Currently, there is considerable interest in constitutive relations for accurate heat transfer prediction during cryogenic feed line chill-down to optimize the cryogen requirement and quench time. The present work develops a new database for transient quench flow film boiling with 14,143 inverted annular (IAFB), 9299 inverted slug (ISFB), and 3257 dispersed flow (DFFB) film boiling data from LH2, LN2, LAr, LOX, and LCH4 chill-down experiments with transfer line length ranging from 0.11 m to 11 m, inner diameter from 0.4 cm to 2.2 cm, and under various orientations and gravity levels. Moreover, in the collected data, mass flux spans 8.8–1805 kg/m2s and local pressure from 72.3 kPa to 1246.2 kPa. The database is used to analyze 33 existing film boiling correlations to understand their predictive performances and limitations. Accordingly, new IAFB, ISFB, DFFB, and combined IAFB and ISFB inverted flow regime heat transfer correlations are proposed using mutual information and conditional mutual information statistical filter algorithms and genetic algorithm-based optimization. The predictions from the new correlations are in good agreement with the experimental results with a mean absolute error of less than 29%. Hence the proposed correlations can be added to thermal-hydraulic codes to support the design of reliable and energy-efficient cryogenic fluid transfer.