The genome-wide association studies identified genes associated with many diseases, but the identification and verification of disease variants are still challenging due to small effects and large number of individual variants. In this paper, we propose a powerful method that first quantifies the strength of regional associations at each single nucleotide polymorphism and converts these measures into time series data before using a change point detection algorithm to identify significant regions. In our extensive simulation study, the proposed method consistently demonstrates greater power than existing alternatives, achieving a relative increase of over 20% in challenging scenarios where true causal variants are sparse and multiple association regions exist at the same time, while maintaining a lower false positive rate.