A risk assessment system for cross matching difficulties

Abstract

The application discloses a risk assessment system for cross matching difficulty, and relates to the technical field of clinical medical examination and blood transfusion management. The system comprises a data input module, a risk calculation module, a risk stratification module and a result output and suggestion module. The data input module obtains key information of a patient, including irregular red blood cell antibody screening results, multiple myeloma diagnosis, hematology department medical history, age of 60 years or older, pregnancy-related disease state and blood type identification difficulty; the risk calculation module adopts a preset scoring algorithm to assign weights to each factor and calculate a total risk score; the risk stratification module divides the patient into three grades of low risk, medium risk and high risk; and the result output module provides a differentiated clinical management path corresponding to the risk grade. The application realizes early warning of cross matching difficulty through quantitative assessment, can optimize the efficiency of the blood transfusion process and reduce clinical risk, has good prediction performance (AUC=0.819) and a high negative predictive value (93.7%).

Description

Technical Field

[0001] This invention relates to the field of clinical medical testing and transfusion management technology based on computer models, and in particular to a risk assessment system for crossmatching difficulties, applicable to patients with irregular red blood cell antibody positivity, enabling quantitative assessment of crossmatching difficulties risk, automatic stratification, and early warning during the transfusion application stage. Background Technology

[0002] Blood transfusion is an indispensable life support measure in modern medicine. Before transfusion, crossmatching is a crucial step to ensure the compatibility of donor and recipient blood. When a patient has irregular antibodies against red blood cells, finding compatible blood products becomes complex and time-consuming, a condition known as "crossmatching difficulties." This can lead to transfusion delays, increase treatment risks for patients (especially those with acute and critical illnesses), and may even trigger hemolytic transfusion reactions.

[0003] Currently, the management of patients with positive irregular red blood cell antibodies mainly relies on a passive response model, which involves first screening and identifying irregular red blood cell antibodies, and then attempting to find compatible blood. Existing research largely focuses on describing the antibody positivity rate, specific distribution, and its association with single factors such as transfusion history and pregnancy history. For example, it is known that patients with multiple myeloma or those who have received multiple transfusions are more prone to developing complex antibodies. However, clinical practice lacks a comprehensive, quantitative risk assessment tool that integrates key patient information (such as diagnosis, department, age, antibody response intensity, and difficulty of blood typing). Physicians and transfusion staff often rely on personal experience to make judgments, lacking objective risk stratification standards. This can lead to delayed transfusions for high-risk patients due to insufficient early warning, while low-risk patients may consume excessive medical resources.

[0004] The drawbacks of existing technology are:

[0005] (1) The management model is passive and lagging: the existing process only begins the complex manual or special blood matching work after a problem is discovered (such as a positive screening for irregular antibodies in red blood cells), and cannot make forward-looking predictions and early warnings of the risk of "difficulty in cross-matching" at the blood transfusion application stage.

[0006] (2) The assessment method is subjective and non-quantitative: clinical judgment mainly relies on personal experience and lacks objective and unified risk assessment standards. The judgment results of different physicians vary greatly, making it difficult to achieve standardized management.

[0007] (3) Insufficient information integration: The independent risk factors of multiple dimensions, such as the patient's clinical diagnosis (e.g., multiple myeloma), department affiliation (e.g., hematology), demographic characteristics (e.g., age), and laboratory indicators (e.g., antibody response intensity, difficulty of blood type identification), were not systematically integrated and quantitatively analyzed.

[0008] (4) Lack of hierarchical management tools: Existing technologies do not provide a simple and easy-to-use clinical risk scoring tool to transform complex statistical models into intuitive scores and risk levels, thereby guiding differentiated blood preparation and resource allocation schemes.

[0009] Chinese Invention No. CN114783607A discloses a method for constructing a surgical transfusion risk prediction model and its online calculator. The method utilizes a univariate logistic regression model to calculate the correlation between various clinical indicators and transfusion risk, selects clinical indicators related to transfusion risk for multivariate logistic regression calculation, and identifies independent predictive factors to construct the transfusion prediction model. An online calculator is generated based on the formula for the weight coefficient scores of each predictive factor in the prediction model. A dataset of key information about surgical patients is collected and input into the online calculator to obtain the predicted transfusion probability. The transfusion prediction model constructed by this invention can predict the probability of perioperative transfusion for surgical patients before surgery. The online calculator generated based on this model can be applied to electronic medical records, providing automated electronic decision-making for preoperative planning and blood management. However, it only predicts whether a transfusion is needed and does not address the difficulties of cross-matching.

