A method, device, and program product for predicting the number of days of optimized perioperative bed turnarounds based on postoperative complications

Abstract

The application relates to the field of intelligent medical treatment, in particular to a method, equipment and program product for predicting the days of optimized bed turnover in the perioperative period based on postoperative complications. The method comprises the following steps: obtaining respiratory clinical data and surgical invasiveness parameters of a patient to be hospitalized for surgery, wherein the respiratory clinical data at least include whether there is a history of asthma, whether there is hyperlipidemia, gender, a surgical site, whether there is pulmonary embolism, and whether there is reoperation; calculating a respiratory risk probability through the respiratory clinical data, and predicting a risk probability after PNB intervention; calculating a coupling factor through the respiratory risk probability and the risk probability after PNB intervention; calculating an invasiveness adjustment coefficient through the surgical invasiveness parameters; calculating an expected bed saving day through the coupling factor and the invasiveness adjustment coefficient; and dynamically adjusting the bed turnover according to the expected bed saving day. The application can be used to predict the bed days saved after using PNB, and has good application value for hospital bed management.

Description

Technical Field

[0001] This application relates to the field of intelligent healthcare, specifically to a method, device, program product, and computer-readable storage medium for predicting the number of days of optimized perioperative bed turnover based on postoperative complications. Background Technology

[0002] In clinical practice, the prediction and management of postoperative complications are crucial for optimizing hospital bed turnover and improving patient recovery. Existing technologies, such as reference patents CN119446505A and CN115775630A, provide methods for predicting postoperative pulmonary complications after cardiac and major vascular surgery based on machine learning models and methods for predicting the probability of postoperative pulmonary complications based on sleep stage data, respectively. These methods have achieved significant results in predicting specific postoperative complications. However, these technologies mainly focus on predicting single types of surgery or specific complications and do not delve into how to optimize bed turnover based on the prediction results. Especially in complex situations involving multiple types of surgery and various complications, the generalization ability and practical application effectiveness of their prediction models still need improvement. Furthermore, existing technologies often lack a deep integration of patient individual characteristics and surgical invasiveness parameters, which limits the accuracy and applicability of the prediction models. In particular, insufficient consideration is given to the interaction between surgical methods (open or minimally invasive) and specific patient vulnerability characteristics (such as asthma and hyperlipidemia), which may lead to excessively high misjudgment rates in practical applications, thus failing to effectively guide clinical decision-making, especially in optimizing bed turnover. Summary of the Invention

[0003] Based on the above analysis, this application aims to address the insufficient generalization ability of existing postoperative complication prediction models in predicting multiple types of surgeries and various complications, and the key issue of how to effectively optimize bed turnover based on prediction results. By deeply integrating patient individual characteristics with surgical invasiveness parameters, a more accurate prediction model is constructed, thereby achieving efficient management of bed turnover. Especially in dealing with complex postoperative complication scenarios, by integrating specific patient vulnerability characteristics and surgical procedures, the prediction of perioperative bed turnover days is optimized to improve the accuracy of clinical decision-making and the utilization rate of bed resources. Specifically, a method for optimizing perioperative bed turnover day prediction based on postoperative complications is provided, including: Acquire respiratory clinical data and surgical invasive parameters of hospitalized patients awaiting surgery. The respiratory clinical data shall include at least the following: history of asthma, presence of hyperlipidemia, gender, surgical site, presence of pulmonary embolism, and need for reoperation. The respiratory clinical data are used to calculate the respiratory risk probability and predict the risk probability after PNB intervention. The coupling factor is calculated using the respiratory risk probability and the risk probability after PNB intervention; The invasiveness adjustment coefficient is calculated using the surgical invasiveness parameters; The expected number of bed-saving days is calculated using the coupling factor and the invasive adjustment coefficient.

[0004] Bed turnover is dynamically adjusted based on the expected number of days saved per bed.

[0005] The invasiveness moderating coefficient is dynamically selected based on the surgical invasiveness parameters of respiratory clinical data. It is calculated as the mediating proportion of the reduction in hospital stay due to prevention of pulmonary complications within the overall reduction in hospital stay resulting from PNB intervention when any type of surgical invasiveness parameter is implemented. The surgical invasiveness parameters include open surgery or minimally invasive surgery. The mediating proportion is obtained through multivariate mediation analysis based on the Baron-Kenny framework. The confidence interval is estimated using the Bootstrap method (5000 iterations), and the resulting mediating proportions are point estimates: λ = 0.295 for minimally invasive surgery and λ = 0.40 for open surgery. When the surgical type does not belong to either of these two categories, λ is taken as the default value of the total mediating proportion point estimate of 0.279.

