The second wave of COVID-19 was unrelenting on India's healthcare workers, hospitals, governments and citizens. Valuable lessons have been learnt during this period, with the hope that the country doesn't see another devastating wave of COVID, but if it does, first responders are more prepared. COVID-19 spurred a series of technology-driven innovations in healthcare. Last year witnessed the unbridled rise of telemedicine, and as lockdowns continue to restrict mobility, telemedicine is witnessing rising adoption levels even in rural areas now. This year, a highly potent 'double mutant' variation did insurmountable damage to the health of Indians, in rural and urban areas alike. While urban Indians still have access to quality healthcare, the second wave disproportionately affected India's villages and small towns. 

A novel solution called XraySetu is now being considered a key tool in the arsenal of solutions needed to beat COVID19, and aid doctors and healthcare workers in India's towns and villages.

India has only 10,000 radiologists. In rural areas, this surmounts to 1 radiologist for every 100,000 people. Meanwhile, WhatsApp worldwide has over 2 billion users. Of them, 390 million are in India alone - making it the single biggest user nation for WhatsApp. Rural health infrastructure is far from developed but with COVID-19 wreaking havoc across the Indian hinterland, it was imperative to develop a screening or testing solution that can deployed through an existing network with a simple interface. AI researchers at ARTPARK in collaboration with Niramai Health Analytix and IISc developed XraySetu, an AI-driven solution for the early intervention of COVID-19 cases across India with chest Xray interpretation over WhatsApp. 

Umakant Soni, Cofounder and CEO of Artpark explains, "In some of India's biggest cities, COVID19 tests are taking more than a week to get administered. So you can only imagine the plight of rural geographies – home to about two-thirds of the Indian population. Moreover, false negatives produced by RT-PCR tests exacerbates the crisis. This wave has left several people with extensive lung damage despite being declared COVID19-negative on an RT-PCR. On the other hand, there are 80,000 Xray machines that smaller towns and cities can use. We thought if we build a solution that leverages existing infrastructure in rural India via a simple user interface like WhatsApp, it can save more lives."

How AI and ML Work To Make Xray Setu Highly Effective, Accurate and Accessible 

XraySetu is a cloud-hosted AI-based service that analyses chest Xrays using machine learning algorithms and generates a patient report showing suspicious abnormal regions in the lungs and detects if the person is likely positive for COVID, pneumonia or other lung abnormalities. The solution has so far tested and validated 125,000 X Ray images from standard Open Source NIH COVID19 dataset with sensitivity rate of 98.86%, specificity of 74.74% and AUC of 0.9316. (AUC suggests no discrimination and an AUC score above 0.9 is considered outstanding.) The model is tuned to process low-resolution images of X-rays as well, suitable for Whatsapp-based delivery. In addition, an explainable AI provision allows for semantic annotations to be reviewed by doctors, and a localised heatmap of affected areas allows for easy review. The automatically generated report gives annotated images marking 15 different lung abnormalities along with probability of COVI-19 and pneumonia, including atelectasis, cardiomegaly, effusion, infiltration, mass, nodule, pneumonia, pneumothorax, consolidation, edema, emphysema, fibrosis, pleural thickening, hernia and COVID19. 

In response to the rising load on diagnostic centres to administer RT-PCR tests, XraySetu is a viable alternative for doctors to screen patients for Covid-19, plan timely intervention and avoid over medication, added Soni. 

Doctors & Diagnosticians Wish They Had This Solution Earlier So More Lives Could Be Saved 

So far, 300+ doctors have been connected with XraySetu in the last 10 months, mostly from parts of rural India. Close to 1,000 Xrays have been analysed so far on WhatsApp. "We are getting queries mostly from Medical officers at PHCs. Since its launch it has already served close to 1,200+ reports so far from the interior parts of India," says Soni. 

Dr Padmanabh Kamath, Prof & HOD Cardiology, KMC, Mangalore, who has been an early advisor and user of XraySetu, remarked, “I have been a biggest proponent of taking healthcare and technology to the underprivileged and rural areas. I hope XraySetu gets nationwide acclaim and recognition in bridging the gaps for the end user and for the healthcare sector." 

In Pune's Baramati Taluka, Om Sai Diagnostic's technician Shubham Kalyan wishes this service was made available a month earlier so more lives could be saved. "We read about XraySetu in the newspaper and started using the service for COVID-19 patients in Baramati taluka. We started sharing pictures of Xrays on the XraySetu WhatsApp number and started getting the report in few minutes. It's very fast and easy. We validate each XraySetu report with in-house doctors, visiting radiologist and RT PCR test. We have been very satisfied with the outcome of the reports. So far, we have shared more than 20 reports in the last two weeks and all the reports are correct and validated."

To date, Xraysetu has received 7,000+ scans from over 2,000 unique users after it was launched as a chatbot. Overall, Xraysetu has serviced 3500+ doctors and Xray technicians from all parts of India. More recently, the solution is getting traction from overseas as well.

Soni says, "The second wave of COVID-19 truly exposed the frailty of our current healthcare ecosystem. Scaling manpower is hard and often futile in a big and populous country like India. We need to build the future of our healthcare system with technology to scale to 1.36bn people. XraySetu is one tiny step in this direction."

As far as product enhancements go, the team at XraySetu is working to add vernacular language support for the chatbot, to enhance its usability for healthcare workers in rural areas.