Here right now in Myanmar, there are thousands of new covid-19 cases day by day. And the diagnostic tests are not available in most parts of the country. Anyway, in some cities, we use the RDT-antigen test for diagnosis. It costs around about 5 dollars. In places where the RDT test is not available, we just use clinical diagnosis by observing obvious symptoms such as fever and cough, sometimes along with the history of contact to a confirmed positive person, and then started the clinical management. Admittedly speaking, as a clinician, I won’t use this machine learning algorithm for the diagnosis in my setting. I don’t know much about machine learning. But I believe this paper delivers other important points to the covid-19 issue. I think it alone might not give much sense in terms of a practical impact. But, I think this is one of the first steps and a part of the bigger picture that the machine learning field would help us fight against the pandemic. What I mean is it has been already cited by the other 11 articles and giving inputs to the existing literature. And it might encourage the other IT professionals to do more research about the covid-19. On a bigger picture, being a part of the many steps may offer something we do not see and perhaps deliver a new concept regarding Covid-19 diagnosis.

I completely agree with Matt’s comment. The paper alone does not have impact in diagnosis of Covid-19, however, it gives insights into future solutions and research for limited resource setting. I’m thankful for your fascinating presentation, Naphat.

I don’t have any experience about machine learning. It’s very interesting topic to know about machine learning model.

The current situation of COVID-19 in Thailand is not good, there are many new cases per day and deaths as well. In my opinion think that this model from the article would not fit with this situation because there are many factors that should be considered such as a who was diagnoses with no any symptoms, or who got vaccine. Other reason is that the sensitivity and specificity of this model is still low compare to the available method and the rapid diagnosis test is available which would more effectiveness than the machine learning model.

Thank you for presentation with new learning topic to us.
1. It might be better if we could do random confirmation to tested cases with RDT only.
2. I do not have such experience and I’m learning through your paper and presentation as well as university lectures.