How a Singaporean startup is using AI to help speed up Covid-19 diagnosis

Interview with Dr Jeevesh Kapur and David Quail, co-founders of Medo.

15 years ago, patients had to go to trained medical staff for a blood pressure reading. Fast forward to today - with leaps in healthcare tech, just about anyone can purchase a blood pressure monitor rather affordably.

Tech has helped to make complex and specialised medical tools more accessible. One Singaporean startup, Medo, has built an AI tool to simplify the analysis of lung scans. This could help to increase accessibility to more reliable scans and reduce strain on hospitals around the world.

GovInsider spoke to Dr Jeevesh Kapur and David Quail, co-founders of Medo, to find out more.

How it works

 

Medo has come up with an AI-powered tool called Medo Lung which can quickly analyse lung scans. The algorithm powering Medo Lung is trained to inspect lung ultrasounds for the presence of lung abnormalities, which indicate the severity of possible Covid-19 infection or pneumonia.

Medo Lung can complement current handheld ultrasound scanners, transforming how healthcare is brought to the patient. “Just put it in your pocket”, Dr Jeevesh says. The clinician will be able to scan the patient wherever they are, upload it to the cloud-based system and get a diagnosis within seconds, he explains.

The tool is in the process of getting regulatory approvals. Medo is working with hospitals and clinics in Canada for trials.

More accessible healthcare

Ultrasound is a complicated tech that requires extensive training, notes Dr Jeevesh. Inexperienced clinicians could mis-diagnose or miss something altogether.

Medo has simplified the process significantly to place less burden on the caregivers who are using ultrasound. Taking a scan is as easy as taking a “panoramic view on your phone”, he notes.

This innovation brings portability and ease-of-use to help smaller clinics lacking in expertise and facilities. It could even be used in ambulances, allowing patients to access an expert-level diagnosis from the AI before entering the hospital.

“The cloud has allowed us to greatly improve patient outcomes, including in remote areas that would not otherwise have the infrastructure needed to perform or manage such patients,” Dr Jeevesh says.

“Covid-19 really crystallised our vision of democratising medical imaging and bringing our solution to as many people as we can,” Quail shares. Medo’s tool can “be administered by any caregiver at the point of care where it’s needed the most, keeping patients comfortable and safe”.

Medo Lung presents a simpler alternative to the usual way of managing Covid patients. Traditionally, doctors would conduct CT scans to diagnose lung conditions.But this is “ not only a highly radiating modality, it is very time consuming and very cumbersome”, he notes.

Patients would have to move out of their ward to where the CT machines are, making infection control tricky. Some may also be too weak to move. Plus, the machines would have to be thoroughly sanitised after every use, explains Quail.

Ease patient loads

Medo hopes to ease the overburdened healthcare system with this AI tool. Covid-19 patients may experience a spectrum of symptoms ranging from mild, moderate or severe. Patients in a less severe condition can usually be sent home to recuperate in isolation, Dr Jeevesh explains.

The AI tool analyses the specifics and severity of a lung infection. Medical staff can then prioritise care for patients in critical conditions.

He predicts that this will have an “extremely huge impact for patient management”. Overwhelmed hospitals will be able to manage their patient numbers and reduce fatigue among healthcare workers.

How AWS has helped

AWS’s cloud services has allowed Medo to be “experimental” with their innovation, Quail says. Its development team was able to make changes “quickly and radically” when testing out new algorithms without having to worry about pricey server costs.

Its AI tools came in helpful as well. The team used Amazon SageMaker to train AI models.

Amazon Textract has also allowed Medo to read and extract information from medical images more easily. The team could then focus on more novel machine learning challenges, such as disease classifications. AWS’s tools have helped Medo create a tool that will impact lives “in both the short term and long after the pandemic subsides”, Quail believes.

This project is supported by AWS’s Diagnostic Development Initiative, which was established to accelerate research for the detection of Covid-19. AWS has earmarked US$20 million for this two-year programme.

The Initiative has supported 87 organisations, ranging from startups, nonprofits, research institutions, and businesses, across 17 countries in its first year. In 2021, the Initiative has expanded beyond diagnostics to include early disease detection, prognosis to better understand how a disease progresses over time, and public health genomics. AWS continues to accept applications for the program at www.aws.amazon.com/ddi.

Innovation towards more accessible healthcare will help more people receive the care that they need. As we step into the age of information, AI and cloud tech will lead the charge towards better healthcare for all.

Featured image of an ultrasound probe using Medo’s technology by Medo.