The presence of all five of the virulence-associated genes iucA, iroB, peg-344, rmpA, and rmpA2 is presently the most accurate genomic means for predicting hypervirulent Klebsiella pneumoniae (hvKp-p). With this longitudinal cohort study, we firstly provide novel insights into the clinical and genomic characteristics of hvKp-p in high-risk regions. Through propensity score matching, we show that hvKp-p is less likely to acquire antimicrobial resistance but develops more severe disease and result in increased mortality. HvKp-p are predominantly isolated from hospital settings and caused pneumonia in majority of the cases. ST23 and KL1 are the most common types in the hvKp-p cohort. Community-acquired and healthcare-associated infections are also identified as independent risk factors for hvKp-p. This genomic definition, albeit imperfect, offers a practical and efficient alternative to murine models, allowing for early identification and timely intervention in clinical settings.