[0010] Chinese Invention No. CN113611401A discloses a system and method for perioperative blood management, including an information acquisition module, an information processing module, an artificial intelligence calculation module, and a clinical support result display module. The information acquisition module acquires patients' medical data in real time; the information processing module extracts key features for blood use assessment based on the patients' medical data and converts them into features recognizable by an artificial intelligence engine through data cleaning; the artificial intelligence calculation module constructs a classification model based on the features recognizable by the artificial intelligence engine to achieve preoperative anemia assessment, perioperative transfusion prediction, and postoperative adverse transfusion event risk prediction, obtaining blood use assessment results; the clinical support result display module integrates the blood use assessment results and provides early warnings and recommendations based on them. This achieves multi-stage coverage and dynamic monitoring of perioperative blood management, minimizing transfusion risks. However, it focuses on the blood use process and data acquisition, lacking a dedicated scoring model and weighting design for cross-matching difficulties. Summary of the Invention

[0011] The technical problem to be solved by the present invention is to provide a risk assessment system for crossmatching difficulties, which can integrate multi-dimensional clinical and laboratory information of patients, and realize quantitative assessment, prospective prediction and early warning and automatic stratification of the risk of "crossmatching difficulties" before blood transfusion, so as to overcome the shortcomings of the existing technology.

[0012] This invention provides a risk assessment system for crossmatch difficulties, comprising:

[0013] The data input module is used to obtain key patient information related to crossmatching difficulties. The key patient information includes: red blood cell irregular antibody screening results, whether it is a diagnosis of multiple myeloma, whether it has visited a hematology department, whether the patient is ≥60 years old, whether it is a pregnancy-related disease, and whether blood type identification is difficult.

[0014] The risk calculation module is configured to assign corresponding weights to the patient's key information and calculate the total risk score according to a preset risk scoring algorithm.

[0015] The risk stratification module is configured to classify patients into at least three levels: low risk, medium risk, and high risk, based on the numerical range of their total risk score.

[0016] The Results Output and Recommendations module is used to output the risk level and the corresponding differentiated clinical management pathway.

[0017] Furthermore, the assignment rules used in the risk scoring algorithm are as follows:

[0018] Three positive wells are assigned a score of +2.

[0019] Multiple myeloma diagnosis, assigned a score of +3;

[0020] Hospitalization in the hematology department, +1 point;

[0021] For those aged 60 or older, add 1 point.

[0022] Pregnancy-related diseases are assigned a score of -3.

[0023] Difficult blood type identification will be assigned a +3 point value.

[0024] Total risk score = 2×X1 + 3×X2 + 1×X3 + 1×X4 - 3×X5 + 3×X6, where X1 to X6 are binary variables, with 1 for yes and 0 for no.

[0025] Furthermore, the three levels are: a total risk score of ≤2 points indicates low risk, a total risk score in the range of 3-4 points indicates medium risk, and a total risk score of ≥5 points indicates high risk.

[0026] Furthermore, the key patient information was based on a retrospective clinical cohort study. The study screened candidate variables with a p-value <0.1 through univariate analysis, and then determined six independent predictors through multivariate logistic regression. Among them, the results of irregular red blood cell antibody screening, whether it is a diagnosis of multiple myeloma, and whether blood type identification is difficult are the independent predictors with the strongest risk, while whether it is a pregnancy-related disease state shows a unique protective effect.

[0027] Furthermore, the risk calculation module determines the weights and values ​​of each factor among the six independent predictors based on the results of multivariate logistic regression analysis of a retrospective clinical cohort. Specifically, the relative importance of each factor among the six independent predictors is first determined based on the regression coefficients of the multivariate logistic regression analysis. The regression coefficients reflect the independent influence of the factor on the risk of crossmatching difficulties. The larger the absolute value of the regression coefficient, the higher the weight. Then, the regression coefficients are converted into integer risk integrals proportionally for assignment.

[0028] Furthermore, the results of the irregular antibody screening for red blood cells include two types: one or two positive wells and three positive wells.

[0029] Furthermore, the aforementioned risk levels and corresponding differentiated clinical management pathways are specifically as follows:

[0030] Low-risk level: Standard blood transfusion procedures are followed, and antibody identification and antigen-negative blood screening are performed routinely, without the need for special acceleration or resource allocation;

[0031] Medium risk level: Implement early warning and preparation procedures. Upon receiving the application, the blood transfusion department will immediately initiate expanded screening, prioritize the allocation or reservation of rare blood types that may be needed, and notify clinical departments to prepare for possible delays.

[0032] High-risk level: Implement emergency intervention and multidisciplinary collaboration procedures. The system will automatically trigger the highest level of warning, notify relevant personnel to immediately activate the department's difficult blood matching plan, contact the regional blood center to seek rare blood sources, and assess the risk and benefit ratio of emergency blood transfusion in the case of incomplete blood matching.

[0033] The present invention has the following technical effects:

[0034] (1) Achieving a shift from a passive to a proactive management model: It enables forward-looking prediction and early warning of the risk of crossmatching difficulties at the blood transfusion application stage, changing the previous situation of passively responding after discovering problems, and winning valuable time to ensure timely blood transfusion for critically ill patients. In particular, for the first time in a prospective risk prediction model, the objective and early-obtained pattern indicator of "all 3 wells positive" has been established as an independent quantitative predictor. This marks a shift in risk assessment from relying on the final "antibody identification result" to focusing on the initial "screening response pattern", winning more time for clinical early warning and preparation.

