Objective To explore the relationship between serum ischemia-modified albumin （IMA）， soluble growth stimulation expressed gene 2 protein （sST2）， apolipoprotein B （ApoB）/ApoA1 and poor ST segment regression in electrocardiogram after percutaneous coronary intervention （PCI） in patients with acute ST-segment elevation myocardial infarction （STEMI）. Methods A total of 102 STEMI patients who received PCI treatment in the emergency department of the Second Affiliated Hospital of Anhui Medical University between September 2022 and September 2025 were selected. These patients were then divided into the poor regression group （with an ST-segment regression rate < 50%， n = 31） and the good regression group （with an ST-segment regression rate ≥50%， n = 71） based on the ST-segment regression rate of the postoperative electrocardiogram. The levels of serum IMA， sST2， and ApoB/ApoA1 were compared between the two groups before PCI and 2 hours after PCI. Demographic data， medical history data， and PCI-treatment-related data of the two groups were collected to screen for the influencing factors of poor ST-segment regression after STEMI. The predictive value of preoperative serum IMA， sST2， and ApoB/ApoA1 for poor ST-segment regression was analyzed. Results Among the 102 patients， 31 （30.39%） exhibited poor electrocardiogram regression 2 hours after the operation， and 71 （69.61%） showed good regression. The percentages of patients with a time from onset to admission > 6 hours， left anterior descending coronary artery infarction， and Killip grade ≥ Ⅱ were higher in the poor-regression group， whereas the percentage of TIMI blood flow grade 3 was lower （P < 0.05）. When comparing the same group before and 2 hours after the operation， serum IMA， sST2， and ApoB/ApoA1 levels decreased in both groups （P < 0.05）. The levels of serum IMA， sST2， and ApoB/ApoA1 in the poor-regression group before and 2 hours after the operation were higher （P < 0.05）. The area under the curve， sensitivity， and specificity of preoperative serum IMA for predicting poor ST-segment regression in STEMI patients after PCI were 0.818， 80.65%， and 73.24%， respectively； those of sST2 were 0.778， 64.52%， and 95.77%， respectively； those of ApoB/ApoA1 were 0.866， 70.98%， and 90.14%， respectively； and those of the combined prediction were 0.931， 77.42%， and 98.59%， respectively. Among them， the combined prediction had the highest efficacy （P < 0.05）. Left anterior descending coronary artery infarction （OR = 2.962）， preoperative IMA （OR = 2.354）， preoperative sST2 （OR = 2.517）， and preoperative ApoB/ApoA1 （OR = 2.863） were risk factors influencing ST-segment regression after PCI in STEMI patients （P < 0.05）. A multi-factor logistic regression model was constructed： logit（P） = -2.568 + 0.245 × Time from Onset to Admission （h）+1.086 × Infarction of the Left Anterior Descending Coronary Artery + 0.702 × Killip Classification + 0.935 × TIMI Flow Grade + 0.856 × Preoperative IMA + 0.923 × Preoperative sST2 + 1.052 × Preoperative ApoB/ApoA1. The area under the curve of this model for predicting poor ST-segment resolution after PCI in STEMI patients was 0.956 （95% CI： 0.897 - 0.987）， with a sensitivity of 87.10% and a specificity of 92.96%， Z = 22.366， P < 0.001. The goodness-of-fit test of the model yielded a χ² value of 5.971， with P > 0.05， which indicates good calibration of the model. The DCA curve showed that when the threshold probability was between 6% and 92%， using the model as a reference could achieve high clinical benefits. Conclusions Infarction of the left anterior descending coronary artery， preoperative high ratios of IMA， sST2， and ApoB/ApoA1 are all associated with poor ST-segment regression after PCI in STEMI patients. Moreover， the combined model shows relatively high feasibility in predicting the occurrence of adverse events related to ST-segment regression.