Objective:To screen the independent influencing factors of restoration of spontaneous circulation (ROSC) in patients after cardiopulmonary resuscitation (CPR) and establish a predictive model, and explore its clinical value.Methods:A retrospective case control study was conducted. The clinical data of cardiac arrest patients admitted to the emergency department of Tangdu Hospital of Air Force Military Medical University and received CPR from January to July 2023 were analyzed, including general information, blood biochemical indicators, main cause of cardiac arrest, whether it was defibrillation rhythm, duration from admission to CPR, and whether ROSC was achieved. The clinical data between the patients whether achieved ROSC or not were compared. The binary multivariate Logistic regression analysis was used to screen the independent influencing factors of ROSC in in-hospital CPR patients. According to the above influencing factors, the ROSC prediction model was established, and the receiver operator characteristic curve (ROC curve) was drawn to evaluate the predictive value of the model for ROSC.Results:A total of 235 patients who received CPR in the emergency department were enrolled, including 153 cases (65.11%) of in-hospital CPR and 82 cases (34.89%) of out-of-hospital CPR. The ROSC ratio was 30.21% (71/235). Among all patients, the majority were aged 61-80 years [40.43% (95/235)], and cardiogenic disease was the main cause of cardiac arrest [32.77% (77/235)]. Among 153 patients with in-hospital CPR, 89 were non-ROSC and 64 were ROSC with ROSC rate of 41.83%. Compared with the non-ROSC group, the patients in the ROSC group had lower blood lactic acid (Lac), N-terminal pro-brain natriuretic peptide (NT-proBNP), Lac/albumin (Alb) ratio (LAR), and ratio of non-defibrillation rhythm [Lac (mmol/L): 5.50 (2.33, 9.65) vs. 7.10 (3.50, 13.35), NT-proBNP (μg/L): 0.87 (0.20, 8.68) vs. 3.00 (0.58, 20.17), LAR: 0.14 (0.07, 0.29) vs. 0.19 (0.10, 0.43), non-defibrillation rhythm ratio: 68.75% (44/64) vs. 93.26% (83/89)], higher actual base excess (ABE) and Alb [ABE (mmol/L): -3.95 (-12.75, 0.23) vs. -7.50 (-13.50, -3.35), Alb (g/L): 38.13±7.03 vs. 34.09±7.81], and shorter duration from admission to CPR [hours: 3.25 (1.00, 14.00) vs. 8.00 (2.00, 27.50)], the differences were statistically significant (all P < 0.05). Binary multivariate Logistic regression analysis showed that LAR [odds ratio ( OR) = 0.037, 95% confidence interval (95% CI) was 0.005-0.287], non-defibrillation rhythm ( OR = 0.145, 95% CI was 0.049-0.426), and duration from admission to CPR ( OR = 0.984, 95% CI was 0.972-0.997) were independent influencing factors for ROSC in hospitalized CPR patients (all P < 0.05). Based on the above influencing factors, a ROSC prediction model was constructed through regression analysis results. The ROC curve analysis showed that the area under the ROC curve (AUC) for predicting ROSC in in-hospital CPR patients was 0.757 (95% CI was 0.680-0.834), Yoden index was 0.429, sensitivity was 76.6%, and specificity was 66.3%. Conclusions:LAR, non-defibrillation rhythm and duration from admission to CPR were independent influencing factors for ROSC in patients with in-hospital CPR. The ROSC prediction model established based on the above influencing factors has a good predictive value for ROSC of CPR patients in hospital, and can guide clinicians to evaluate the prognosis of patients through relevant indicators as early as possible.