[0035] (2) Provides quantitative and objective assessment criteria: Based on large sample clinical data and rigorous statistical models, it provides a unified and repeatable quantitative scoring standard, which eliminates individual experience differences and makes risk assessment more scientific and accurate.

[0036] (3) Good predictive performance: After internal validation, the area under the curve (AUC) of this risk scoring model reached 0.819 (95% CI: 0.766-0.872), with a sensitivity of 74.4% and a specificity of 76.5%, showing good predictive accuracy and clinical practical value.

[0037] (4) Simple to operate and easy to promote: The complex regression model is simplified into a scoring rule that includes only six easily accessible clinical and laboratory variables. No special software or additional tests are required, and clinicians and transfusion staff can quickly master and apply it.

[0038] (5) Optimize resource allocation and improve transfusion safety: Through clear risk stratification and differentiated management paths, a complete closed loop from "risk identification" to "tool creation" and then to "path formulation" has been completed, guiding blood banks to allocate resources scientifically and efficiently. This avoids excessive intervention for low-risk patients while ensuring that high-risk patients receive the most timely and sufficient attention and resources, ultimately significantly improving the safety and timeliness of transfusions.

[0039] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0040] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0041] Figure 1 This is a structural block diagram of the system of the present invention.

[0042] Figure 2 This is a flowchart illustrating the screening, grouping, and analysis process for patients with irregular red blood cell antibody positivity in the system of this invention.

[0043] Figure 3 This is a schematic diagram of the hierarchical management path based on risk level in the system of the present invention.

[0044] Figure 4 This is a forest plot of the multi-factor logistic regression system of the present invention and a ROC curve of the model's predictive performance.

[0045] Figure 5 The data provided are baseline characteristics of 521 patients with irregular erythrocyte antibody positivity in this embodiment of the invention.

[0046] Figure 6 This is a univariate analysis of factors related to blood matching difficulties in this embodiment of the invention (based on the initial screening mode).

[0047] Figure 7This is a multivariate logistic regression analysis of independent risk factors for blood matching difficulties in this embodiment of the invention.

[0048] Figure 8 This is a table showing the risk stratification and treatment measures for patients with irregular red blood cell antibody positivity in this invention.

[0049] Figure 9 This is a table showing the β coefficient, aOR, and risk score conversion for crossmatching difficulties in this invention.

[0050] Figure 10 This is a table showing the correspondence between risk stratification and treatment measures in embodiments of the present invention. Detailed Implementation

[0051] This application provides a risk assessment system for crossmatching difficulties, which integrates multi-dimensional clinical and laboratory information of patients to quantitatively assess, prospectively predict, provide early warning and automatic stratification of the risk of "crossmatching difficulties" before blood transfusion, thereby overcoming the shortcomings of the prior art.

[0052] The overall approach of the technical solution in this application is as follows: Based on the results of multivariate logistic regression analysis of a retrospective clinical cohort, a risk assessment system for crossmatching difficulties is constructed. This system integrates six independent predictive factors through a data input module: irregular red blood cell antibody screening results (1-2 or 3 positive wells), clinical diagnosis (whether it is multiple myeloma), department visited (whether it is hematology), age (whether it is ≥60 years old), pregnancy-related disease status (protective factor), and difficulty of blood typing (whether it is difficult). The risk calculation module assigns specific weights to each factor based on the logistic regression coefficient and calculates the total risk score using a formula. Then, the risk stratification module classifies the risk level according to a preset threshold. Finally, the result output and suggestion module output the corresponding differentiated clinical management path (low-risk standard process, medium-risk early warning preparation, high-risk emergency intervention) to achieve prospective quantitative assessment, early warning, and stratified management of crossmatching difficulties. Example

[0053] This embodiment provides a risk assessment system for crossmatching difficulties, which is built based on the results of multivariate logistic regression analysis of a retrospective clinical cohort (521 patients with positive erythrocyte irregular antibody screening). The system integrates six independent predictive factors to calculate a risk score and output corresponding risk levels and management recommendations.

[0054] like Figure 1 As shown, the risk assessment system in this embodiment mainly consists of the following four modules:

[0055] The data input module is used to acquire key patient information related to crossmatching difficulties, including:

[0056] Red blood cell irregular antibody screening results (divided into: 1-2 positive well group or 3 positive well group);

[0057] Is it a diagnosis of multiple myeloma?

[0058] Did you visit a hematology department?

[0059] Is the age ≥60 years old?

[0060] Is there a pregnancy-related medical condition (as a protective factor)?

[0061] Is blood type identification difficult?

