Severe patient organ failure assessment and risk stratification method based on occurrence time point of blood flow infection

Abstract

The invention discloses a severe patient organ failure assessment and risk stratification method based on the occurrence time point of blood flow infection, and relates to the field of severe medicine prognosis risk assessment modeling. Comprising the following steps: firstly, collecting daily sequential organ failure evaluation scores of a blood flow infection patient during an infection occurrence period, and dividing data into a training set and a verification set; importing the training set into a group trajectory model for fitting, determining the optimal grouping number of trajectory groups, and verifying the model through a verification set; then, according to a verification result, the patients are allocated to the track group with the highest membership probability for research; finally, a Cox proportional risk regression model is adopted, the correlation between different track groups and the death rate is evaluated, and decisive factors influencing prognosis are determined. According to the method, a robust framework is provided for hierarchical management of ICU blood flow infection patients, the death rate of the patients can be independently predicted, reference is provided for early intervention and formulation of personalized management strategies, and the clinical outcome of the patients is finally improved.

Description

Technical Field

[0001] The present invention relates to the field of critical care medicine prognosis risk assessment modeling, and in particular to a method for organ failure assessment and risk stratification in critically ill patients based on the time of bloodstream infection. Background Art

[0002] Bloodstream infection (BSI) is a common and high-risk infection in the intensive care unit (ICU). It often leads to multiple organ dysfunction syndrome (MODS), significantly increasing patients' hospitalization time, treatment costs, and mortality risk. In this clinical context, how to conduct early and accurate prognostic assessment of BSI patients has become a key issue in ICU management.

[0003] Currently, the Sequential Organ Failure Assessment (SOFA) score is widely used clinically as a standard tool to measure the degree of organ dysfunction. Studies have shown that the dynamic trajectory of the SOFA score better reflects the disease progression than static scores, leading to the development of prognostic modeling methods based on trajectory analysis.

[0004] For example, some studies have attempted to use Group-Based Trajectory Modeling (GBTM) to analyze the SOFA scores of ICU patients within 72 hours of admission, and to predict the patients' clinical outcomes based on this. This type of method has improved the ability to identify disease heterogeneity to a certain extent, but it also has key technical limitations. Specifically, setting the starting point of data collection as "ICU admission time" is a management anchor point rather than a pathophysiological event that reflects the essential progression of the disease. This anchor point selection may cause the model to misjudge the true inflammatory process (such as immune imbalance and organ failure peaks), thereby affecting the model's accuracy in predicting the risk of death.

[0005] Therefore, this field urgently needs to solve a technical problem that has not been fully addressed: how to use the actual occurrence time of bloodstream infection as an anchor to construct a data collection time window that is more closely aligned with the pathophysiological process, thereby improving the trajectory modeling method's ability to predict the prognosis of ICU patients. Summary of the Invention

[0006] Existing prognostic assessment methods for critically ill patients with bloodstream infections (BSIs) generally use the time of ICU admission as the starting point for trajectory modeling. This is a managerial anchoring approach that fails to accurately reflect the peak of the inflammatory response during pathophysiological processes. This can lead to delayed identification of changes in organ function, thus limiting the accuracy and clinical utility of risk stratification. Therefore, a trajectory modeling strategy anchored by the time of BSI onset is urgently needed to better align with the rhythm of disease progression and improve the predictive performance of dynamic organ failure assessment.

[0007] To solve the above technical problems, the present invention provides a method for assessing and stratifying the risk of organ failure in critically ill patients based on the time of bloodstream infection, comprising the following steps:

[0008] S1. Data Collection: Patient data for the training and validation sets were collected, including demographic characteristics, clinical data, biological indicators, and clinical outcomes. Patients without significant influencing factors were excluded, and cases with first-time bloodstream infection were screened to form the training and validation sets respectively.

[0009] S2. Sequential Organ Failure Assessment (SOFA) score collection: The patients were assessed for organ failure using the Sequential Organ Failure Assessment (SOFA) score, and daily SOFA scores were collected from 1 day before to 3 days after the bloodstream infection.

