Not yet reviewed by Pith; the record is open.

Pre-merger gravitational-wave (GW) sky-localisation of binary neutron star (BNS) and neutron star black hole (NSBH) coalescence events, would enable telescopes to capture precursors and electromagnetic (EM) emissions around the time of the merger. We propose a novel astrophysical scenario that could provide early-warning times of hours to days before coalescence with sub-arcsecond localisation, provided that these events are gravitationally lensed. The key idea is that if the BNS/NSBH is lensed, then so must the host galaxy identified via the EM counterpart. From the angular separation of the lensed host galaxy images, as well as its redshift and the (foreground) lens redshift, we demonstrate that we can predict the time delays assuming a standard lens model. Encouraged by the non-trivial upper limits on the detection rates of lensed BNS/NSBH mergers that we estimate for upcoming observing runs of the LIGO-Virgo-Kagra and third generation networks, we assess the feasibility and benefits of our method. To that end, we study the effect of limited angular resolution of the telescopes on our ability to predict the time delays. We find that with an angular resolution of $0.05''$, we can predict time delays of $> 1$ day with $1\sigma$ error-bar of $\mathcal{O}$(hours) at best. We also construct realistic time delay distributions of detectable lensed BNSs/NSBHs to forecast the early-warning times we might expect in the observing scenarios we consider.