[0062] The risk calculation module is configured to assign corresponding weights to the patient's key information and calculate a total risk score based on a preset risk scoring algorithm. This risk scoring algorithm assigns weights based on the logistic regression coefficients (β values) of the aforementioned six predictive factors. The specific scoring rules are as follows:

[0063] Three positive wells are assigned a score of +2.

[0064] Multiple myeloma diagnosis, assigned a score of +3;

[0065] Hospitalization in the hematology department, +1 point;

[0066] For those aged 60 or older, add 1 point.

[0067] Pregnancy-related diseases are assigned a score of -3.

[0068] Difficult blood type identification will be assigned a +3 point value.

[0069] Total risk score = 2×X1 + 3×X2 + 1×X3 + 1×X4 - 3×X5 + 3×X6; where X1 to X6 are binary variables, 1 for yes and 0 for no.

[0070] The risk stratification module is configured to classify patients into at least three levels: low risk, medium risk, and high risk, based on the numerical range of their total risk score.

[0071] (1) Low risk level: Total risk score ≤ 2 points.

[0072] (2) Medium risk level: The total risk score is 3-6 points.

[0073] (3) High risk level: Total risk score ≥ 7 points.

[0074] The results output and suggestion module is used to output the risk level and the corresponding differentiated clinical management pathway. This module can be integrated into the hospital's laboratory information system.

[0075] Interconnection between modules: The modules communicate with each other through data interfaces or APIs, forming a closed data flow loop of "data input module → risk calculation module → risk stratification module → result output and suggestion module".

[0076] Key technical features of this invention:

[0077] For the first time, six factors were combined as predictive factors to assess the risk of crossmatching difficulties: "3 positive wells", "diagnosis of multiple myeloma", "hospitalization in hematology department", "age ≥60 years", "pregnancy-related diseases" and "difficult blood typing".

[0078] A simplified clinical risk scoring model was constructed by assigning specific integer scores to each predictor based on logistic regression coefficients.

[0079] Specific risk level thresholds were defined (low: ≤2 points; medium: 3-4 points; high: ≥5 points), which were determined based on the optimal cutoff values ​​of approximately 74.4% sensitivity and 76.5% specificity of the model in the derivation cohort. Risk scores were directly linked to specific, differentiated clinical management pathways.

[0080] like Figure 2 As shown, the operation process of this embodiment of the invention is as follows:

[0081] (1) Data acquisition: When a patient applies for a blood transfusion, the status of the above six independent predictors is automatically or manually entered through the data input module.

[0082] (2) Risk calculation: The system transmits the data to the risk calculation module, which automatically executes the risk scoring algorithm, assigns values ​​to each factor, and calculates the total risk score.

[0083] (3) Risk stratification: The risk stratification module converts the total risk score into one of three levels: “low risk”, “medium risk” or “high risk” based on a preset threshold.

[0084] (4) Results output: The results output module generates a report that clearly displays the risk level and corresponding management recommendations.

[0085] like Figure 3 As shown, the differentiated clinical management path corresponding to this system is as follows:

[0086] (1) Low risk level (≤2 points): Follow the standard blood transfusion procedure. Perform antibody identification and antigen-negative blood screening as usual, without special acceleration or resource allocation.

[0087] (2) Medium risk level (3-4 points): Implement early warning and preparation procedures. Upon receiving the application, the blood transfusion department will immediately initiate expanded screening, prioritize the allocation or reservation of rare blood types that may be needed, and notify clinical departments to prepare for possible delays.

[0088] (3) High-risk level (≥5 points): Implement emergency intervention and multidisciplinary collaboration procedures. The system will automatically trigger the highest level warning and notify the head of the blood transfusion department and the attending physician. Immediately activate the department's difficult blood matching plan, contact the regional blood center to seek rare blood sources, and assess the risk-benefit ratio of emergency blood transfusion in the case of incomplete blood matching.

[0089] like Figure 4 As shown, in one embodiment of the present invention, a six-factor risk model for predicting crossmatching difficulties was successfully constructed and validated through a retrospective analysis of clinical data from 521 patients with irregular red blood cell antibodies. The model demonstrated good discriminative ability (AUC=0.819) and calibration. Figure 4 A shows a forest plot of independent risk factors affecting crossmatch difficulties, including screening patterns, multiple myeloma, hematology hospitalization, age ≥60 years, pregnancy-related diseases, and difficulties in blood typing. Figure 4 B is the ROC curve of the logistic regression model, with AUC = 0.819 (95% CI: 0.766–0.872).

[0090] The following describes the research process of an embodiment of the present invention:

[0091] The study in this embodiment of the invention is a single-center retrospective cohort study aimed at constructing and preliminarily validating a clinical risk model for predicting crossmatching difficulties in patients with irregular red blood cell antibody positivity.