[0006] The coupling factor is calculated by first calculating the difference between the respiratory risk probability and the risk probability after PNB intervention, and then performing a weighted product on the difference.

[0007] The weights for the weighted product calculation are as follows: obtain the respiratory clinical dataset of patients with pulmonary complications; input the respiratory clinical dataset into the first regression prediction model to calculate the regression coefficients, where the independent variable of the first regression prediction model is postoperative pulmonary complications and the dependent variable is the length of hospital stay.

[0008] The respiratory risk probability is calculated using a second regression model to determine the probability of postoperative pulmonary complications based on respiratory clinical data. The respiratory clinical data includes factors such as history of asthma, hyperlipidemia, gender, surgical site, pulmonary embolism, and reoperation. The respiratory clinical data only includes variables that are available before or during surgery to meet the clinical application scenarios of preoperative risk prediction.

[0009] The probability of risk after PNB intervention is simulated by the ratio of respiratory clinical data to the predicted respiratory risk after PNB intervention. The specific calculation formula is: P_PNB = P_basic × OR_PNB / (1 Pbase + Pbase × OR_PNB), where OR_PNB = 0.395, is derived from a multifactor logistic regression model.

[0010] The method also includes bed turnover early warning, and the respiratory risk probability is calculated by the first regression prediction model to predict the additional hospitalization days due to postoperative pulmonary complications; when patients are predicted to have a hospitalization duration of more than N days due to postoperative pulmonary complications, they are marked as key patients and a mandatory preoperative nerve block list is generated.

[0011] The purpose of this invention is to provide a computer program product that includes a computer program or instructions, which are executed by a processor to implement the above-mentioned method for predicting the number of days of perioperative bed turnover based on postoperative complications.

[0012] The purpose of this invention is to provide a computer device comprising a memory, a processor, and a computer program or instructions stored in the memory, wherein the computer program or instructions are executed by the processor to implement the above-described method for predicting the number of days of perioperative bed turnover based on postoperative complications.

[0013] The purpose of this invention is to provide a computer-readable storage medium having a computer program or instructions stored thereon, which are executed by a processor to implement the above-described method for predicting the number of days of perioperative bed turnover based on postoperative complications.

[0014] Advantages of this invention: 1. To address the current lack of in-depth exploration of how to optimize bed turnover based on prediction results, this invention utilizes regression coefficients to construct a postoperative pulmonary complication (PPC)-derived hospital stay penalty index (coupling index). The coupling index solves the pain point of clinical inability to quantify the time cost of complications. It calculates and predicts the reduction of hospital stay after PNB intervention, providing a management basis for clinical bed turnover management, guiding patients in preoperative intervention, and improving the quality of prognosis.

[0015] 2. Based on subgroup analysis, this invention found that the mediating weight (40%) of pulmonary complications in open surgery was significantly higher than that in minimally invasive surgery. Utilizing this finding, this invention establishes a differentiated anesthesia decision-making logic based on 'invasiveness stratification', which is unprecedented in existing technologies. This provides a basis for anesthesia decisions, thereby improving the accuracy and reliability of predicting postoperative hospital stay and the reduction in hospital stay days after PNB intervention. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic diagram of the process for predicting the number of days of perioperative bed turnover based on postoperative complications, provided in an embodiment of the present invention; Figure 2 This is a mediating analysis diagram of the effect of preoperative peripheral nerve block (PNB) on postoperative hospital stay provided in an embodiment of the invention. The total effect of preoperative PNB on postoperative hospital stay is represented by path c. After controlling for the incidence of postoperative pulmonary complications, the direct effect of preoperative PNB on postoperative hospital stay is represented by path cc. The effect of preoperative PNB on the incidence of postoperative pulmonary complications is represented by path a. After controlling for preoperative PNB, the effect of postoperative pulmonary complications on postoperative hospital stay is represented by path b. Through the mediating effect of postoperative pulmonary complications, the indirect effect of preoperative PNB on postoperative hospital stay is calculated as the product of paths a and b (a×b). Figure 3 A schematic diagram of a system for predicting the number of days of perioperative bed turnover based on postoperative complications, provided in an embodiment of the present invention; Figure 4 A schematic diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0019] In some of the processes described in the specification, claims, and accompanying drawings of this invention, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as S101, S102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0020] Figure 1 The schematic diagram of the method for predicting the number of days of perioperative bed turnover based on postoperative complications provided in this embodiment of the invention specifically includes: S1: Obtain respiratory clinical data and surgical invasive parameters from hospitalized patients awaiting surgery; In one embodiment, respiratory clinical data includes a history of asthma, presence of hyperlipidemia, gender, surgical site, presence of pulmonary embolism, and whether reoperation is required. All variables are clinical information available preoperatively or intraoperatively, excluding events that can only be confirmed postoperatively (such as deep vein thrombosis in the lower extremities or unplanned transfer from the ward to the ICU). Surgical invasiveness refers to the degree of invasiveness involved in diagnostic or treatment procedures that directly enter the body through surgery, instruments, etc.; for example, minimally invasive surgery is less invasive than open surgery. The surgical invasiveness parameter (open / minimally invasive) is used to determine the invasiveness moderating coefficient λ, rather than as a direct input variable for the logistic regression model.

