SARS-CoV-2 is a multi-organ disease characterized by a wide range of symptoms, which also causes the severe acute respiratory syndrome. When it initially began, it rapidly spread from its origin to adjacent nations, infecting millions of people around the globe. In order to take appropriate preventative and precautionary actions, it is necessary to anticipate positive COVID-19 instances in order to better comprehend future risks. Therefore, it is vital to building mathematical models that are resilient and have as few prediction mistakes as feasible. This research recommends an optimization-based Least Square Support Vector Machines (LSSVM) for forecasting COVID-19 confirmed cases along with the daily total vaccination frequency. In this work, a novel hybrid Barnacle Mating Optimizer (BMO) via Gauss Distribution is combined with the Least Squares Support Vector Machines algorithm for time series forecasting. The data source consists of the daily occurrences of cases and frequency of total vaccination from February 24, 2021, to July 27, 2022, in Malaysia. LSSVM will thereafter conduct the prediction job with the optimized hyper-parameter values using BMO via Gauss distribution. This study concludes, based on its experimental findings, that hybrid IBMOLSSVM outperforms cross validations, original BMO, ANN, and a few other hybrid approaches with optimally optimized parameters.