[0092] This invention retrospectively reviewed the erythrocyte irregular antibody (ERA) test records of all inpatients at a tertiary hospital in Fujian Province from January 2020 to June 2025, identifying 574 patients with positive ERA screening results. Subsequently, screening was conducted according to the following inclusion and exclusion criteria:

[0093] The missing rate of key clinical variables (such as diagnosis, department, age) exceeded 20% (n=18);

[0094] For patients who tested positive multiple times, only the first positive record was included, and subsequent duplicate data were excluded (n=26).

[0095] Patients whose outpatient or emergency observation period is less than 24 hours (n=9).

[0096] Inclusion criteria: Patients who tested positive for irregular red blood cell antibodies after meeting the above exclusion criteria. Ultimately, a total of 521 patients were included in the final analysis cohort.

[0097] Study endpoint and grouping: The observation endpoint of this study was crossmatching difficulty. The procedure was defined as follows: using three randomly selected bags of blood from donors with the same blood type, a standard crossmatch test was performed. If all three bags were incompatible, it was considered "crossmatching difficulty"; if at least one bag was compatible, it was considered "crossmatching successful". Based on the initial screening results, all 521 patients were divided into two groups:

[0098] 1-2 well positive group: In the erythrocyte irregular antibody screening test, only 1 or 2 screening cell wells showed a positive reaction (n=428).

[0099] 3-well all-positive group: In the erythrocyte irregular antibody screening test, all 3 screening cell wells showed positive reactions (n=93).

[0100] Data Collection and Variable Definition: The following variables were collected through the hospital information system and laboratory information system:

[0101] Demographic characteristics: age (groups: <60 years, ≥60 years), sex.

[0102] Clinical characteristics: Department of admission (please specify whether it is "Hematology Department").

[0103] Clinical diagnosis (highlight “multiple myeloma” and “pregnancy-related disease”).

[0104] Laboratory data: Irregular antibody screening pattern for red blood cells (1-2 wells positive vs. 3 wells all positive).

[0105] The existence of "difficulty in blood type identification" is defined as the occurrence of discrepancies between the forward and reverse typing results in the initial ABO / RhD blood type test, requiring additional testing for confirmation.

[0106] Crossmatch results (difficult / successful).

[0107] Other exposure history: history of blood transfusion, history of pregnancy.

[0108] Screening and identification methods for irregular red blood cell antibodies: All plasma samples were screened for irregular red blood cell antibodies using standard microcolumn gel card technology. Commercially available reagent cells were used for screening. All procedures were strictly performed in accordance with the manufacturer's instructions and laboratory standard operating procedures, and the results were interpreted independently by two experienced technicians.

[0109] Statistical Analysis and Model Building: Data analysis was performed using SPSS 28.0 software. Baseline Characteristic Description and Univariate Analysis: Groups were formed based on the difficulty of crossmatching, and the characteristics of each group were described. Categorical variables were expressed as the number of cases (percentage), and chi-square tests were used for inter-group comparisons. Continuous variables conforming to a normal distribution were expressed as mean ± standard deviation, and t-tests were used; otherwise, the median was used, and nonparametric tests were employed. Multivariate Logistic Regression Analysis: Variables with p-values ​​<0.1 in the univariate analysis were used as candidate variables. A multivariate binary logistic regression analysis was performed using a stepwise backward method to screen for independent risk factors for crossmatching difficulties, and adjusted odds ratios and their 95% confidence intervals were calculated. Risk Scoring System Development: Based on the β-regression coefficients of each independent factor in the final multivariate model, they were proportionally converted to integer risk scores to construct a simplified clinical risk scoring tool. Model Performance Evaluation: Discrimination: Receiver operating characteristic (ROC) curves were plotted, and the area under the curve was calculated. Risk stratification and management pathway development: Based on the actual distribution of risk scores and the opinions of clinical experts, we define the score thresholds for low, medium and high risks, and develop corresponding standardized blood preparation and communication management pathways for each risk level.

[0110] Baseline characteristics of the study population: This study ultimately included 521 patients who tested positive for irregular erythrocyte antibodies. Detailed baseline characteristics can be found in [link to study population]. Figure 5 Based on the irregular erythrocyte antibody screening reaction pattern, 428 patients (82.1%) were classified into the 1-2 well positive group and 93 patients (17.9%) into the 3-well all-positive group. The mean age of all patients was 58.3 ± 16.2 years, of which 54.7% (285 / 521) were male. Clinically, a significant proportion of patients were hematology inpatients (32.4%) and multiple myeloma patients (13.1%). Intergroup comparisons showed that the mean age of patients in the 3-well all-positive group was significantly higher than that in the 1-2 well positive group (61.2 ± 15.3 years vs. 57.6 ± 16.3 years, P = 0.041). Meanwhile, the proportions of hematology hospitalizations (53.8% vs. 27.8%, P<0.001), diagnoses of multiple myeloma (43.0% vs. 6.5%, P<0.001), and difficulties in blood typing (22.6% vs. 8.4%, P<0.001) were significantly higher in the 3-well all-positive group. Conversely, the proportion of pregnancy-related diseases was lower in the 3-well all-positive group (5.4% vs. 13.6%, P=0.023). There were no significant differences between the two groups in terms of sex, ABO blood type distribution, transfusion history, or pregnancy history.

