to deliver an Azure high-performance computing (HPC) solution that’s adeptly orchestrated with NVIDIA Triton Inference Server on graphics processing units. The result: the incredibly fast and accurate MagnifEye solution.

"Our initial benchmark of 500 tests per second is impressive. But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure."

Alan Payne, Chief Information and Product Officer, Sensyne Health

Suppose you could know with a very high degree of accuracy and in just seconds if you have COVID-19. How might that information affect the spread of the virus? The National Health Service (NHS), the United Kingdom’s publicly funded healthcare system, asked that question as it dug into the extraordinary struggle to control the COVID-19 outbreak. It wanted to overcome the issues of response time and the fallibility of reading Lateral Flow Tests (LFTs) with the human eye.

Sensyne Health came up with an answer: its MagnifEye solution, an app for mobile devices that use a device’s camera to capture the LFT stick image and analyze it in tenths of seconds with extraordinary accuracy. It was a matter of combining the right technologies: high-performance computing (HPC) on Microsoft Azure and Azure Cognitive Services, along with NVIDIA graphics processing unit (GPU) acceleration and NVIDIA Triton Inference Server.

Exercising a novel approach to the thorniest health problems

Until recently, analyzing enough data to solve some of the biggest issues in healthcare was a dream—timely analysis was an insurmountable barrier. But current AI and machine learning technology is changing that.

Sensyne Health was built on a unique premise: the ethical use of data with AI technology to combat threats to human health. “We’re an ethically driven safe haven and docking station between healthcare data and pharmaceutical research,” says Alan Payne, Chief Information and Product Officer at Sensyne Health. “We work only with fully anonymized data that has a clear and definitive purpose.” The company uses anonymized data from the United Kingdom’s NHS and United States’ healthcare systems to build rich datasets—more than 22.5 million records as of this writing. That data holds incredible potential to combat disease.

Also unique is Sensyne Health’s partnership with the NHS, which is not only a Sensyne Health stakeholder but a shareholder, too, owning equity in Sensyne and receiving a share of revenues to reinvest back into patient care. Sensyne Health was an obvious candidate when the NHS needed proposals to develop a mobile web service that could improve on LFT readings and make COVID-19 testing more accurate and convenient.

As one of three finalists from an initially crowded field, the company received a dataset to use as a prototype. The objective for the finalists: provide a way to read LFTs at 95 percent specificity and 95 percent sensitivity (false positives and false negatives). The stakes for such a tool are high. “If you’re doing a million tests and 5 percent are false positives, you’re going to create a huge compression on the health system,” explains Payne. Likewise, false negatives allow the virus to spread.

The difficulty stems from precisely reading the horizontal lines on the LFT stick, which resemble the thin lines on a pregnancy test. No two people see in necessarily the same way; vision acuity differs, and human brains receive and interpret images in varying ways. Sensyne Health’s plan was to train its algorithm on hundreds of thousands—even millions—of images. It just needed the right technology to analyze that vast sea of data.

Putting the chemistry between Microsoft and NVIDIA to work

A Microsoft Partner Network member, Sensyne Health has extensive experience with the Azure platform. It used Azure services to run data cost-effectively and in a highly secure manner for MagnifEye, supplying compute power with Azure HPC. “Microsoft has invested heavily in the cybersecurity aspect of the Azure platform,” says Payne. Sensyne needed multiple certifications for the product to comply with the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom, the General Data Protection Regulation (GDPR) in the European Union, and the Health Insurance Portability and Accountability Act of 1996 (HIPAA) in the United States. “The regulated products we build must follow strict medical device and software guidelines,” he explains. “With the better modeling and traceability through Azure logging mechanisms, we can identify all the source data that makes up the end product, which is key to MHRA requirements—we need to prove that auditability. For that reason alone, using Azure saved a lot of product development time.”