[0021] In one specific embodiment, the interaction between peripheral nerve block (PNB), postoperative pulmonary complications, and postoperative hospital stay in elderly patients undergoing major thoracic and abdominal surgery remains unclear. This invention clarifies the mediating role of postoperative pulmonary complications in the relationship between preoperative PNB and postoperative hospital stay.

[0022] This invention employs a retrospective cohort study design and has been approved by the Clinical Research Ethics Committee of Peking University First Hospital (Approval No.: 2024

[224] ; Beijing, China). Given that data were retrospectively extracted from electronic medical records during the study and there was no postoperative follow-up, the Ethics Committee waived the requirement for written informed consent. The confidentiality of patient information was strictly maintained. This study was prospectively registered with the Chinese Clinical Trial Registry (chictr.org.cn, ChiCTR2400087610) on July 31, 2024.

[0023] Patient identification and screening were conducted from December 1, 2021 to July 31, 2023 using the electronic medical record system of Peking University First Hospital (Beijing, China). Inclusion criteria included patients aged 65 years and older undergoing elective major non-cardiac thoracic and abdominal surgery lasting at least two hours under general anesthesia. Exclusion criteria were as follows: (1) American Society of Anesthesiologists (ASA) Class IV or above; (2) the type of surgery (such as laparoscopic cholecystectomy and complication-free hernia repair) generally did not require thoracolumbar nerve block or epidural-general anesthesia; (3) a history of neurological disorders such as epilepsy, Parkinson's disease, or myasthenia gravis; (4) reoperation or organ transplantation during the same hospitalization; and (5) incomplete data.

[0024] Anesthesia and perioperative management: Intraoperative monitoring included pulse oximetry, electrocardiogram, bispectral index (BIS), nasopharyngeal temperature, arterial blood pressure, inhaled / exhaled sevoflurane and carbon dioxide concentrations, and urine output. The patient received general anesthesia with or without peripheral nerve block (PNB). Anesthesia was induced with sufentanil, propofol / etomidate, and rocuronium bromide, followed by maintenance via intravenous infusion of propofol and remifentanil, sufentanil injection, and inhaled sevoflurane as needed. The target BIS range was 40 to 60. Mechanical ventilation was initiated with an oxygen-air mixture, and fluid resuscitation was performed according to standard protocol. Vasopressors were used if necessary to maintain mean arterial pressure within 30% of baseline.

[0025] On each workday during the study period, two anesthesiologists were responsible for administering nerve blocks. Peripheral nerve blocks (PNBs) were performed at the discretion of the attending anesthesiologist, usually preoperatively, and included thoracic paravertebral blocks for lung surgery, transversus abdominis plane blocks for gastrointestinal and prostate surgeries, and transversus abdominis plane blocks, erector spinae plane blocks, or quadratus lumborum plane blocks for kidney / ureteral surgeries. These blocks were administered preoperatively by the designated anesthesiologist using 0.375% ropivacaine solution, guided by a GE ultrasound system with linear or curved array probes. The choice of PNB type and injection volume was based on the type of surgery, patient condition, and ultrasound-guided diffusion of the injection. For example, patients undergoing lobectomy would receive a unilateral thoracic paravertebral block using 20 mL of 0.375% ropivacaine, while patients undergoing gastrointestinal surgery would receive bilateral transversus abdominis plane blocks using 40 mL of 0.375% ropivacaine.