[0111] Incidence and univariate analysis of crossmatching difficulties: Among all 521 patients, 78 (15.0%) experienced crossmatching difficulties. The incidence in the 3-well all-positive group (20.4%) was higher than that in the 1-2 well positive group (13.8%), but the difference was not statistically significant (P=0.095). The results of the univariate analysis of crossmatching difficulties are shown below. Figure 6 Compared to the group with successful crossmatching (n=443), the group with difficult crossmatching (n=78) had a higher proportion of patients aged ≥60 years (69.2% vs. 56.4%, OR=1.74, 95%CI: 1.03–2.93, P=0.036) and a greater proportion of patients from the hematology department (47.4% vs. 29.8%, OR=2.15, 95%CI: 1.32–3.51, P=0.002). Regarding disease diagnosis, multiple myeloma (OR=4.74, P<0.001) and autoimmune diseases (OR=2.06, P=0.037) were significantly associated with an increased risk of difficult crossmatching, while pregnancy-related diseases showed a protective effect (OR=0.35, P=0.045). Among laboratory indicators, difficulty in blood typing is a strong predictor of crossmatching difficulties, increasing the risk by approximately 4 times (OR=5.06, P<0.001).

[0112] Multivariate Logistic Regression Analysis and Identification of Independent Risk Factors: To control for confounding factors, we conducted a multivariate binary logistic regression analysis. Variables with p-values ​​< 0.1 in the univariate analysis were included. Six independent predictors of crossmatching difficulties were ultimately identified using a stepwise backward method (likelihood ratio test). See [link to relevant documentation]. Figure 7 The analysis showed that a 3-well all-positive screening pattern (adjusted odds ratio [aOR]=2.34, P=0.015), diagnosis of multiple myeloma (aOR=5.05, P<0.001), hematology hospitalization (aOR=2.25, P=0.005), age ≥60 years (aOR=1.97, P=0.019), and difficulty in blood typing (aOR=4.57, P<0.001) were independent risk factors for increased crossmatching difficulties. Pregnancy-related diseases were the only protective factor (aOR=0.24, P=0.011). The forest plot of the model is shown below. Figure 4 A.

[0113] Regarding the six independent predictors:

[0114] (1) Irregular antibody screening response pattern on red blood cells: deepening the predictive value from "presence or absence" to "pattern". This invention, for the first time in a prospective risk prediction model, establishes "all 3 wells positive" as an objective and early-obtainable pattern indicator as an independent quantitative predictor. This marks a shift in risk assessment from relying on the final "antibody identification result" to focusing on the initial "screening response pattern", providing clinicians with more time for early warning and preparation. Therefore, before antibody specificity is clear, the screening pattern can already provide key early risk signals, guiding blood banks to prioritize resource allocation.

[0115] (2) Multiple myeloma: A strongly confirmed key risk factor. This study confirms that a diagnosis of multiple myeloma (MM) is one of the strongest risk factors for predicting crossmatching difficulties (aOR=5.05). The underlying causes are multifaceted: firstly, MM patients often have immune dysfunction, which may lead to more complex and difficult-to-clear irregular antibody production; secondly, these patients often receive treatment regimens containing proteasome inhibitors, immunomodulators, etc., which may affect erythrocyte antigen expression or alter the immune response; thirdly, MM patients frequently undergo repeated blood transfusions due to anemia and chemotherapy requirements, significantly increasing the chance of alloimmunization. This study, through multivariate analysis, after controlling for factors such as transfusion history and age, still clearly demonstrates the independent contribution of MM, highlighting the importance of the disease's inherent characteristics (such as immune microenvironment dysregulation), going beyond the explanation scope of simple "transfusion exposure."

[0116] (3) Hematology hospitalization and age ≥60 years: pointing to "complex clinical conditions" and "immunosenescence". Hematology hospitalization and age ≥60 years are often considered confounding variables in previous studies, but they have independent predictive value in this model. The label "hematology hospitalization" may actually aggregate multiple high-risk characteristics: in addition to the aforementioned MM, it also includes other diseases requiring frequent transfusions (such as aplastic anemia, myelodysplastic syndrome) or accompanied by complex immune states (such as autoimmune hemolytic anemia, post-hematopoietic stem cell transplantation). This suggests that patients from specific high-risk departments should be pre-judged by the transfusion department as a group requiring more attention and resource preparation. Age, as a risk factor, may be related to "immunosenescence". With increasing age, the function of the immune system changes, which may lead to changes in the immune response pattern to transfused red blood cell antigens, producing more durable or clinically significant antibodies. In addition, elderly patients have more comorbidities and more opportunities for cumulative medical exposure (such as surgery, transfusion). This study further clarifies its direct impact on the operational outcome of "blood matching difficulties".