Sensyne Health used Azure Kubernetes Service (AKS) with NVIDIA V100 GPUs to provide the cost-effective scalability and speed it needed to deliver lightning-fast results. “We can automate spinning up a Kubernetes node literally in seconds,” explains Payne. “That’s pivotal because most requests for results occur between 7:30 and 8:30 in the morning for people who need to present their results to enter their workplaces or schools. Other spikes occur in the evening and at midday.” The company further speeds that process by using Azure Machine Learning to predictively spin up that infrastructure. Azure continuous integration and continuous delivery (CI/CD) expedites automated model updates. The millisecond response time and instant scalability of Azure Cosmos DB made it the perfect vehicle for test requests and results.

Figure 1. Managing LFT testing spikes at different times of the day using Azure Machine Learning within AKS combined with Azure continuous integration and continuous delivery for expediting automated model updates. Azure Cosmos DB response and scalability capabilities made it the perfect vehicle for LFT test requests and results.

The solution would also have to deploy at scale to multiple concurrent end users. NVIDIA supplied the other pieces of that puzzle, Triton Inference Server, which runs on NVIDIA V100 GPUs in AKS clusters to infer results for multiple test images in parallel—initially meeting the NHS standard of 500 tests per second. “Our initial benchmark of 500 tests per second is impressive,” asserts Payne. “But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure.” The machine response time to deliver the result for the user was a tenth of a second—effectively in real time for the user. “The performance we achieved in terms of the sensitivity of the algorithm is fantastic by any academic standard,” he adds. “With the Microsoft technology stack, we have incredible performance that we couldn’t have gotten otherwise.”

Figure 2. Sensyne Health achieved 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs and on AKS, providing the cost-effective scalability and speed needed to deliver lightning-fast results.

As an added bonus, Triton Inference Server was four times as fast as another comparably priced solution. “NVIDIA provided a price/performance ratio sea change,” says Payne. “It was remarkable throughput.” Triton Inference Server continues to perform impressively amidst rising data volumes. “We have three servers that service hundreds of thousands of tests a day, and we can easily scale to 30 to 40 million a week,” he adds. “This technology gives governments an enormous opportunity to get a handle on how the virus is performing and adapting and in which locations.”

“Seeing” with AI

Much of the project focused on anomaly detection to help ensure that the software accurately perceived the control and test lines of the LFT stick window and correctly interpreted the results. The Sensyne Health team trained the anomaly framework for several weeks, eventually creating nine anomaly checks that run in parallel. Those anomaly checks tell the user that the image isn’t in focus or is too dark or too light.

The complicating factor? That same training needs to be repeated for every test that a different manufacturer creates. “We’ve now trained the models on tests from more than six manufacturers, but we still retain a 99.6 percent false positive rate,” says Payne. “That’s astounding considering the variety of pictures we get from real-world use and the volume of images.”

Sensyne Health runs 10 models for each processed test—one to complete the LFT reading and the other nine for anomaly detection. That intensive processing brought up the question of whether to use central processing units (CPUs) or graphics processing units (GPUs) for inference support. Sensyne Health decided to use NVIDIA NC6 v3 CPUs with the NVIDIA V100 GPU. “The hybrid approach turned out to be far and away the most cost-effective, and it’s very straightforward to operate,” reports Payne. “We effectively run multiple models simultaneously to deliver an accurate result and identify the test manufacturer, giving us ‘one algorithm to rule them all.’“

The company creates rich data graphics with Power BI that NHS officials and elected officials can use for fast policy decisions. MagnifEye soon made a difference in the spread of COVID-19 in the United Kingdom thanks to the fast, accurate readings it provides and the invaluable information it makes available to policy makers.

Saving lives, enhancing healthcare resources for everyone

Andrew Beggs, Consultant Surgeon and Professor of Surgery and Cancer Genetics at the University of Birmingham and University Hospitals Birmingham and advisor to the UK Department of Health and Social Care, has been leading COVID-19 testing for NHS staff in his region. He emphasizes that end user interpretation was a major issue—until MagnifEye. “MagnifEye breaks the chain of infection by reducing the spread of COVID-19,” he says. “We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS.”

