For 5G New Radio services with the requirement of high reliability and low latency, the HARQ loop can only afford one or a couple of retransmissions within the latency bound, or none at all, depending on the system configuration. In response to such stringent demand on the link adaptation algorithm, improved algorithms for predicting or estimating downlink channel conditions are desired. This paper describes the problem and proposes a solution where channel state statistics are reported by user equipment to aid the base station for more accurate downlink scheduling. Simulation results are provided to demonstrate the benefit of the new approach.