[0117] (4) Pregnancy-related diseases: a unique "protective" factor. One of the most noteworthy and unique findings of this study is that pregnancy-related diseases (including normal pregnancy and complications) are a significant protective factor (aOR=0.24). This seems to contradict many studies that emphasize pregnancy as one of the main pathways of red blood cell alloimmunization. However, upon in-depth analysis, this "contradiction" reveals the complexity of clinical practice. Possible explanations include: First, pregnancy-related antibodies are most commonly anti-D, followed by anti-E, anti-c, and other Rh system antibodies, as well as anti-M. For these common antibodies, blood banks usually have mature screening, identification, and antigen-negative blood reserve procedures, resulting in relatively high matching efficiency. Second, obstetricians are usually highly vigilant about pregnant women with irregular antibodies and will communicate with the transfusion department in advance to develop transfusion plans. This "anticipatory management" greatly reduces the difficulty of emergency blood matching. Third, the endpoint of this study is "difficulty in crossmatching," rather than "the presence of antibodies." Many pregnancy-induced antibodies may have low titers or react at room temperature, limiting their clinical significance and not necessarily causing incompatibility in crossmatching with anti-human globulin mediators. Therefore, this study does not deny that pregnancy is a risk factor for alloimmunization, but rather emphasizes that, given the presence of antibodies, the risk of crossmatching difficulties caused by pregnancy-triggered antibodies is significantly lower than that of antibodies produced in the context of hematologic malignancies or other diseases. This finding has significant practical implications for risk stratification, suggesting greater vigilance should be exercised regarding antibody-positive patients with non-pregnancy-related causes (especially hematologic malignancies).

[0118] (5) Difficulty in Blood Typing: A Strong Early Warning Signal That Has Been Overlooked. Difficulty in blood typing is one of the most predictive indicators in this study (aOR=4.57), a point that has not been adequately emphasized in previous studies. Difficulty in blood typing often stems from ABO subtypes, Rh variants, acquired B antigens, erythrocyte polyagglutination, or interference from autoantibodies, which in itself indicates the complexity of the patient's erythrocyte blood group system. The results show that challenges encountered in the initial stage of the transfusion process—blood typing—are a very strong warning signal that subsequent crossmatching may encounter difficulties, suggesting that the patient's immunohematological background may be different from the norm, requiring immediate initiation of a more in-depth investigation and advanced blood preparation plan. This study provides clear quantitative evidence for its importance.

[0119] Model Innovation and Clinical Translation: From Statistical Association to Action Guidelines. Unlike many previous studies that merely identified risk factors or constructed complex statistical models, the greatest innovation of this invention lies in completing a full closed loop from "risk identification" to "tool creation" and then to "pathway development." Our five-factor risk scoring system (with independent calculations excluding protective factor scores) is simple and intuitive. All variables can be obtained at the time of transfusion request or during blood typing, requiring no additional testing and possessing strong clinical operability. This model focuses on solving the most pressing issue in the daily work of blood transfusion departments: "blood matching efficiency and safety." Based on the scoring-based three-color, four-tier management path, limited blood resources and expert attention are precisely directed to high-risk patients, achieving a paradigm shift from "homogenized processing" to "precise triage response." This is expected to significantly shorten the waiting time for high-risk patients and systematically reduce the safety risks caused by hasty blood matching.

[0120] Performance evaluation of the predictive model: The multivariate logistic regression model showed good predictive performance. The Hosmer-Lemeshow goodness-of-fit test results (χ²=5.13, P=0.743) indicated good model calibration. The model had excellent discrimination, with an area under the receiver operating characteristic curve (AUC) of 0.819 (95% CI: 0.766–0.872) for predicting crossmatch difficulties. Figure 4 B). At the optimal cutoff value, the model's predictive sensitivity was 74.4%, specificity was 76.5%, and overall accuracy was 76.2%. The model had a high negative predictive value (93.7%), indicating that patients predicted to be low-risk are unlikely to experience actual crossmatching difficulties.

[0121] Risk scoring system establishment and hierarchical management strategy: To facilitate rapid clinical application, we transformed the multifactor model into an integer risk scoring system based on six independent factors. (See [link to relevant documentation]). Figure 8 Each factor is assigned a corresponding score based on its regression coefficient weight, and a total risk score is calculated. For example, Figure 9 As shown, based on the β regression coefficients of each independent factor in the final multifactor model, they are proportionally converted into integer hazard integrals. The specific conversion steps are as follows:

[0122] 1) Based on the highest β coefficient: Multiple myeloma (β≈1.619) and blood typing difficulties (β≈1.519) have the highest and closest β values, so they are both assigned the highest score of 3 points;

[0123] 2) Proportional scaling and rounding: The β value of all 3 wells positive (β≈0.850) is about 52% of the highest β (0.850 / 1.619≈0.52), and is assigned 2 points after rounding; the β value of hematology inpatients (β≈0.811) and patients aged ≥60 years (β≈0.678) is about less than 50% of the highest β, so is assigned 1 point;

[0124] 3) Negative scoring for protective factors: The β value for pregnancy-related diseases is negative (≈-1.427), and its absolute value is close to the highest β value, so -3 points are assigned to offset the risk.

