Thank you for Subscribing to CIO Applications Weekly Brief
AI Developer Opu Labs Releases App to Check for COVID-19 Skin Rashes
As recent as February of this year, cases began to be reported of COVID-19 patients experiencing skin conditions many of which appeared before the onset of other symptoms. Also, known as the “COVID Rash”, as many as twenty percent of the patients testing positive for COVID have now been found to exhibit this rash as one of the symptoms. Furthermore, this same rash is appearing in roughly half of these patients prior to the emergence of any other symptoms.
In order to aid in the battle to identify COVID-19 as early as possible, Opu Labs has built a product from the ground up to aid in the recognition of the symptoms of COVID-19 cutaneously.
Pandemics caused by viral infection are occurring in more regular cycles, and the successful management of outbreaks requires efficient testing and tracing. However, adequate resources for widespread testing can be limited and, for emerging infectious diseases, non-existent. Approximately 20% of individuals who have tested positive for COVID-19 have presented with viral exanthem (i.e. widespread skin rashes), and half of these occurred prior to the onset of more common symptoms (e.g., cough, fever, shortness of breath). To reduce the burden on resources for testing, an automated smartphone-based classification system could provide pre-screening for cutaneous manifestations consistent with viral infection (e.g., maculopapular rash). These test results are the key to enabling at-home surveillance, and accelerating actions to self-quarantine or seek further medical advice. Recent advances in deep learning make possible the necessary performance of the artificial intelligence (AI) for efficient identification of viral exanthems, independent from other “non-viral” skin conditions (e.g., acne) and from normal skin.
An automated skin analyzer that is informative and simple to use enables at-home surveillance of potentially infected individuals, and contributes to the arsenal of tools essential for expediting a clinical diagnosis of specific viral infection, thus slowing the spread of potentially lethal viral infections.
Recent advances in deep learning technology over the past decade have improved the precision of image recognition to specific sectors, known as “vertical” AI, and have enabled increasingly effective detection of specific skin conditions. Opu AI uses a novel classification approach based on neural network architecture and images that undergo custom image-processing that includes de-noising and region-of-interest segmentation. Opu Labs has created an ecosystem with more than 150,000 active participants from more than 200 countries who want to track their skin health and interface with doctors, brands, and skincare experts. Opu AI can already detect acne, wrinkles, redness, and hyperpigmentation on images of facial skin. Targets for classification of viral infection include the maculopapular rash (the most common type of rash in a viral infection) as well as related Koplik spots, livedoids, cutis marmorata, perniosis/chilblains, and residual brown skin. To “train” Opu AI to recognize indications of viral infection, we previously collected more than 50,000 general facial skin images and have already collected more than 400 de-identified images related to COVID-19 using an interactive crowdsourcing platform appealing to participants’ altruistic tendencies during the pandemic to support COVID-19 diagnosis and risk mitigation research.
Using its extensive background in technology development, Opu Labs immediately began crafting something to be used as a point solution using its skin scan AI technology. This new app called Opu Health has been developed to gather data and combine it with scans taken of the body, face or anywhere there is skin irritation. The AI will then analyse the data and provide a summary score with recommendations for action such as “see a doctor” or “get tested” or “monitor your skin condition closely”.
The free version allows the user to do up to three scans ensuring that the technology is available to everyone with a smartphone around the world who is concerned with their condition. The premium version of the app unlocks features to enable unlimited scans, connect with a doctor online, the and the ability to share biometric data captured in the app with medical professionals who may be assisting with patient tracking. The price of the premium version is $1.99 per year on the app store which makes the basic app highly affordable.
The solution is attractive wherever there is a need to address the public anxiety that has been becoming more prevalent as people venture into public spaces. Opu Labs has launched the app on the web as a “public beta” version for research purposes to help better train the AI for increased accuracy. The CEO Marc Bookman commented, “the current state of the image recognition is well over seventy percent accurate for known COVID-19 images, however, as the AI learns and grows with each scan, we anticipate the accuracy to increase substantially over a short period of time”. While no AI will ever be 100% accurate for recognizing COVID-19 (no other test is either), when it comes to machine learning, the more data that can be provided, when combined with other tests such as a smell and taste test, the PCR test, and/or the antibody test – medical professionals can identify a larger number of asymptomatic people.
Opu Labs is working alongside other technology partners and governments by providing an API to allow access to the AI for many of the frontline groups addressing the COVID-19 symptom detection. Information can be found at oplabs.com/covid-19-api/. Special licensing programs are available for organizations who are serving disadvantaged populations. Bookman commented, “It’s very important for us to get this into the hands of as many people who need it as possible, with as few barriers to access as possible, and do our part in the fight against COVID-19”.