[0010] S3. Group trajectory model establishment: The training set data is imported into the group trajectory model for fitting, the optimal number of groupings of trajectory groups is determined, and the accuracy of the model is verified by the validation set. The patients are assigned to the trajectory group with the highest probability of membership for subsequent analysis;

[0011] S4. Statistical analysis: The Cox proportional hazards regression model was used to evaluate the association between different trajectory groups and in-hospital mortality. Statistically significant and clinically meaningful covariates were selected for univariate and multivariate analyses. Significant variables in the multivariate model were adjusted to obtain a highly predictive Cox regression model. Survival curves were drawn, and the 28-day and 90-day mortality rates of the groups were compared. Key factors influencing prognosis were evaluated.

[0012] Furthermore, the exclusion criteria for patients in step S1 included: (a) age <18 years; (b) hospitalization time <24 hours; (c) survival time <24 hours; (d) incomplete or missing data; and (e) non-pathogenic bloodstream infection.

[0013] Furthermore, in step S1:

[0014] The demographic data included age, sex, race, and body mass index;

[0015] The clinical characteristics include comorbidities, potential risk factors before bloodstream infection, whether the infection is hospital-acquired, and treatment after infection;

[0016] The biological indicators include vital signs, SOFA score and laboratory test results on the day of bloodstream infection, and the laboratory test results include blood routine, biochemical indicators, coagulation function, arterial blood gas results, C-reactive protein and procalcitonin;

[0017] The primary clinical outcome measures were in-hospital mortality and 28-day mortality, and the secondary outcomes were 90-day mortality and survival time.

[0018] Furthermore, the occurrence of hospital-acquired bloodstream infection was defined as the patient's first positive blood culture sample was obtained more than 48 hours after admission, and there was no clinical evidence of infection at admission.

[0019] Furthermore, in step S3, the criteria for determining the best trajectory group include:

[0020] (1) Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC): The smaller the absolute values ​​of BIC and AIC, the better the model fitting effect and the lower the model complexity;

[0021] (2) Average posterior probability: its value is greater than or equal to 0.7, indicating that the model is reliable;

[0022] (3) Entropy: The closer its value is to 1, the clearer the division of groups is and the better the separation effect is;

[0023] (4) Group distribution: If the proportion of any subgroup in the whole is less than 5%, the model is excluded;

[0024] (5) Correct classification rate: The probability of correct classification should be greater than 5.0.

[0025] Furthermore, in step S4, the log-rank test was used to evaluate the differences between the groups.

[0026] Furthermore, after patients were assigned to prognostic risk trajectory groups, intensive intervention treatment plans were administered to patients assigned to high-risk trajectory groups.

[0027] Furthermore, when the method is validated on an independent validation set, it can effectively perform risk stratification for patients.

[0028] The principle and beneficial effects of this technical solution:

[0029] 1. Compared with the existing technology that generally uses ICU admission time as the modeling starting point, this invention proposes to use the actual occurrence time of bloodstream infection as the time anchor point to establish a SOFA score data window that is more in line with the pathophysiological process, thereby more accurately capturing the dynamic changes in organ function as the infection progresses.

[0030] 2. Through group trajectory modeling, this method can identify patient subgroups with obvious clinical heterogeneity, achieve better risk stratification, and obtain consistent trajectory distribution results on two independent datasets, verifying its stability and applicability in different medical settings.

[0031] 3. Furthermore, this study reveals the dominant role of organ-specific indicators (such as cardiovascular scores) in trajectory grouping, providing a basis for the identification of subsequent intervention targets. This method has strong operability and clinical translational potential, and can assist ICU physicians in providing personalized early warning and precision treatment, thereby improving the survival rate and overall outcome of patients with severe bloodstream infections. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Flowchart of the admission and exclusion of the MIMIC cohort (a) and the SAHZU cohort (b);

[0033] Figure 2 Schematic diagram of the SOFA total score trajectory of each group in the MIMIC cohort (a) and SAHZU cohort (b);

[0034] Figure 3 Schematic diagram of the SOFA subscore trajectory for each group in the MIMIC cohort (a) and the SAHZU cohort (b);

[0035] Figure 4 Schematic diagram of the 28-day mortality survival curves of the MIMIC cohort (a) and the SAHZU cohort (b);

[0036] Figure 5 Schematic diagram of the 90-day mortality survival curves of the MIMIC cohort (a) and the SAHZU cohort (b). DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments and the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0038] It should also be noted that, in order to avoid obscuring the present invention due to unnecessary details, the accompanying drawings only show structures and / or processing steps closely related to the solutions according to the present invention, while other details that are not closely related to the present invention are omitted.

