Infrared　(IR)　nonlinear　optical　(NLO)　crystals　are　essential　for　laser　applications　in　mid-　or　far-IR　regions.　The　common　mid-IR　NLO　crystals　are　inadequate　under　high-power　laser　mainly　due　to　low　laser-induced　damage　threshold　which　is　closely　related　to　the　lattice　thermal　conductivity　(κL).　Here　a　combined　method　involving　automatic　structure　prediction　algorithms,　first-principles　calculations　and　machine　learning　model　is　used　to　predict　and　evaluate　a　new　IR　NLO　material　family　with　the　chemical　formula　of　Li2ZnS2.　Totally,　three　Li2ZnS2　structures　(or　phases)　are　designed,　where　one　phase　is　expected　to　meet　the　stringent　requirements　for　mid-IR　NLO　materials　pumped　by　high-power　laser,　with　the　high　thermal　conductivity,　decent　NLO　coefficient,　wide　bandgap　and　phase-matching　ability　in　the　mid-IR　region.　This　study　presents　an　efficient　approach　to　exploring　mid-IR　NLO　materials　with　high　κL　based　on　accumulated　knowledge　concerning　the　structure-property　relationship　in　the　NLO　field　and　provides　inspiration　to　improve　κL　prediction.　©　2025　Wiley-VCH　GmbH.