Transit predictions are not as simple as many assume. Low Earth orbits are unstable, and even in the absence of an orbital altitude boost, the predictions are associated with error. The prediction is based upon a simplified ballistics calculation using a model called the SGP4 Propagator, which uses orbital element data that is supplied by NASA and NORAD, called two line elements (TLEs). These are published every few hours for the ISS, and the prediction will change with each new set of data. Predictions more than one week out from the event should be viewed with skepticism, and will often shift by significant amounts, enough to cause a local event to disappear. This is especially true if an orbital boost event occurred during the intervening period. It's best to continually check the prediction, and only set up your equipment based upon data that was published in the 24 hours preceding the transit. And even in these cases, you will often notice an error of several hundred meters along the ground, up to 1km, depending on how stringent your target window is. Some of this information is included in a few posts I've made in the past year.
Edited by Tom Glenn, 05 March 2021 - 01:25 PM.