The core of our machine learning is Ax=b :grins:
More seriously, the main reason why traditional sanger sequencing can't be used for COVID-19 testing is because it would be unclear whether a lack of signal is truly due to lack of virus, or if it is just because the assay failed (happens all the time!)
What we've done is introduce a reference sequencing signal that is biochemically very similar to viral RNA, but produces a distinct vector of electrical signals that is different from the signals emitted by viral RNA. Since we know what both the reference and viral signals look like, we can perform linear regression analysis to fit the linear combination of viral and reference signals that best match our data.
