NeuTigers, an artificial intelligence company spun out of Princeton University, has launched CovidDeep, a clinically validated solution that can triage those needing further testing for SARS-CoV-2/COVID-19 using physiological sensors data derived from wearable devices.
The CovidDeep app is 90%+ accurate in predicting whether a person is virus-free or virus-positive. Early in the pandemic, NeuTigers realised its StarDeep Smart Healthcare Platform, which delivers AI- powered solutions to augment healthcare professionals’ productivity, could be used in the fight against COVID-19. CovidDeep was developed through clinical research and validation trials beginning at San Matteo Hospital in Pavia, Italy, during the acute phase of the COVID-19 pandemic in April 2020, with further expanded prospective field studies in hospitals in France and the United States to explore its robustness.
CovidDeep is already deployed in B2B settings including multiple nursing homes and assisted living facilities in America and Europe. It works by collecting physiological sensor data from the Empatica E4 wristband, blood pressure and blood oxygen levels from any off-the-shelf, standalone monitors as well as personal health symptoms from a brief questionnaire. The technology is already being adapted to work with connected health products from Fitbit, Withings, Apple, Samsung, and other devices. The CovidDeep consumer app is expected in early 2021.
Deep Neural Networks
COVID-19 affects people’s biometrics and physiological markers in both obvious and nearly imperceptible ways. Using advanced algorithms of machine learning, CovidDeep can detect changes in these physiological patterns before they are felt by the patient and all with real- time analysis.
CovidDeep is powered by cutting-edge AI deep neural networks that mimic how the human brain perceives, learns, and interprets the world. NeuTigers’ research co-founders at Department of Electrical Engineering at Princeton University used proprietary deep neural networks to learn from hundreds-of- thousands of digital health data points and a specific questionnaire in SARS-CoV-2-positive and healthy participants. They identified patterns in the sensor physiological readings such as Galvanic Skin Response (GSR), Skin temperature, Heart Inter-beat Interval (IBI), Blood pressure, and Blood oxygen saturation levels (SpO2) that are consistent with how COVID-19 impacts the body. CovidDeep can recognise the ‘digital signature’ of SARS-CoV-2/COVID-19 and identify if a person is COVID- positive, even if they are asymptomatic.
CovidDeep is shown from the controlled clinical study at San Matteo hospital to predict SARS-CoV- 2/COVID-19 with upwards of 90% accuracy.
Adel Laoui, CEO and founder of NeuTigers, said: “Advances in machine learning and the proliferation of medical-grade sensors in everyday consumer wearables has led to a new era in which we can predict and identify the onset of a myriad of diseases. Before the pandemic, our teams were using predictive models in our StarDeep health platform to monitor and screen for diabetes and mental health conditions.”
“Realising our AI technology could be valuable in the fight against COVID-19, we quickly pivoted and developed a new solution after rigorous clinical research trials. Initially meeting the urgent need for mass screening in the business environment, CovidDeep is set to expand to a wider consumer offering in early 2021.”
Users simply answer a questionnaire regarding symptoms and health history (based on CDC guidelines) and input their health sensor’s data. Data is entered by connecting CovidDeep to an Empatica E4 Wristband as well as inputting blood pressure and blood oxygen readings using any off-the-shelf device. All data remains local to the device, is never shared, and stays under the user’s control.
CovidDeep then analyses the data and provides a prediction as to whether someone is likely to be negative or positive for SARS CoV-2/COVID-19. The process takes around two minutes, allowing one Empatica device, blood pressure monitor and pulse oximeter to screen unlimited numbers of people after being sanitised between usages.
Available globally from January 12, 2021, on Android with iOS availability coming soon, users can try CovidDeep for free with unlimited screening results for seven days.
Laoui added: “Whether helping frontline healthcare workers, employers keeping the workplace safe, assisting airlines in reestablishing consumer confidence in travel, helping students to stay in schools, or providing fast access to sporting and entertainment venues, the applications of CovidDeep are endless.”
NeuTigers is continuing extended clinical field testing worldwide to further hone the app’s accuracy to ensure that no geographic, demographic, or ethnic biases occur. As well as completed trials in Italy, validation field-testing studies are currently ongoing in the United States, France and planned in North Africa, Switzerland, Japan, and India.