He’s optimistic about the impact that the technologies behind MagnifEye can bring to healthcare. “Machine learning and AI opens up a world of potential for clinicians in the NHS, including rapid screening of chest X-rays for COVID-19, increased accuracy of staging of cancer scans, and earlier detection of patient deterioration in a ward-based environment,” he explains. “We also hope to reduce delays in cancer diagnosis by deploying the technology to a triage-based environment where patients with cancer symptoms can be automatically and rapidly referred into the right investigation.”

Payne is adamant that without exceptional teamwork, the high-performance app couldn’t have been developed in such a tight time frame. “The cultural fit of the teams from Microsoft, NVIDIA, and Sensyne Health made MagnifEye possible,” says Payne. “We were in the middle of a pandemic, but everyone pulled together to rise to a great purpose. We’re grateful to Microsoft teams both in the United Kingdom and the United States, working with NVIDIA to bring their technologies together to help us create the solution we’re so proud of. We couldn’t have done it without them.”

"We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS."

Andrew Beggs: Consultant Surgeon and Professor of Surgery and Cancer Genetics
University of Birmingham and University Hospitals Birmingham, and advisor to the UK Department of Health and Social Care

View the case study on the Microsoft customer stories page: https://customers.microsoft.com/en-us/story/1430377058968477645-sensyne-health-partner-professional-services-azure-hpc

Arrange to meet us
Case study

Sensyne Health aids National Health Service in the COVID-19 struggle with Microsoft HPC and AI technologies

November 2, 2021

When the United Kingdom’s National Health Service envisioned an app that could read COVID-19 tests at scale with an accuracy and speed not possible by human readers, Sensyne Health answered the call.

The app connects users to a web service that uses AI to deliver an Azure high-performance computing (HPC) solution that’s adeptly orchestrated with NVIDIA Triton Inference Server on graphics processing units. The result: the incredibly fast and accurate MagnifEye solution.

"Our initial benchmark of 500 tests per second is impressive. But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure."

Alan Payne, Chief Information and Product Officer, Sensyne Health

Suppose you could know with a very high degree of accuracy and in just seconds if you have COVID-19. How might that information affect the spread of the virus? The National Health Service (NHS), the United Kingdom’s publicly funded healthcare system, asked that question as it dug into the extraordinary struggle to control the COVID-19 outbreak. It wanted to overcome the issues of response time and the fallibility of reading Lateral Flow Tests (LFTs) with the human eye.

Sensyne Health came up with an answer: its MagnifEye solution, an app for mobile devices that use a device’s camera to capture the LFT stick image and analyze it in tenths of seconds with extraordinary accuracy. It was a matter of combining the right technologies: high-performance computing (HPC) on Microsoft Azure and Azure Cognitive Services, along with NVIDIA graphics processing unit (GPU) acceleration and NVIDIA Triton Inference Server.

Exercising a novel approach to the thorniest health problems

Until recently, analyzing enough data to solve some of the biggest issues in healthcare was a dream—timely analysis was an insurmountable barrier. But current AI and machine learning technology is changing that.

Sensyne Health was built on a unique premise: the ethical use of data with AI technology to combat threats to human health. “We’re an ethically driven safe haven and docking station between healthcare data and pharmaceutical research,” says Alan Payne, Chief Information and Product Officer at Sensyne Health. “We work only with fully anonymized data that has a clear and definitive purpose.” The company uses anonymized data from the United Kingdom’s NHS and United States’ healthcare systems to build rich datasets—more than 22.5 million records as of this writing. That data holds incredible potential to combat disease.