[0039] It should be emphasized that the term "include / comprises" when used herein refers to the existence of features, elements, steps or components, but does not exclude the existence or addition of one or more other features, elements, steps or components.

[0040] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the accompanying drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0041] It should be emphasized here that the step marks mentioned below do not limit the order of the steps, but it should be understood that the steps can be executed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be executed simultaneously.

[0042] Example 1

[0043] The following uses Staphylococcus aureus bloodstream infection (SA-BSI) as an example to illustrate a method for assessing and stratifying organ failure in critically ill patients based on the time of bloodstream infection. The method includes the following steps:

[0044] S1. Data Collection: Patient data for the training and validation sets were collected, covering demographic characteristics, clinical data, biological indicators, and clinical outcomes. Patients without significant influencing factors were excluded, and cases with first-time bloodstream infection were screened to form the training and validation sets, respectively.

[0045] Clinical characteristics include comorbidities, potential risk factors before bloodstream infection, whether the infection was hospital-acquired, and post-infection treatment;

[0046] Biological indicators include vital signs, SOFA score, and laboratory test results on the day of bloodstream infection, including blood routine, biochemical indicators, coagulation function, arterial blood gas results, C-reactive protein, and procalcitonin;

[0047] The primary clinical outcome measures were in-hospital mortality and 28-day mortality, and the secondary outcomes were 90-day mortality and survival time.

[0048] In this case, the MIMIC-IV database, a large public anonymous dataset from the intensive care unit (ICU) of Beth Israel Deaconess Medical Center (BIDMC) in the United States, was used as the training set, denoted as the MIMIC cohort; clinical cases from the ICU of the Second Affiliated Hospital of Zhejiang University School of Medicine were used as the validation set, denoted as the SAHZU cohort.

[0049] First, individuals without significant influencing factors were excluded to identify first-time SA-BSI cases. Exclusion criteria included: (a) age <18 years; (b) hospitalization time <24 hours; (c) survival time <24 hours; (d) incomplete or missing data; and (e) nonpathogenic S. aureus infection. If S. aureus was cultured in only one of the culture sets and the culture time exceeded 48 hours, the infection was classified as nonpathogenic S. aureus.

[0050] refer to Figure 1 , 834 ICU patients from the MIMIC cohort and 151 patients from the SAHZU cohort were included in the final analysis.

[0051] In the MIMIC cohort, a total of 2171 patients with SA-BSI were identified. Of the 1260 non-ICU cases, 25 with a hospital stay of <24 hours, 3 with a survival time of <24 hours, and 49 with incomplete data, 834 patients were ultimately included. Due to a lack of relevant information in the MIMIC-IV database, the number of cases with nonpathogenic S. aureus infection could not be determined.

[0052] In the SAHZU cohort, a total of 641 patients with SA-BSI were identified, 6 cases aged <18 years, 434 non-ICU cases, 5 cases with hospital stay <24 hours, 6 cases with survival time <24 hours, 22 cases with incomplete data, and 17 cases with nonpathogenic S. aureus infection were excluded, and 151 cases were finally included.

[0053] S2. Collection of SOFA scores: Sequential Organ Failure Assessment (SOFA) was used to assess organ failure in patients, and daily SOFA scores were collected from 1 day before to 3 days after the bloodstream infection.

[0054] The severity of organ failure was defined using the SOFA score. SA-BSI was defined according to the Centers for Disease Control and Prevention (CDC) criteria. SA-BSI onset (day 0) was the time the first culture-positive blood sample was collected. Hospital-acquired SA-BSI was defined as a patient whose first positive blood culture was obtained more than 48 hours after admission, with no clinical evidence of infection at the time of admission. The period from day -1 to day +3 was set as a dynamic observation window, with daily SOFA scores recorded. This time window was based on the following clinical logic: the inflammatory response typically peaks within a few days before and after the onset of infection; blood culture reporting is delayed an average of 3–5 days, long after the actual infection has occurred; therefore, this method achieves more accurate capture of the disease process by anchoring pathological events.

[0055] Because some patients missed individual examination items during the 5-day period, not all patients had complete SOFA total and sub-item score data for the 5 days. However, this method uses the group trajectory model (GBTM) to supplement missing data through maximum likelihood estimation, which is robust to missing data, so no additional interpolation was performed.

