Intelli-Care: A Frontline Tool for Fighting COVID-19

Sankar Ramachandran
DeepTek
Published in
5 min readMay 4, 2020
Images: yuri.samoilov.online, www.flickr.com/photos/gleonhard/33661764490

DeepTek, which has developed core competency in using artificial intelligence (AI) solutions for the healthcare sector, has developed an AI-powered solution for fighting the COVID-19 pandemic. Intelli-Care is a-state-of-the-art, physician-oriented solution designed for vanguard service in the fight against COVID-19. It features a holistic suite of capabilities that can critically aid healthcare professionals by offering end-to-end assistance in screening, diagnostics, and workload management. Intelli-Care was developed leveraging on the company’s expertise in the field of tuberculosis eradication programs, its experience in the field of machine learning, and the skill of its core team. So far, our AI lead radiology offerings are being leveraged by over 50 hospitals and imaging centers across India.

When utilized in the field, Intelli-Care can act as a scout, guide, and force multiplier for healthcare professionals as they tackle this pandemic.

AI-Powered Imagery Analysis

An aggressive and comprehensive testing regime has now been adopted as the primary means of fighting this pandemic. While viral testing remains at the forefront of the testing regime, it has some key drawbacks which impose limits on its standalone effectiveness in fighting this pandemic. Viral testing using RT-PCR and other modes is expensive, not widely available, and takes time in getting results. Furthermore, the RT-PCR test is very specific but has a lower sensitivity. Research coming out of China suggests that it may be accurate only 70% of the time. In other words, a viral test can fail even when the patient is infected.

A smarter testing model that addresses these critical drawbacks incorporates radiology to augment biochemical confirmatory tests. Since COVID-19 patients start exhibiting very specific lung abnormalities and pneumonia-related symptoms within a few days of getting infected, many health centres are using chest X-rays and CT scans to screen suspected patients. This is where our AI-powered solution can help.

Intelli-Care uses its proprietary algorithms to quickly scan radiological data. Our extensive research into machine learning means its solutions are precise and accurate in its scanning ability. The neural network model powering Intelli-Care correctly predicts whether a case had the disease more than 90% of the time using either X-ray or CT scan images of the person. It attains a specificity of over 90% while also maintaining sensitivity at 90%.

The Human Touch

Image: www.af.mil/News/Photos/igphoto/2000598750/

We appreciate the dynamic and fluid nature of machine learning and are dedicated to its constant augmentation. In many cases, the initial cache of data is inadequate to build a robust and reliable AI model. Human expertise and instinct can significantly improve the system, which is why we have an integrative approach that allows a human radiologist to verify the algorithm’s results. Each time a radiologist corrects the AI prediction, the rectifications are fed back into the machine learning algorithm. Over time, the AI learns from the human radiologist and performs more accurately.

While our algorithms are advanced, we have also believed that human expertise and instinct, which is developed through years of learning and practice, must also supplement AI solutions. Combining human expertise with speed and precision of AI, we believe, is the right formula for any smart AI solution. That is why we have the Expert in the Loop as a part of Intelli-Care. Our team of board-certified radiologists will constantly scrutinize and provide feedback contributing to the auxiliary diagnosis of certain conditions.

Turning the Tide

An outbreak of a viral infection in a certain area launches a tidal wave of persons suspected of having contracted the infection rushing to the closest testing lab or hospital. The rapid rate of infection, as well as its confusing pathologies, contribute to the panic. This throws a wrench into existing testing and diagnostic systems and strains human and technical resources to the maximum. As the key to successful management of the pandemic is rapid and accurate testing, isolating, and treating of COVID-19 patients, there is no room for error. Yet such a high tempo response will possibly lead to physician burnout and omissions in diagnosis and treatment.

Triaging is the key here and Intelli-Care helps do this. Its algorithms deal with this deluge of cases by quickly helping physicians identify high probability cases and prioritising them. Intelli-Care creates a smart worklist that prioritizes cases to be reported based on the likelihood of the patient needing attention.

Better Situational Awareness

Intelli-Care’s physician-focussed smart design dashboard is crafted to provide high situational awareness to physicians and other healthcare workers in an intense, changing work environment. The interface provides quick access to critical information and continuous monitoring of a large number of cases.

Smart Response

Intelli-Care supports workflow integration and follow up over a suite of communication tools like SMS, WhatsApp, and email. These enable the physicians to better manage patients through quick contact and instant alert and enhance their capability to coordinate their responses with others in the healthcare system.

Radiologist at Your Service

Many screening centres do not have radiologists on call and may find it expensive to analyse the large volume of X-rays and CT scans. Furthermore, the time factor is crucial, as patients must be identified quickly. Our AI results are available in a few seconds. Through a highly optimized radiology workflow system, the AI results are reviewed and confirmed by an expert radiologist. The confirmation by an expert Indian board-certified radiologist is delivered within one hour. Our in-house radiology services handle large volumes in a cost-effective and efficient manner.

--

--