I've seen many blogs and articles saying the artificial intelligence (AI) will save the humankind from the 2019-nCoV (a.k.a. COVID-19) pandemic. I'm sorry for breaking it, but AI will not save us from the Coronavirus. Physical distancing and handwashing will. However, what it can help with is flattening the curve. We badly need to slow down the rate of the virus spread in each and every community to give local hospitals time to deal with both the infected patients and the capacity to handle the ever-growing loads of patients. And that's where AI can come in handy. On a global scale, effective AI and machine learning solutions for the Coronavirus timely detection and control give time to hundreds of R&D teams all over the globe working to create a vaccine against the virus.
To assist in the effort, both public organizations and private companies are contributing open datasets that can be used by others to analyze trends, launch own projects and build digital solutions for a plethora of the pandemic related things, from detecting fake information on social media to identifying individuals with a high risk of being exposed to COVID-19.
I've analyzed some of the most interesting and promising solutions in terms of their potential to slow down the spread of the infection, reduce death rates, and improve information hygiene in these turbulent times.
An Israeli startup RADLogics is working on the development of automated AI-based CT images analysis tools for the Coronavirus detection, quantification and monitoring. The ultimate project goal is to differentiate infected patients from non-infected patients.
RADLogics uses multiple datasets from all over the world, including data from infected areas in China, to build 2D and 3D deep learning models and combine them with the clinical context. The company conducted several retrospective experiments to analyze system performance in the detection of suspected COVID-19 thoracic CT features and to evaluate disease evolution in each patient over time using a 3D volume review, generating a score.
At the time of this writing, RADLogics has analyzed 157 Chinese and U.S. patients and achieved 98.2% sensitivity (likelihood of correct identification of individuals with Coronavirus) and 92.2% specificity (likelihood of correct identification of individuals without Coronavirus).
The platform built by RADLogics takes only 30 seconds to process 400 CT images, while it normally takes up to 2 minutes per patient to do a CT scan of the chest. As of now, the system output enables quantitative measurement for smaller opacities (diameter, volume) and visualization of the larger opacities in a slice-based heat map.
This initial study demonstrates that rapidly developed deep learning-based algorithms for CT images analysis can have very high precision in the coronavirus diagnostics as well as its dynamics monitoring.
A Ukrainian startup ViTech Lab Healthcare is working on an AI-based solution to provide relief to medical staff throughout the world that works under high pressure due to the Coronavirus outbreak. Many healthcare providers are struggling to diagnose patients fast and correctly, while proper and timely diagnostic of the virus is crucial for the decline of the pandemic and to avoid collapse of many healthcare systems. The solution aims to lower the pressure on medical personnel by reducing diagnostic time and efforts and fostering more informed and data-based decision making.
ViTech's solution is trained to analyze CT files of many formats including DICOM, png, jpeg and others, and is able to identify and classify such illnesses as ARDS, Bacterial Pneumonia, Fungal Pneumonia, Viral Pneumonia, Legionella, Klebsiella and COVID-19 among others.
This is just one of many use cases:
COVID-19 causes viral pneumonia, while other germs like staphylococcic typically cause bacterial pneumonia. By at least determining a type of pneumonia in patients, it's possible to exclude a number of illnesses.
Image courtesy of ViTech Lab Healthcare
At the time of this writing, ViTech Lab is inviting all healthcare organizations and hospitals that have patient X-rays and CT images at their disposal and/or want to contribiute to research to join the project and request early access to see the system in action.
'Software-based diagnostic tools can serve as a valuable, virtual second opinion for medical professionals, especially in parts of the world where medical teams are short-staffed,' Barath Narayanan, a scientist at the University of Dayton Research Institute.
Barath has developed a specific software code that can detect the disease just by scanning chest X-rays. His system was adapted from existing medical diagnostic software in just a few hours and then licensed in less than three days.
This deep learning-based algorithm continues to train itself with more and more new X-rays being loaded into the system; its precision rate is now exceeding 99%.