[0056] S3. Group trajectory model establishment: The training set data is imported into the group trajectory model for fitting, the optimal number of groupings of trajectory groups is determined, and the accuracy of the model is verified by the validation set. The patients are assigned to the trajectory group with the highest probability of membership for subsequent analysis;

[0057] For the MIMIC-IV cohort, the trajectory of the continuous SOFA score over 5 days was modeled. The model was fitted with a polynomial function, and the best trajectory trend (linear, quadratic, or cubic) was determined by likelihood ratio testing. Models with 1–5 potential groups and different polynomial shapes were tested to obtain the best-fitting model. Criteria for determining the best model included:

[0058] (1) Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC): The smaller the absolute values ​​of BIC and AIC, the better the model fitting effect and the lower the model complexity;

[0059] (2) Average posterior probability (AvePP): its value is greater than or equal to 0.7, indicating that the model is reliable;

[0060] (3) Entropy: The closer its value is to 1, the clearer the division of groups is and the better the separation effect is;

[0061] (4) Group distribution: If the proportion of any subgroup in the whole is less than 5%, the model is excluded;

[0062] (5) OCC: The probability of correct classification should be greater than 5.0.

[0063] S4. Statistical Analysis: The association between different trajectory groups and in-hospital mortality was assessed using a Cox proportional hazards regression model. Statistically significant and clinically relevant covariates were selected for univariate and multivariate analyses. Significant variables in the multivariate model were adjusted to generate a robust Cox regression model. Survival curves were constructed to compare 28-day and 90-day mortality rates among the groups, and key prognostic factors were evaluated.

[0064] The trajectory group to which the patient belonged was used as the primary exposure factor, and covariates were selected based on statistical significance or clinical relevance.

[0065] Covariates included demographics (age, sex), comorbidities (hypertension, diabetes, chronic lung disease, circulatory system disease, cerebrovascular disease, kidney disease, liver disease, and malignancy), potential risk factors (invasive ventilation, invasive procedures, renal replacement therapy (RRT), and central venous catheter (CVC) / peripherally inserted central catheter (PICC) use), infection-related factors (whether hospital-acquired infection), and baseline SOFA score. Variables with a p-value < 0.05 in univariate Cox regression were included in the multivariate analysis; in the multivariate analysis, a final Cox proportional hazards regression model was constructed by adjusting for factors with a p-value < 0.05. The final model was used to estimate 28-day and 90-day mortality in each group and was described by Kaplan-Meier survival curve analysis. Differences were tested using the log-rank test.

[0066] All data processing and analysis tools include:

[0067] Database management: Navicat Premium Lite 17 (PremiumSoft Corporation, Hong Kong, China) and Microsoft Excel (Microsoft Corporation, Redmond, Washington, USA);

[0068] Statistical analysis and graphics: R software (version 4.4.2, Lucent Technologies, USA), Zstats v1.0 (www.zstats.net), SPSS26.0 (IBM, Armonk, New York, USA) and Stata / MP 17 (Stata Corp, College Station, Texas);

[0069] For normally distributed data, continuous variables were expressed as mean ± standard deviation (SD); for non-normally distributed data, continuous variables were expressed as median and interquartile range (IQR), and categorical variables were expressed as percentage (%). Continuous variables were compared using Student's t-test or Mann-Whitney U test, while Pearson chi-square test was used. 2 Categorical variables were compared using the 2-sided test or Fisher's exact test. A p-value < 0.05 was considered statistically significant.

[0070] result

[0071] 1. GBTM Modeling Results

[0072] Table 1 shows the fitting statistics of the GBTM model for different numbers of trajectory groups. To determine the optimal trajectory model, polynomial functions of varying orders were used to model and compare 1 to 5 trajectory groups. The results showed that all models had AvePP greater than 0.7 and OCC greater than 5.0, indicating good classification stability.

[0073] Based on the modeling of the MIMIC cohort, the SAHZU cohort was introduced as an out-of-sample (OOS) validation set. The most likely trajectory group assignments and corresponding posterior probabilities of SAHZU patients were inferred based on the model fitted in the MIMIC cohort. The evaluation results of the posterior probability showed that the classification accuracy of the SAHZU cohort was also high, with the maximum posterior probability (maxPP) exceeding 0.7 for 139 of the 151 patients, and the maxPP of all patients was higher than 0.5. Ultimately, the AvePP of groups 1 to 3 in the SAHZU cohort were 0.93, 0.90, and 0.94, respectively, further confirming the good adaptability and reliability of the model in two independent datasets.

