Initial Prototype Video Comments (Part 1 of 2)

We’re starting to receive comments about the about the initial prototype videos we made. I’ll summarize the comments we currently have.

The first reviewer did not understand why there was a separate cell phone application and a self-contained diagnostic machine. The reviewer liked the idea of test strips, but did not see the need for the gradient, saying that the strips did not need to be most accurate.

My response: We did not make it clear that the colorimetric assay could not be precisely read by the naked eye. Having a machine to read it (by converting it into grayscale and then comparing it against a standardized gradient) is more accurate. But then, one could argue that our test strips do not need to be that accurate – having 6 boxes is enough. I would say, though, that we should not limit our initial conceptual designs to be “just approximate enough” when we know how to make it better – if we are successful, we might even be able to market our product in developed countries. So perhaps we should make different technologically advanced levels of our product for different types of consumers.

The second reviewer liked the idea of having different levels of reading, which allows for redundancy – the lowest is reading the strips with the naked eye, the middle is using the cell phone, the highest level is using the CCD camera. The reviewer also liked the idea that one could save data on cell phone or send it to a central database (for the doctor to monitor the patient’s health).

The reviewer misunderstood how we plan to calibrate the strips. We are not using the white background to calibrate the strips - instead, we are using the gradient that appears next to the test strip for calibration. The reviewer makes a good point about the fact that shadows could significantly distort the reading – perhaps we could place two standardized calibration gradients on the strip – one on either edge (left and right).

The reviewer also makes the great point that we should have some detection method that asks the user to retake the photo if some condition fails (the background is not uniform because of a shadow). Indeed, having “sanity checks” increases the reliability of our program – it ensures that the reading falls into a certain range. Perhaps our program can even return a confidence estimate. Another simple check would be to display the result to the user and wait for a confirmation response.

The reviewer questioned how we were going to exactly develop the box – the reviewer pointed to the fact that it could potentially be cheaper (in addition to being more accurate), but it might be too complex, given the limited time that we have.