[0125] The larger the absolute value of the β coefficient (e.g., 1.619 for multiple myeloma, 1.519 for blood typing difficulties), the higher the corresponding aOR (5.05, 4.57), and the higher the score (3 points), reflecting the logic of "higher risk, higher score". Converting the β coefficient to an integer score avoids complex logistic regression calculations, allowing clinicians to quickly assess risk through "addition / subtraction" (e.g., if a patient has both "multiple myeloma" and "blood typing difficulties", the total score = 3 + 3 = 6 points, directly indicating high risk).

[0126] Then as Figure 10 As shown, based on the total risk score, patients are divided into three risk levels (low-risk group 0-2 points, medium-risk group 3-4 points, and high-risk group ≥5 points) and corresponding management paths are established: the low-risk group uses a green label and follows the routine process, the medium-risk group uses a yellow label and initiates the priority process, and the high-risk group uses a red label and initiates the special process, thus forming a management closed loop.

[0127] While specific embodiments of the present invention have been described above, those skilled in the art should understand that the specific embodiments described are merely illustrative and not intended to limit the scope of the present invention. Equivalent modifications and variations made by those skilled in the art in accordance with the spirit of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A risk assessment system for crossmatch difficulties, characterized in that: include: The data input module is used to obtain key patient information related to crossmatching difficulties. The key patient information includes: red blood cell irregular antibody screening results, whether it is a diagnosis of multiple myeloma, whether it has visited a hematology department, whether the patient is ≥60 years old, whether it is a pregnancy-related disease, and whether blood type identification is difficult. The risk calculation module is configured to assign corresponding weights to the patient's key information and calculate a total risk score based on a preset risk scoring algorithm; the specific assignment rules used in the risk scoring algorithm are as follows: Three positive wells are assigned a score of +2. Multiple myeloma diagnosis, assigned a score of +3; Hospitalization in the hematology department, +1 point; For those aged 60 or older, add 1 point. Pregnancy-related diseases are assigned a score of -3. Difficult blood type identification will be assigned a +3 point value. Total risk score = 2×X1 + 3×X2 + 1×X3 + 1×X4 - 3×X5 + 3×X6 X1 to X6 are binary variables, with 1 for yes and 0 for no; The risk stratification module is configured to classify patients into at least three levels: low risk, medium risk, and high risk, based on the numerical range of their total risk score. The Results Output and Recommendations module is used to output the risk level and the corresponding differentiated clinical management pathway.

2. The system according to claim 1, characterized in that: The three levels are: a total risk score of ≤2 points indicates low risk, a total risk score between 3 and 4 points indicates medium risk, and a total risk score of ≥5 points indicates high risk.

3. The system according to claim 1, characterized in that: The key patient information was based on a retrospective clinical cohort study. The study screened candidate variables with a p-value <0.1 through univariate analysis, and then determined six independent predictors through multivariate logistic regression. Among them, the results of irregular red blood cell antibody screening, whether it was a diagnosis of multiple myeloma, and whether blood type identification was difficult were the strongest independent predictors of risk, while whether it was a pregnancy-related disease state showed a unique protective effect.

4. The system according to claim 1, characterized in that: The risk calculation module determines the weights and values ​​of each factor among the six independent predictors based on the results of multivariate logistic regression analysis of a retrospective clinical cohort. First, the relative importance of each factor among the six independent predictors is determined based on the regression coefficients of the multivariate logistic regression analysis. The regression coefficients reflect the independent influence of the factor on the risk of crossmatching difficulties. The larger the absolute value of the regression coefficient, the higher the weight. Then, the regression coefficients are converted into integer risk integrals according to the proportion and assigned values.

5. The system according to claim 1, characterized in that: The results of the irregular antibody screening for red blood cells include two types: 1-2 positive wells and 3 positive wells.

6. The system according to claim 1, characterized in that: The specific risk levels and corresponding differentiated clinical management pathways are as follows: Low-risk level: Standard blood transfusion procedures are followed, and antibody identification and antigen-negative blood screening are performed routinely, without the need for special acceleration or resource allocation; Medium risk level: Implement early warning and preparation procedures. Upon receiving the application, the blood transfusion department will immediately initiate expanded screening, prioritize the allocation or reservation of rare blood types that may be needed, and notify clinical departments to prepare for possible delays. High-risk level: Implement emergency intervention and multidisciplinary collaboration procedures. The system will automatically trigger the highest level of warning, notify relevant personnel to immediately activate the department's difficult blood matching plan, contact the regional blood center to seek rare blood sources, and assess the risk and benefit ratio of emergency blood transfusion in the case of incomplete blood matching.

Images