{We use a 362 fb$^{-1}$ sample of $e^+ e^-$ collisions at $\Upsilon(4S)$ resonance collected with Belle II detector at SuperKEKB collider to search for the rare decay $B^+ \to K^+ \nu \bar{\nu}$. The main strategy consists in exploiting the inclusive properties of the other $B$ to suppress the background. Another analysis is based on a conventional hadronic reconstruction of the accompanying B meson and is used to corroborate the first strategy. Both the procedures have been validated with several control samples.
A maximum likelihood fit is used to extract the branching ratio, which results in $[2.7\pm 0.5\text{(stat)} \pm 0.5 \text{(syst)}] \times 10^{-5} $ for the main analysis and $[1.1^{+0.9}_{-0.8}\text{(stat)}^{+0.8}_{-0.5} \text{(syst)}] \times 10^{-5}$ for the support analysis.
The combination of the two analyses gives a branching fraction of
$[2.3 \pm 0.5\text{(stat)}^{+0.5}_{-0.4} \text{(syst)}] \times 10^{-5}$ which corresponds to the first evidence of the decay with 3.5 standard deviations and 2.7 standard deviations above the standard model expectation.