With additional research and investments, the above technologies can be fine-tuned and polished to identify even the slightest anomalies on CT images and X-rays, which are impossible to see with the human eye. They will help speed up the time to diagnose and treat patients.
As the impact of the Coronavirus is being felt across nearly all industries and niches, it's crucial to take care of our information hygiene. With social distancing policies in place, virtual communication has become an important source of both information and misinformation. Identifying the latter is crucial, as the spread of fake news and false data about the disease can have a drastic effect on the mental health of thousands of people having to stay under strict home lockdown for weeks.
The University of Southern California (USC) Melady Lab is addressing this issue in collaboration with New Voices of Academies of Science, Engineering and Medicine. They've designed an interactive dashboard to track misinformation about the Coronavirus on Twitter. The dashboard adds visibility into Twitter posts and discussions around the pandemic by verifying the quality of information being shared. Researchers analyze more than 5 million real-time tweets related to COVID-19 and provide up-to-date insights from the social media lens.
Image source: Githab
They're collecting streaming data through the Twitter API and provides an analysis of topic clusters and social sentiments related to such hashtags as #socialdistancing and #workfromhome. By tracking emerging hashtags over time and doing time-sensitive analysis of sentiments, the project provides a method for detection of fake, misleading or clickbait content from Twitter information cascades.
The same Ukrainian startup ViTech Lab Healthcare is working on another interesting project in a bid to combat COVID-19. This one addresses the following challenge: healthcare providers are seeking faster and more efficient control of PPE compliance during the pandemic to prevent further spread of the virus.
Hospitals all over the world are overwhelmed with the coronavirus patients, which makes PPE compliance even more topical. It's crucial to ensure safety of the medical staff, as every healthy and able-bodied specialist counts!
This PPE detection solution is based on a machine learning algorithm and uses data from surveillance cameras to help detect the absence of certain PPE items like gloves or masks on a medical empoloyee. When a violation is detected, the safety engineer or controller gets an automatic report generated by the algorithm.
Besides, ViTech Lab's solution can help identify people with high temperature and can also log the body temperature in real time. Each time the temperature is higher than normal, medical staff get an alert to take an action.
Last but not least, the solution is capable of face recognition and can be used to identify safety rules violators among the personnel (e.g., people who don't have a respirator or a face shild on).
You can request a demo of this AI-based PPE compliance tool here.
Ancora.ai is using technology and a patient-first approach to democratize clinical trial access and foster informed decision-making. One of the company goals is to provide unbiased clinical trial information by collecting data from public registries so that patients, their healthcare providers, and their loved ones can evaluate all potential treatment options.
Clinical trials are crucial as they provide the opportunity to try innovative ways to treat diseases months and years before they reach the market. Alternatively, they give patients the opportunity to try a new type of treatment or medication without waiting for it to be approved or reimbursed by insurances.
To contribute to the urgent fight against the Coronavirus, Ancora now supports and facilitates the search for the COVID-19 clinical trials to help develop treatments and diagnostics.
"We use a custom parser and NLP model to restructure eligibility criteria and then use this to develop questionnaires featuring the most relevant eligibility criteria in a standard way (e.g, those that exclude the highest % of trials)," says Emily Rose Jordan, PhD, COO at Ancora.
Then they can match patients to suitable trials where they fit the criteria based on their responses to a simple questionnaire. This brings down search results from 100+ per patient to just a dozen that are highly relevant to them.
Here's how it works.
If you see a trial that might be a good fit for you, you can then ask to be connected with the sponsors to take the next steps towards enrollment. Using Ancora is free and you do not have to create an account or save your details in order to search for trials. However, if you create a profile, the company can provide more specific trial matching with an advanced set of questions, connect you to trial sites, and share alerts about new opportunities that you might be interested in.
I encourage you to take a look and participate in the COVID-19 clinical trials. By participating in clinical trials, we all can help advance medical research while gaining access to the latest treatment options available.
Are you using technology to contribute to the global fight against the Coronavirus? Get in touch, and I'll be happy to add your case to this article.