[0074] Furthermore, as the number of trajectory groups increased, the absolute values ​​of the model's AIC and BIC gradually decreased, while the entropy also showed a downward trend, suggesting improved model fit, but blurred classification boundaries and reduced accuracy. Notably, the five-group model contained a subgroup with a proportion of less than 5%, suggesting a risk of overfitting. Considering both statistical robustness and clinical interpretability, the three-group model was ultimately selected as the optimal trajectory model.

[0075] Table 1 Fitting statistics of different number of trajectory groups

[0076]

[0077] In the table, Loglik: log likelihood value; AIC: Akaike information criterion; BIC: Bayesian information criterion; AvePP: average posterior probability of each group; OCC: correct classification rate.

[0078] refer to Figure 2 and Figure 3 , showing three subgroups of patients with significantly different SOFA score dynamic trajectories divided by the method of the present invention, verifying the effectiveness of this method in identifying clinical heterogeneity. Figure 3 a shows the SOFA subscore trajectory of the MIMIC cohort. In the third trajectory group, the six organ scores increased significantly, especially the cardiovascular and renal scores, which were at high levels from the beginning, suggesting that they played a leading role in the increase of SOFA scores; the respiratory score increased rapidly over time and was an important driving factor for the increase in the total SOFA score. In the SAHZU cohort ( Figure 3In group b), the cardiovascular subscore was also the main factor for higher SOFA scores, but the respiratory score played a more significant role than the renal score in group 3. Overall, all SOFA subscores in this group showed an increasing trend, reflecting severe and progressive organ dysfunction, which is consistent with the trend in the MIMIC cohort.

[0079] 2. Baseline Characteristics

[0080] As the trajectory group increased from group 1 to group 3, multiple organ failure (OD)-related biomarkers showed a gradual upward trend. Taking the MIMIC cohort as an example, the levels of total bilirubin (TBil), blood urea nitrogen (BUN), and serum creatinine (SCr) increased from 11.97 (7.27, 18.81) μmol / L, 21.00 (14.50, 34.50) mg / dL, and 88.40 (61.88, 132.60) μmol / L in group 1 to 32.49 (17.10, 69.68) μmol / L, 58.00 (41.00, 72.75) mg / dL, and 274.04 (167.96, 499.46) μmol / L in group 3, respectively.

[0081] As shown in Table 2, inflammatory markers, including C-reactive protein (CRP) and procalcitonin (PCT), key indicators of systemic inflammatory response, also showed an increasing trend across all trajectory groups. For example, the median CRP level increased from 82.55 (44.12, 139.75) mg / L in Group 1 to 200.80 (73.55, 252.90) mg / L in Group 3; the median PCT level increased from 0.28 (0.13, 1.12) ng / mL in Group 1 to 7.47 (0.97, 14.75) ng / mL in Group 3, suggesting a more pronounced systemic inflammatory response in patients in the high-trajectory group. Furthermore, D-dimer, a marker of coagulation dysfunction, was also significantly elevated in the high-trajectory group. The median D-dimer level of patients in group 3 was 6170.00 (3440.00, 8220.00) μg / L, which was significantly higher than that in other groups, suggesting that patients in this group had more severe coagulation abnormalities.

[0082] Table 2 Other indicators of the SAHZU cohort

[0083]

[0084]

[0085] In the table: IQR: interquartile range; ECMO: extracorporeal membrane oxygenation; BSI: bloodstream infection; SD: standard deviation; P / F oxygenation index; CRP: C-reactive protein; PCT: procalcitonin; APACHE: Acute Physiology and Chronic Health Evaluation.

[0086] 3. Conclusion

[0087] Refer to Tables 3-1 and 3-2, which show the Cox univariate and multivariate regression results of in-hospital mortality in the MIMIC and SAHZU cohorts, respectively.