[0026] Postoperatively, depending on the patient's clinical and hemodynamic status, the patient is transferred to the post-anesthesia recovery room or the intensive care unit (ICU). Morphine or sufentanil is administered via patient-controlled intravenous analgesia, with nonsteroidal anti-inflammatory drugs, oxcontin, or other analgesics given as needed for supplemental analgesia.

[0027] Data collection and results: Data were retrospectively extracted from the electronic medical record system. Collected demographic data included age, sex, and body mass index (BMI). Preoperative data included surgical diagnosis, comorbidities (Charlson Comorbidity Index 20, CCI), laboratory test values, smoking and alcohol consumption history, and ASA classification. Intraoperative data included surgical type and duration, anesthesia type and duration, nerve block type and location, fluid management, blood transfusions, and urine output. Postoperative data included ICU admission, surgery-related complications, postoperative pulmonary complications, and length of hospital stay. Postoperative pulmonary complications were defined as pneumonia, atelectasis, and respiratory failure.

[0028] Statistical analysis: Patients were divided into two groups based on whether PNB was used during anesthesia. Normally distributed data are expressed as mean ± standard deviation and analyzed using Student's t-test. Non-normally distributed data are expressed as median and interquartile ranges, and differences were assessed using the Mann-Whitney U test. Categorical variables were compared using the chi-square test or Fisher's exact test.

[0029] Generalized linear models were used to analyze factors associated with postoperative hospital stay. Variables with p-values ​​<0.20 from univariate analysis, as well as clinically significant variables, were included in multivariate linear regression models. Logistic regression models were applied to analyze perioperative factors associated with postoperative pulmonary complications, including the relationship between PNB and postoperative pulmonary complications.

[0030] To elucidate the mechanism by which postoperative pulmonary embolism (PNB) may affect postoperative hospital stay, a mediation analysis was conducted. This analysis aimed to determine the extent to which the effect of PNB on hospital stay is mediated by postoperative pulmonary complications. Using the methodological framework of Baron and Kenny, the study described the causal pathway from PNB to hospital stay, with postoperative pulmonary complications as a potential mediating variable. This involved assessing the direct effect of PNB on hospital stay (pathway cc) and its indirect effect through postoperative pulmonary complications (pathways a and b). Figure 2 (As shown). The total effect (path c), indirect effect, and direct effect were all quantified using the "mediation" software package, and confidence intervals were estimated using the bootstrap method (5000 iterations). The process was repeated after adjusting for confounding factors. The adjusted coefficients show the total, direct, and indirect effects after adjusting for potential confounding factors. Confounding factors were selected based on: (1) independent predictors of length of hospital stay in the multivariate linear regression model; (2) variables with or without PNB imbalance between groups; and (3) clinically significant variables, such as ASA classification. In cases where multicollinearity exists in the inducing factors, a representative variable was selected.

[0031] The mediation analysis model included variables such as age, sex, BMI, CCI, chronic obstructive pulmonary disease (COPD), hemiplegia, asthma, chronic smoking, hemoglobin, alanine aminotransferase, albumin, ASA classification, high-sensitivity cardiac troponin I (hs-cTnI), surgical site, surgical type, surgical duration, ICU admission, reoperation, sepsis, and unplanned ICU transfer from general wards. Subgroup analyses were performed based on PNB type, surgical type (minimally invasive vs. open surgery), and surgical site (thoracic vs. abdominal surgery).

[0032] This invention also included a sensitivity analysis of propensity score matching. All collected baseline and intraoperative variables were included in propensity score matching. Propensity scores were calculated using a logistic regression model to predict the probability of accepting PNB. Propensity score matching was performed at a 1:1 ratio without substitution, using the nearest neighbor matching algorithm with a caliper width of 0.2. Between-group balance was analyzed by absolute standardized differences (ASDs), defined as the absolute difference of the mean, average rank, or proportion divided by the pooled standard deviation, calculated using the formula published by Austin. At that time, an imbalance was considered to exist between the two groups. Missing data was not replaced.

[0033] Power Calculation: In a fixed cohort of 1,840 patients, this invention performed a post-hoc power analysis of the mediating pathway using the "mediation" R software package, with α=0.05, the standard deviation of postoperative pulmonary complications (mediating variable) SD=0.254, and the regression coefficient of the mediating variable equal to 6.654 [in linear regression y]. i = b0 + b1x i + b2m i + In the above model, the standard deviation of the random error term was 7.221, and the correlation between PNB (predictor variable) and the mediator variable was -0.100. This study achieved 100% power. Statistical significance was set as a two-tailed test with P < 0.05, and all analyses used R 3.6.0.