Also unique is Sensyne Health’s partnership with the NHS, which is not only a Sensyne Health stakeholder but a shareholder, too, owning equity in Sensyne and receiving a share of revenues to reinvest back into patient care. Sensyne Health was an obvious candidate when the NHS needed proposals to develop a mobile web service that could improve on LFT readings and make COVID-19 testing more accurate and convenient.

As one of three finalists from an initially crowded field, the company received a dataset to use as a prototype. The objective for the finalists: provide a way to read LFTs at 95 percent specificity and 95 percent sensitivity (false positives and false negatives). The stakes for such a tool are high. “If you’re doing a million tests and 5 percent are false positives, you’re going to create a huge compression on the health system,” explains Payne. Likewise, false negatives allow the virus to spread.

The difficulty stems from precisely reading the horizontal lines on the LFT stick, which resemble the thin lines on a pregnancy test. No two people see in necessarily the same way; vision acuity differs, and human brains receive and interpret images in varying ways. Sensyne Health’s plan was to train its algorithm on hundreds of thousands—even millions—of images. It just needed the right technology to analyze that vast sea of data.

Putting the chemistry between Microsoft and NVIDIA to work

A Microsoft Partner Network member, Sensyne Health has extensive experience with the Azure platform. It used Azure services to run data cost-effectively and in a highly secure manner for MagnifEye, supplying compute power with Azure HPC. “Microsoft has invested heavily in the cybersecurity aspect of the Azure platform,” says Payne. Sensyne needed multiple certifications for the product to comply with the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom, the General Data Protection Regulation (GDPR) in the European Union, and the Health Insurance Portability and Accountability Act of 1996 (HIPAA) in the United States. “The regulated products we build must follow strict medical device and software guidelines,” he explains. “With the better modeling and traceability through Azure logging mechanisms, we can identify all the source data that makes up the end product, which is key to MHRA requirements—we need to prove that auditability. For that reason alone, using Azure saved a lot of product development time.”

Sensyne Health used Azure Kubernetes Service (AKS) with NVIDIA V100 GPUs to provide the cost-effective scalability and speed it needed to deliver lightning-fast results. “We can automate spinning up a Kubernetes node literally in seconds,” explains Payne. “That’s pivotal because most requests for results occur between 7:30 and 8:30 in the morning for people who need to present their results to enter their workplaces or schools. Other spikes occur in the evening and at midday.” The company further speeds that process by using Azure Machine Learning to predictively spin up that infrastructure. Azure continuous integration and continuous delivery (CI/CD) expedites automated model updates. The millisecond response time and instant scalability of Azure Cosmos DB made it the perfect vehicle for test requests and results.

Figure 1. Managing LFT testing spikes at different times of the day using Azure Machine Learning within AKS combined with Azure continuous integration and continuous delivery for expediting automated model updates. Azure Cosmos DB response and scalability capabilities made it the perfect vehicle for LFT test requests and results.

The solution would also have to deploy at scale to multiple concurrent end users. NVIDIA supplied the other pieces of that puzzle, Triton Inference Server, which runs on NVIDIA V100 GPUs in AKS clusters to infer results for multiple test images in parallel—initially meeting the NHS standard of 500 tests per second. “Our initial benchmark of 500 tests per second is impressive,” asserts Payne. “But after tuning our algorithm and working with Microsoft to use the latest technology from Azure Cognitive Services, we achieved more than 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs in Azure.” The machine response time to deliver the result for the user was a tenth of a second—effectively in real time for the user. “The performance we achieved in terms of the sensitivity of the algorithm is fantastic by any academic standard,” he adds. “With the Microsoft technology stack, we have incredible performance that we couldn’t have gotten otherwise.”

Figure 2. Sensyne Health achieved 1,000 tests per second running the predictions model on a six-node machine using Triton Inference Server on NVIDIA GPUs and on AKS, providing the cost-effective scalability and speed needed to deliver lightning-fast results.

