Currently, few research institutes and universities are working on this problem of detecting people affected with CoVID 19 in the vicinity using voice recognition models. But are there any alternate pathways? If yes, then what are they and how to make a data survey pertaining to each of them, so as to obtain some useful inferences?
In addition to this, it is not always feasible to detect such just by asking for personal details like whether the person has visited any foreign country, or whether he has come in contact with any CoVID 19 patient. So, given these limitations, can anyone please elaborate on the said matter?
Body temperature scanning may be helpful once the patient is actively infected.
The challenge is detecting asymptomatic patients who are infected and can spread the virus.
In machine learning, we may use images analysis, via classification/segmentation/clustering or object tracking. If we consider images of viruses, we will have to distinguish it from other viruses e.g., influenza, Ebola etc via image classification technique etc. Another, way is to consider images of the affected parts of the patient by comparing it with the confirmed affected parts images of the previously affected patients. Also a good way is to use blood vessels segmentation/analysis etc.
Due to very high undocumented and asymptomatic COVID-19 people, almost everyone should be suspected of having had infection in a person's vacinity.
Based on antibody testing data shared by the Governor of New York, more than 10 times the number of people who tested positive have had exposure to virus - indicating 90% undocumented (or asymptomatic) infected COVID-19 cases(1).
Therefore, most of the people already infected would not be picked up if we look for fever, dry cough, sore-throat, myalgias.
However, screening suspected people at the airports, businesses, utility stores, hospitals may pick up people by temperature checking and quick questionnaire (notwithstanding the limitations of commonly used thermal temperature guns)(2)
Therefore the definite way is "test, test, and test" as recommended by WHO by RT-PCR initially and a validated antibody test for SARS-CoV-2 from week 2 to 4 weeks.
Therefore distance measures, use of a mask, and hand hygiene after any contact is the order of the day and time.
1. https://www.ny1.com/nyc/all-boroughs/news/2020/04/23/governor-cuomo-daily-coronavirus-update-with-first-antibody-test-results
2. https://www.todayonline.com/world/thermometer-guns-covid-19-front-lines-are-notoriously-not-accurate
Article On Facing the SARS-CoV-2 (COVID-19) with Combination of Nano...
https://www.who.int/news-room/commentaries/detail/advice-on-the-use-of-point-of-care-immunodiagnostic-tests-for-covid-19
Subhamoy Biswas , I would like to know about the voice recognition model you have mentioned.
As mentioned by Mrs. Mousumi Basu , there is a website designed by Carnegie Mellon University where they are requesting the public openly to record their voice clips and upload in their website, so that they can perform their study.
The link for the website is: https://cvd.lti.cmu.edu/
Thank you Subhamoy Biswas very much for mentioning the website. It is indeed an unique approach.
Dr. Mousumi Basu
May be this will be of help:
Article IoT-Q-Band: A low cost internet of things based wearable ban...
https://www.who.int/news-room/commentaries/detail/advice-on-the-use-of-point-of-care-immunodiagnostic-tests-for-covid-19
CoVID 19 patients will not show immediate symptoms and even some will not show physically observable symptoms at all. Therefore it makes it challenging to derive simple parameters for effective classification. Proper screening is the best way to confirm even when symptoms appear.
From a research point of view, for general judgement, there are protocols to check for symptoms that could recommend if a screening is required. These qualitative or quantitative observations could be mapped with screening results to create machine learning models to observe if good predictions could be made.
Encourage the government to do an anonymous consensus of all according to precise location (not home address). A survey of all possible encounters (visited this shop at this time of day bought this vegetable; came into contact with these products since contraction; my diet consists of; I have these health issues; questions that can articulate to AI how each occurrence occurred.
Most people who are infected with the SARS-CoV-2 Coronavirus show no symptoms, do not have Covid-19 disease, and do not know that they are infected. In individual countries, according to various estimates, it is from 70 to 80, and sometimes even about 90 percent. infected who have this infection completely asymptomatically. The remaining few - several percent. days infected with Coronavirus shows symptoms of a mild cold, which usually recovers quickly (after a few days) without any medical treatment. In contrast, a relatively small percentage of infected people who develop Covid-19 show symptoms of general weakness, raised body temperature, breathing problems, and other symptoms of serious respiratory disease. Besides, there may also appear other disease symptoms that are associated with other diseases and comorbidities. A complete diagnosis to check whether a specific person is infected with Coronavirus can only be carried out using specialized tests carried out in medical conditions with the use of specially manufactured, licensed, and certified verification tests for the presence of Coronavirus. In addition, applications installed on smartphones, created in individual countries in cooperation with public health care institutions and / or public institutions of the crisis management system, which inform smartphone users about the scale of the risk of potential Coronavirus infection in a specific local, regional, city, district cities on the basis of a constantly updated database of contacts of a specific person with other people in public space, at work, at home, etc.
Best regards,
Dariusz Prokopowicz
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Artificial Intelligence