[0034] In one specific embodiment, from December 1, 2021 to July 31, 2023, a total of 45,737 patients were screened, of whom 29,102 were excluded (8,117 due to non-general anesthesia surgery or incomplete information, 19,434 due to non-thoracic / abdominal surgery, and 1,551 due to emergency surgery). Of the 16,635 patients who underwent elective thoracic / abdominal surgery, 14,644 were further excluded (10,117 due to age under 65 years, 4,180 due to surgery time less than 2 hours, and 347 due to cardiac surgery). Of the remaining 1,991 patients, 151 were further excluded based on various criteria, including ASA IV (57 patients), patients for whom PNB was not typically required during surgery (34 patients), patients requiring combined epidural analgesia (11 patients), patients with preoperative epilepsy (4 patients), patients with preoperative Parkinson's syndrome (22 patients), patients with preoperative myasthenia gravis (9 patients), patients undergoing secondary surgery during hospitalization (6 patients), patients who had received kidney transplantation (2 patients), and patients with incomplete data (6 patients). Ultimately, a total of 1,840 patients were included, of whom 1,306 (71.0%) received PNB and 534 (29.0%) did not.

[0035] Patients receiving PNB had a lower proportion of males (63.5% vs. 68.9%, P=0.026), a lower incidence of chronic obstructive pulmonary disease (COPD) (6.0% vs. 9.0%, P=0.020), and lower CCI scores (2.5±0.7 vs. 2.6±0.8, P=0.025). They also had a lower ICU admission rate (23.4% vs. 31.1%, P=0.001), a lower incidence of postoperative pulmonary complications (5.3% vs. 10.9%, P<0.001), and a shorter postoperative hospital stay (7 [5, 9] vs. 7 [5, 11], P<0.001). Other significant differences were also observed between the two groups in baseline and perioperative data (as shown in Table 1).

[0036] Table 1. Baseline and perioperative data

[0037] Data are expressed as mean ± standard deviation, number of cases (%), or median (interquartile range). * indicates P < 0.05.

[0038] a These include atrial fibrillation, frequent (>6 times / min) or multifocal premature ventricular contractions, paroxysmal supraventricular tachycardia, second / third-degree atrioventricular block, and sick sinus syndrome.

[0039] b Smoking ≥10 cigarettes per day for ≥1 year (for former or current smokers).

[0040] c Drinking alcohol ≥ 2 times a day or drinking alcohol equivalent to ≥ 150 mL per week.

[0041] d This includes allogeneic blood transfusion and autologous blood transfusion.

[0042] e Reoperation performed due to postoperative anastomotic leakage, poor wound healing, or bleeding.

[0043] f Confirmed by lower extremity venous ultrasound.

[0044] g New pulmonary thrombosis confirmed by radionuclide scanning and CT angiography.

[0045] h It meets ≥2 criteria for systemic inflammatory response syndrome, has a clear focus of infection, and is accompanied by new functional impairment of at least one organ / system.

[0046] i A new infiltrative shadow was found on chest X-ray, accompanied by a body temperature >38°C and leukocytosis, and confirmed by a pulmonologist.

[0047] j Chest X-ray confirmed hypoxemia, requiring physical therapy to reopen the collapsed lung lobe.

[0048] k If PaO2 < 60 mmHg, or PaO2 / FiO2 < 300, or pulse oxygen saturation < 90%, and oxygen therapy is required.

[0049] Postoperative pulmonary complications risk factors and length of hospital stay: All baseline and perioperative variables were analyzed univariately to identify their association with postoperative pulmonary complications and length of hospital stay, taking into account variable collinearity. A total of 33 variables with P < 0.20 or clinical significance were included in the multivariate linear regression model (Tables 2 and 3). PNB (β = -0.697; 95% CI, -1.234 to -0.16; P = 0.011) and postoperative pulmonary complications (β = 4.634; 95% CI, 3.642 to 5.627; P < 0.001) were both independently associated with postoperative length of hospital stay after adjusting for confounding factors (Table 2). Logistic regression analysis showed that PNB (OR = 0.395; 95% CI, 0.261 to 0.597; P < 0.001) was independently associated with postoperative pulmonary complications, as shown in Table 4.

[0050] Table 2 Univariate and multivariate regression analyses of factors related to postoperative hospital stay.

[0051] β = unstandardized regression coefficient; CI = confidence interval. * indicates P < 0.05.

[0052] a This includes atrial fibrillation and other types of arrhythmias (see notes in Table 1).