As an added bonus, Triton Inference Server was four times as fast as another comparably priced solution. “NVIDIA provided a price/performance ratio sea change,” says Payne. “It was remarkable throughput.” Triton Inference Server continues to perform impressively amidst rising data volumes. “We have three servers that service hundreds of thousands of tests a day, and we can easily scale to 30 to 40 million a week,” he adds. “This technology gives governments an enormous opportunity to get a handle on how the virus is performing and adapting and in which locations.”

“Seeing” with AI

Much of the project focused on anomaly detection to help ensure that the software accurately perceived the control and test lines of the LFT stick window and correctly interpreted the results. The Sensyne Health team trained the anomaly framework for several weeks, eventually creating nine anomaly checks that run in parallel. Those anomaly checks tell the user that the image isn’t in focus or is too dark or too light.

The complicating factor? That same training needs to be repeated for every test that a different manufacturer creates. “We’ve now trained the models on tests from more than six manufacturers, but we still retain a 99.6 percent false positive rate,” says Payne. “That’s astounding considering the variety of pictures we get from real-world use and the volume of images.”

Sensyne Health runs 10 models for each processed test—one to complete the LFT reading and the other nine for anomaly detection. That intensive processing brought up the question of whether to use central processing units (CPUs) or graphics processing units (GPUs) for inference support. Sensyne Health decided to use NVIDIA NC6 v3 CPUs with the NVIDIA V100 GPU. “The hybrid approach turned out to be far and away the most cost-effective, and it’s very straightforward to operate,” reports Payne. “We effectively run multiple models simultaneously to deliver an accurate result and identify the test manufacturer, giving us ‘one algorithm to rule them all.’“

The company creates rich data graphics with Power BI that NHS officials and elected officials can use for fast policy decisions. MagnifEye soon made a difference in the spread of COVID-19 in the United Kingdom thanks to the fast, accurate readings it provides and the invaluable information it makes available to policy makers.

Saving lives, enhancing healthcare resources for everyone

Andrew Beggs, Consultant Surgeon and Professor of Surgery and Cancer Genetics at the University of Birmingham and University Hospitals Birmingham and advisor to the UK Department of Health and Social Care, has been leading COVID-19 testing for NHS staff in his region. He emphasizes that end user interpretation was a major issue—until MagnifEye. “MagnifEye breaks the chain of infection by reducing the spread of COVID-19,” he says. “We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS.”

He’s optimistic about the impact that the technologies behind MagnifEye can bring to healthcare. “Machine learning and AI opens up a world of potential for clinicians in the NHS, including rapid screening of chest X-rays for COVID-19, increased accuracy of staging of cancer scans, and earlier detection of patient deterioration in a ward-based environment,” he explains. “We also hope to reduce delays in cancer diagnosis by deploying the technology to a triage-based environment where patients with cancer symptoms can be automatically and rapidly referred into the right investigation.”

Payne is adamant that without exceptional teamwork, the high-performance app couldn’t have been developed in such a tight time frame. “The cultural fit of the teams from Microsoft, NVIDIA, and Sensyne Health made MagnifEye possible,” says Payne. “We were in the middle of a pandemic, but everyone pulled together to rise to a great purpose. We’re grateful to Microsoft teams both in the United Kingdom and the United States, working with NVIDIA to bring their technologies together to help us create the solution we’re so proud of. We couldn’t have done it without them.”

"We can deliver a result that’s more than 99 percent accurate in less than 2 seconds and make it accessible to anyone with a camera-enabled mobile device and internet access. Earlier identification of COVID-19 infection can help us substantially reduce costs for the NHS."

Andrew Beggs: Consultant Surgeon and Professor of Surgery and Cancer Genetics
University of Birmingham and University Hospitals Birmingham, and advisor to the UK Department of Health and Social Care

View the case study on the Microsoft customer stories page: https://customers.microsoft.com/en-us/story/1430377058968477645-sensyne-health-partner-professional-services-azure-hpc