This is a very insightful read; you should gradually expand the sample size in order to test the accuracy even further.

Also, a query, at what stage does the Covid 19 have to be at in order to be detected by this system? Can it detect cases wherein symptoms have not shown yet?

If you have data sets for different stages of the virus aka pre-symptoms, early-symptoms, post-symptoms, severe/critical cases etc. I think you should train the model on those data sets separately in order to check for accuracy on each of the stages; this may be helpful