How will this work?
Virufy is developing an AI algorithm to predict COVID-19 infection within minutes based on cough sound recordings. Our work is backed by our own results in clinical studies and research from prestigious institutions including CMU, MIT, and Cambridge.
The sound of your cough and breathing has long been utilized by medical professionals as a judgement of respiratory health and for diagnosis of various diseases. Before the COVID-19 pandemic, medical personnel would collect these sounds from a patient and subsequently make a diagnosis through in-person office visits. Although this previously common diagnostic method saves both time and money, the feasibility of an in-person diagnosis of COVID-19 through a patient’s cough and breathing is almost null due to the essentiality of social distancing and limitation of medical personnel during this pandemic.
However, with the spread of Artificial Intelligence (AI) in diagnostic technology, it may be possible to diagnose COVID-19 through a simple recording of a cough into a smartphone app. Due to the ability of AI algorithms to pick up on minute - yet still distinguishable - patterns in audio features, AI has been able to exhibit high sensitivity and specificity in classifying respiratory diseases [1, 2, 3, 4, 5 ]. Previous examples of successful AI diagnostic models include those that diagnose wheezing and rhonchi , pertussis , asthma , and pneumonia  - all of which utilized the unique respiratory signature of each disease to differentiate positive cases.
Similar to all other respiratory diseases, COVID-19 creates a unique respiratory signature in the throat and lungs that is distinct from other respiratory infections which produce a charataristic wet cough. Consequently, previous research suggests that cough sounds can be analyzed to detect COVID-19. Globally, this idea is being actively researched by several prestigious institutions, including CMU, MIT, and Cambridge. For example, a crowdsource research done by University of Cambridge showed that a simple binary machine learning classifier can classify COVID-19 positive patients through breathing and coughing sounds with high accuracy (AUC = 0.7). Similarly, researchers at CMU identified 18 voice features that distinguish positive COVID-19 patients and trained a model to diagnose COVID-19 with a 89.1% accuracy
Based on past research, Virufy is developing an AI algorithm that can be used to accurately predict a COVID-19 infection within minutes based on recordings of cough sounds. However, as opposed to previous COVID-19 cough research that targeted the US population, Virufy aims to collect data from multiple sites across the globe. As a student-run initiative with volunteers spanning several countries, Virufy is developing a COVID-19 diagnostic model with greater racial and spatial inclusivity through data that includes a range of ethnicities and community-specific phonological differences.
Currently, Virufy has developed a model with an 80% accuracy from clinical data derived from several countries and spanning multiple ethnicities. However, while our model is of high accuracy, we recognize that this is not enough. We need your cough in order to refine our model and ultimately develop a free COVID-19 diagnostic model that can easily and instantly provide a COVID-19 diagnosis through a smartphone application.
Below are a few examples of research projects that give us confidence in the scope of developing an AI algorithm to be used for COVID-19 detection:
A Covid-19 Sounds App
A crowdsourced research study conduted by University of Cambridge. They used cough samples and breathing to understand how discernible COVID-19 sounds are from those in asthma or healthy controls. Their results show that even a simple binary machine learning classifier is able to classify correctly healthy and COVID-19 sounds. Our models achieve an AUC above 70% across all tasks. This work inspires further investigation of how automatically analyzed respiratory patterns could be used as pre-screening signals to aid COVID-19 diagnosis .
COVID Voice Detector
A study by Carnegie Mellon University aimed to collect a large number of voice samples to train AI for diagnosis of COVID. The rationale behind the study is that, "the sound of our voice (regardless of language), and the sounds we make when we breathe or cough change when our respiratory system is affected. The changes range from coarse, clearly audible changes, to minute changes -- what we call "micro" signatures, that are not audible to the untrained listener, but are nevertheless present" .
Hi Sigma, do I have the Coronavirus? Call for a New Artificial Intelligence Approach to Support Health Care Professionals Dealing with the COVID-19 Pandemic.
MIT's Department of Mechanical Engineering proposed an idea to detect positive cases of COVID by collecting cough samples through the phone to train artificial intelligence and subsequently build a diagnostic algorithm .
Artificial Intelligence has been used in the past for diagnosis!
Cough Sound Analysis for Pneumonia and Asthma Classification in Pediatric Population
In this study, AI was used to distinguish between the coughs of asthma and pneumonia, in order to provide medical care to developing countries with poor resources. Their method achieved a sensitivity of 89%, specificity of 100% and Kappa of 0.89. These results show the potential use of AI in detection and differentiation of respiratory sounds .
A Cough-Based Algorithm for Automatic Diagnosis of Pertussis
This study examined pertussis cough, croup, and cough containing wheezing sounds corresponding to diseases such as bronchiolitis and asthma to train AI to detect Pertussis. The algorithm was able to diagnose all pertussis cases successfully from audio recordings, automatically detecting individual cough sounds with 92% accuracy and PPV of 97%. Overall, the study supports the use of AI as a potential candidate for differentiating and diagnosing respiratory sounds .
-  Y. Amrulloh, U. Abeyratne, V. Swarnkar and R. Triasih, "Cough Sound Analysis for Pneumonia and Asthma Classification in Pediatric Population," 2015 6th International Conference on Intelligent Systems, Modelling and Simulation,Kuala Lumpur, 2015, pp. 127-131, doi: 10.1109/ISMS.2015.41.
-  “Coronavirus (COVID-19).” National Institutes of Health, U.S. Department of Health and Human Services, 31 Mar. 2020, https://www.nih.gov/health-information/coronavirus.
-  Pramono, Renard & Imtiaz, Anas & Rodriguez-Villegas, Esther. (2016). A Cough-Based Algorithm for Automatic Diagnosis of Pertussis. PloS one. 11. e0162128. 10.1371/journal.pone.0162128.
-  Kvapilova, Lucia, et al. “Continuous Sound Collection Using Smartphones and Machine Learning to Measure Cough.” Digital Biomarkers, vol. 3, no. 3, Oct. 2019, pp. 166–175., doi:10.1159/000504666.
-  Ferrari, Sara, et al. “Cough Sound Analysis to Identify Respiratory Infection in Pigs.” Computers and Electronics in Agriculture, vol. 64, no. 2, 2008, pp. 318–325., doi:10.1016/j.compag.2008.07.003.
-  Maghdid et al., “A Novel AI-Enabled Framework to Diagnose Coronavirus COVID 19 Using Smartphone Embedded Sensors.” https://arxiv.org/abs/2003.07434
-  Covid-19 Sounds App—University of Cambridge. (n.d.). Retrieved April 12, 2020, from http://www.covid-19-sounds.org/
-  COVID Voice Detector—Carnegie Mellon University. (n.d.). Retrieved April 12, 2020, from https://cvd.lti.cmu.edu/
-  “Breath Sounds.” Healthline, https://www.healthline.com/health/breath-sounds .
-  “Hi Sigma, do I have the Coronavirus?: Call for a New Artificial Intelligence Approach to Support Health Care Professionals Dealing With The COVID-19 Pandemic”- MIT Retrieved July 2020, from https://www.researchgate.net/publication/340644305_Hi_Sigma_do_I_have_the_Coronavirus_ Call_for_a_New_Artificial_Intelligence_Approach_to_Support_Health_Care_Professionals_Deal ing_With_The_COVID-19_Pandemicz