The　degradation　of　conventional　steel　reinforcement　in　concrete,　particularly　in　marine　environments,　has　driven　a　shift　toward　using　glass　fiber-reinforced　polymer　(GFRP)　rebars　due　to　their　superior　corrosion　resistance.　However,　GFRP　rebars　are　prone　to　deterioration　in　mechanical　properties　when　exposed　to　prolonged　hygrothermal　conditions.　Understanding　the　impact　of　environmental　and　mechanical　factors　on　the　Tensile　Strength　Retention　(TSR)　of　GFRP　rebars,　both　bare　and　embedded　in　concrete,　is　crucial　for　evaluating　their　durability　in　concrete　structures.　This　study　employs　an　experimental　approach　combined　with　machine　learning　techniques,　specifically　the　XGBoost　model,　to　predict　TSR　under　various　conditions.　The　experimental　procedure　involved　preloading　the　bars,　embedding　some　in　concrete　while　leaving　others　bare,　followed　by　conditioning　at　room　and　elevated　temperatures　(60　degrees　C)　for　up　to　183　days.　Subsequently,　a　total　of　224　data　points　were　obtained　through　accelerated　aging　tests　and　used　to　buit　the　model.　The　latter　achieved　high　accuracy,　with　R2　values　of　0.99　and　0.87　for　the　training　and　testing　datasets,　respectively.　Furthermore,　SHAP　analysis　identified　immersion　time,　temperature,　and　preloading　as　key　factors　influencing　GFRP　degradation.　Notably,　the　10　mm　rebars　embedded　in　concrete　exhibited　greater　degradation　compared　to　those　directly　immersed　in　seawater　with　TSR　of　80.40　%　and　88.3　%　respectively.　In　contrast,　the　gap　between　the　TSR　of　embedded　and　bare　bars　reduced　considerably　for　25　mm　rebars　with　values　of　87.8　%　for　the　former　and　89.8　%　for　the　latter.　This　study　offers　predictive　tools　for　forecasting　the　TSR　of　aged　GFRP　rebars.　It　provides　valuable　insights　for　optimizing　GFRP　bar　performance　under　adverse　conditions,　emphasizing　the　importance　of　environmental　and　mechanical　factors　in　their　practical　application.