Population flow based spatial-temporal eigenvector filtering modeling for exploring effects of health risk factors on COVID-19

•The FLOW-ESTF model consider both spatial-temporal and population flow connectivity.•The FLOW-ESTF model outperformed GLM, ESF and ESTF models in model performance.•The extracted EV project spatial-temporal patterns as possible omitting variables.•The extracted population flow EV helped to explore pandemic transmission nodes.•The EV explain 15% higher variance than Google mobility data in COVID-19 modeling.