[0088] In multivariable Cox regression analysis of the MIMIC cohort, patients in trajectory group 2 had a significantly increased risk of in-hospital mortality compared with those in trajectory group 1 (HR = 1.43, 95% CI: 1.04-1.97, p = 0.029), and the risk in trajectory group 3 was even more significant (HR = 4.49, 95% CI: 2.56-7.90, p < 0.001), demonstrating a gradient of increasing mortality risk across trajectory group. In the SAHZU cohort, trajectory group was also an independent and significant predictor of in-hospital mortality. Patients in trajectory group 3 had a significantly higher risk of death than those in trajectory group 1 (HR = 4.38, 95% CI: 1.22-15.71, p = 0.023). Therefore, trajectory group was significantly associated with in-hospital mortality.

[0089] Table 3-1 Cox univariate and multivariate regression analysis of in-hospital mortality in the MIMIC cohort.

[0090]

[0091]

[0092] Table 3-2 Cox univariate and multivariate regression analysis of in-hospital mortality in the SAHZU cohort.

[0093]

[0094]

[0095] In the table, HR: hazard ratio; CI: confidence interval; RRT: renal replacement therapy; CVC: central venous catheter; PICC: peripherally inserted central catheter; MRSA: methicillin-resistant Staphylococcus aureus.

[0096] Furthermore, after adjusting for significant covariates such as age, circulatory system diseases, cerebrovascular diseases, malignancies, and CVC / PICC use (see Table 4), the trajectory groups remained the strongest independent predictors of in-hospital mortality in both datasets: in the MIMIC cohort, the hazard ratios for Group 2 (HR = 1.56, 95% CI: 1.27 - 1.91, p < 0.001) and Group 3 (HR = 4.60, 95% CI: 3.49 - 6.07, p < 0.001) continued to increase significantly; in the SAHZU cohort, the mortality risk of Group 3 remained significant after adjustment (HR = 5.36, 95% CI: 2.87 - 10.00, p < 0.001), verifying its broad applicability and robustness.

[0097] Table 4 Unadjusted and adjusted Cox regression for the MIMIC and SAHZU cohorts

[0098]

[0099]

[0100] ****: 0 < p-value < 0.0001; ***: 0.0001 < p-value < 0.001; **: 0.001 < p-value < 0.01.

[0101] Kaplan-Meier survival curves (refer to Figure 4 and Figure 5 ) further support the above conclusion. Figure 4 showed a strong correlation between the trajectory groups divided by the method of the present invention and the 28-day mortality rate. The survival rate of the high-risk group was significantly lower than that of the low-risk group, further confirming the prognostic prediction ability of the method of the present invention. There were significant differences in the 28-day and 90-day survival rates among different trajectory groups in both cohorts. Overall, the survival rate of Group 1 was the highest and that of Group 3 was the lowest, and the trajectory grouping had good survival prediction ability. It should be noted that in the pairwise comparison, there was a statistically significant difference between Group 1 and Group 2 in the MIMIC cohort; while this difference did not reach the significant level in the SAHZU cohort (see Table 5), which may be related to factors such as sample size and case heterogeneity.

[0102] Table 5 Pairwise comparison between groups in the training set and validation set using the log-rank test

[0103]

[0104] ****: 0 < p-value < 0.0001; ***: 0.0001 < p-value < 0.001; **: 0.001 < p-value < 0.01.

[0105] The group trajectory model presented in this study provides a robust framework for risk stratification in ICU patients with Staphylococcus aureus bloodstream infection (SA-BSI). By categorizing these groups into trajectory groups, this approach can independently predict mortality and provide a scientific basis for the development of early intervention and personalized management strategies, thereby effectively improving patient outcomes.

[0106] This study revealed the SOFA score progression pattern, organ failure (OD) trajectory and its relationship with prognosis in SA-BSI ICU patients using three trajectory groups identified by group trajectory model (GBTM). In addition to the significant heterogeneity in clinical and biological characteristics among the trajectory groups, the high-risk trajectory group was highly associated with adverse clinical outcomes, and cardiovascular failure assessment played a key role in predicting the risk of patient mortality. In addition, the combination of longitudinal monitoring of inflammatory and coagulation biomarkers further emphasized the potential application value of dynamic biomarkers in improving the accuracy of risk stratification. It is worth noting that even after adjusting for clinical covariates such as age and underlying diseases, trajectory group classification remained a strong independent predictor of in-hospital mortality. This result provides important support for the real-time identification of high-risk patients and the development of personalized ICU care strategies.

[0107] Obviously, the model of the present invention is also applicable to other types of ICU bloodstream infection patients, including bacterial, fungal and viral bloodstream infection patients, and provides a scientific basis and practical guidance for their organ failure assessment and early intervention.

