Objective: To explore the multi-time-window dynamic variation characteristics of aquaporin 4(AQP-4), macrophage inflammatory protein-1α(MIP-1α), and complement C1q/tumor necrosis factor-related protein 12(CTRP12) in patients with acute ischemic stroke(AIS), analyze their association with neurological deterioration trajectory, and to construct a risk stratification and prediction model. Methods: A total of 100 patients with AIS with symptom onset within 24 hours admitted to Tangshan Central Hospital from January 2023 to June 2024 were enrolled. Among them, 58 patients meeting thrombolysis indications received recombinant tissue-type plasminogen activator(rt-PA) therapy(treatment group), while 42 patients beyond the thrombolysis time window or with thrombolytic contraindications did not undergo thrombolysis(untreated group). Peripheral blood samples were collected to detect the three biomarkers at 10 time windows after onset: 3 h, 6 h, 12 h, 24 h, 36 h, 48 h, 3 d, 7 d, 14 d and 30 d. The untreated group received conventional regimens including antiplatelet therapy, lipid-lowering and plaque-stabilizing treatment, as well as blood pressure and glycemic control. Linear mixed-effects models were adopted to analyze the temporal variation trends of biomarkers. Time-series clustering algorithms were used to stratify patients by neurological deterioration risk. Multivariate regression and time-dependent covariate survival analysis were performed to build predictive models, and the dynamic correlation between the three biomarkers and the National Institutes of Health Stroke Scale(NIHSS) scores was quantified. Results: For all subjects, serum AQP-4 peaked at 36 h at(45.3±7.1) pg·mL^-1 and sustained at(24.1±4.2) pg·mL^-1 on day 30; MIP-1α reached the peak concentration of(52.3±7.8) pg·mL^-1 at 6 h, with a residual level of(24.7±3.9) pg·mL^-1 at day 30; CTRP12 declined to the trough value of(8.0±1.0) pg·mL^-1 at 48 h and then rebounded to(13.5±2.2) pg·mL^-1. Intergroup differences at all time windows were statistically significant(P<0.05). rt-PA thrombolysis markedly suppressed the acute upregulation of AQP-4 and MIP-1α. Among patients complicated with neurological deterioration, T3(24 h) AQP-4 was(33.5±7.1) pg·mL^-1 in the untreated group versus(20.5±4.6) pg·mL^-1 in the treatment group; T2(12 h) MIP-1α was(132.5±27.3) pg·mL^-1 in the untreated group versus(85.7±19.3) pg·mL^-1 in the treatment group(all P<0.05). Clustering analysis categorized patients into three deterioration risk subgroups(Cluster 1:28 cases; Cluster 2:42 cases; Cluster 3: 30 cases). The phenotype of severe edema-excessive inflammation-insufficient repair carried the highest risk of neurological worsening. Multivariate regression models revealed that at T1(6 h), AQP-4(r=0.68) and MIP-1α(r=0.72) were strongly positively correlated with NIHSS scores, whereas CTRP12(r=-0.58) presented a negative correlation(adjusted R²=0.45). The concentration inflection points of the three biomarkers were highly synchronized with neurological function turning points. Conclusion: The temporal dynamic fluctuations of AQP-4, MIP-1α and CTRP12 can objectively reflect the progression of blood-brain barrier injury, inflammatory cascade and endogenous neurorepair after AIS. The peak and trough inflection points of the three biomarkers are consistent with the turning points of neurological function, which can provide critical time nodes for early prognostic assessment and individualized precise intervention.