What? I don't run the damn thing, if you want the methodology, you'd need to contact them and see if they're willing to share their models.
And no, the symptoms are not the same. If you respond that you feel ill on the app during the daily check in, it asks you a list of 20 or so questions about your symptoms and also whether or not you've had a Covid test- I posted these earlier in the thread when you were casting shade on it last time.
Again, I don't run it, but I'd be pretty sure it then calculates a probability of the person having Covid based on previous responses correlated against swab test data and then uses that in aggregate to provide its headline figures. As ever, the actual figures are less important than the trend, which is clearly up by any metric.