[0108] It should be understood that the present invention is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted. In the above embodiments, several specific steps are described and illustrated as examples. However, the method of the present invention is not limited to the specific steps described and illustrated. Those skilled in the art may make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present invention.

[0109] In the present invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or replace features of other embodiments.

[0110] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations to the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for assessing and stratifying organ failure in critically ill patients based on the time of bloodstream infection, characterized in that: The steps include: S1. Data Collection: Patient data for the training and validation sets were collected, including demographic characteristics, clinical data, biological indicators, and clinical outcomes. Patients without significant influencing factors were excluded, and cases with first-time bloodstream infection were screened to form the training and validation sets respectively. S2. Sequential Organ Failure Assessment (SOFA) score collection: The patients were assessed for organ failure using the Sequential Organ Failure Assessment (SOFA) score, and daily SOFA scores were collected from 1 day before to 3 days after the bloodstream infection. S3. Group trajectory model establishment: The training set data is imported into the group trajectory model for fitting, the optimal number of groupings of trajectory groups is determined, and the accuracy of the model is verified by the validation set. The patients are assigned to the trajectory group with the highest probability of membership for subsequent analysis; S4. Statistical analysis: The Cox proportional hazards regression model was used to evaluate the association between different trajectory groups and in-hospital mortality. Statistically significant and clinically meaningful covariates were selected for univariate and multivariate analyses. Significant variables in the multivariate model were adjusted to obtain a highly predictive Cox regression model. Survival curves were drawn, and the 28-day and 90-day mortality rates of the groups were compared. Key factors influencing prognosis were evaluated.

2. A method for evaluating and risk stratifying organ failure in critically ill patients based on the time of bloodstream infection according to claim 1, characterized in that: Exclusion criteria for patients in step S1 included: (a) age <18 years; (b) hospitalization time <24 hours; (c) survival time <24 hours; (d) incomplete or missing data; and (e) non-pathogenic bloodstream infection.

3. The method for evaluating organ failure and risk stratification in critically ill patients based on the time of bloodstream infection according to claim 1, characterized in that: In step S1: The demographic data included age, sex, race, and body mass index; The clinical characteristics include comorbidities, potential risk factors before bloodstream infection, whether the infection is hospital-acquired, and treatment after infection; The biological indicators include vital signs, SOFA score and laboratory test results on the day of bloodstream infection, and the laboratory test results include blood routine, biochemical indicators, coagulation function, arterial blood gas results, C-reactive protein and procalcitonin; The primary clinical outcome measures were in-hospital mortality and 28-day mortality, and the secondary outcomes were 90-day mortality and survival time.

4. The method for evaluating organ failure and risk stratification in critically ill patients based on the time of bloodstream infection according to claim 1, characterized in that: Hospital-acquired bloodstream infection was defined as the first positive blood culture obtained more than 48 hours after admission, without clinical evidence of infection at the time of admission.

5. The method for evaluating organ failure and risk stratification in critically ill patients based on the time of bloodstream infection according to claim 1, characterized in that: In step S3, the criteria for determining the best trajectory group include: (1) Bayesian Information Criterion (BIC) and Akaike Information Criterion (AIC): The smaller the absolute values ​​of BIC and AIC, the better the model fitting effect and the lower the model complexity; (2) Average posterior probability: its value is greater than or equal to 0.7, indicating that the model is reliable; (3) Entropy: The closer its value is to 1, the clearer the division of groups is and the better the separation effect is; (4) Group distribution: If the proportion of any subgroup in the whole is less than 5%, the model is excluded; (5) Correct classification rate: The probability of correct classification should be greater than 5.

0.

6. The method for evaluating organ failure and risk stratification in critically ill patients based on the time of bloodstream infection according to claim 1, characterized in that: In step S4, the log-rank test was used to evaluate the differences between the groups.

7. The method for evaluating organ failure and risk stratification in critically ill patients based on the time of bloodstream infection according to claim 1, characterized in that: After patients were assigned to prognostic risk trajectory groups, intensive intervention treatment plans were implemented for patients assigned to high-risk trajectory groups.

8. The method for evaluating organ failure and risk stratification in critically ill patients based on the time of bloodstream infection according to claim 1, characterized in that: When validated on an independent validation set, the method was able to effectively risk stratify patients.

Images