<p>Several studies have extensively explored the axial (uplift and compressive) behavior of helical piles, while lateral behavior—an equally critical aspect in the design of foundation systems— remains relatively uncharted. In this study, key geometric features with significant influence on lateral behavior were explored through machine learning to derive optimization insights. The lateral response of the piles was modelled using the modified p-y springs method, with installation effects accounted for through empirically adjusted geotechnical parameters from field measurements. The results indicate that the optimal helix-to-pile diameter ratio (dH/dP) is dependent on the helix-to-pile embedment ratio (zH/zP) and is strongly influenced by the applied-to-failure load ratio (aL/fL) of each unique configuration. Response surface analyses further reveal that the contribution of the helix to lateral resistance diminishes beyond a zH/zP ratio of 0.2 and becomes negligible past 0.6 — at which point the pile behaves similarly to a conventional shaft without helical reinforcement. The geometric features identified at Test Site I were then applied to guide the pile configurations for Test Sites II and III, with lateral behavior validated through comparison between simulated profiles and strain gauge-derived field measurements, using site-specific geotechnical data. Notably, the derived geometric features demonstrated a degree of independence from localized soil conditions, suggesting their potential applicability across a wide range of ground profiles. Overall, the findings of this study offer both a practical and comprehensive framework for the design of laterally loaded helical piles, enabling practitioners to achieve an optimized balance between geo-structural performance and material efficiency.</p>