[0053] b Reoperation performed due to postoperative anastomotic leakage, poor wound healing, or bleeding.

[0054] c Confirmed by lower extremity venous ultrasound. d Confirmed by radionuclide scanning and CT angiography.

[0055] e Simultaneously meeting ≥2 criteria for systemic inflammatory response syndrome, having a clearly identified focus of infection, and accompanied by ≥1 new functional impairment of an organ / system.

[0056] f Postoperative pulmonary complications included pneumonia, atelectasis, and respiratory failure. Variables with univariate p < 0.20 or clinical significance were included in the multivariate model.

[0057] Table 3. Complete results of univariate analysis of factors related to postoperative hospital stay.

[0058] β = unstandardized regression coefficient; CI = confidence interval. * indicates P < 0.20 (screening criterion for inclusion in multivariate models).

[0059] a The definition of arrhythmia is the same as in the notes of Table 1. a . b The definition of long-term smoking is the same as in the notes to Table 1. b . c Regular drinking is defined as in the notes to Table 1. c .

[0060] d Blood transfusion includes allogeneic and autologous blood transfusion. e The reasons for the reoperation are the same as those noted in Table 1. e . f Deep vein thrombosis in the lower extremities was confirmed by ultrasound.

[0061] g The pulmonary embolism was confirmed by radionuclide scanning and CT angiography. h The definition of sepsis is the same as in the notes to Table 1. h .

[0062] i Postoperative pulmonary complications include pneumonia, atelectasis, and respiratory failure.

[0063] Table 4. Univariate and multivariate logistic regression analyses of risk factors for postoperative pulmonary complications.

[0064] OR = Odds ratio; CI = Confidence interval. * indicates P < 0.05.

[0065] a Reoperation may be necessary due to postoperative anastomotic leakage, poor wound healing, or bleeding. b Deep vein thrombosis in the lower extremities was confirmed by ultrasound.

[0066] c The pulmonary embolism was confirmed by radionuclide scanning and CT angiography.

[0067] Note: PNB (OR=0.395) is the odds ratio source for the P_PNB calculation formula in this invention. Asthma (OR=4.745), hyperlipidemia (OR=2.727), gender (OR=1.633), surgical site, pulmonary embolism (OR=71.531), and reoperation (OR=5.976) are the sources of predictive variables for logistic regression in this invention.

[0068] Intermediary Analysis: After adjusting for potential confounding factors including age, sex, BMI, CCI, COPD, hemiplegia, asthma, long-term smoking, hemoglobin, alanine aminotransferase, albumin, ASA classification, hs-cTnI, surgical site, surgical type, surgical time, ICU admission, reoperation, sepsis, unplanned ICU transfer from ward, year of surgery, and analgesia administration within 72 hours postoperatively, the total and direct effects of PNB on postoperative hospital stay were quantified. The adjusted total effect was β = -0.987 (-1.540 to -0.41; P < 0.001), and the direct effect was β = -0.709 (95% CI, -1.225 to -0.16; P = 0.010). A statistically significant indirect association between postoperative pulmonary complication (PNB) and postoperative hospital stay was observed in postoperative pulmonary complications (adjusted E = -0.278%, 95% CI -0.471% to -0.12%, P < 0.001), accounting for 27.9% of the total effect (95% CI 12.3% to 64.0%) (Table 5).

[0069] Subgroup analyses were performed based on PNB type, surgical procedure type, and surgical site. Consistent with the primary findings, the adjusted model showed a statistically significant mediating effect of postoperative pulmonary complications in the following subgroups: transversus abdominis plane block (adjusted β -0.337, 95% CI -0.548 to -0.1, P<0.001), minimally invasive surgery (adjusted β -0.225, 95% CI -0.489 to -0.08, P<0.001), open surgery (adjusted β -1.49, 95% CI -2.75 to -0.53, P<0.001), and abdominal surgery (adjusted β -0.322, 95% CI -0.557 to -0.16, P<0.001), detailed in Table 6. However, no significant mediating effect was found in the adjusted subgroup analysis of thoracic paravertebral block or thoracic surgery, which may be attributed to insufficient statistical power due to limited sample size (n=482 and 516, respectively).

[0070] Propensity score matching achieved covariate balance between baseline and intraoperative variables, as shown in Table 7. In the matched cohort, the association between PNB and reduced hospital stay remained statistically significant, mediated by postoperative pulmonary complications (adjusted β -0.173%, 95% CI -0.317% to -0.06%, P < 0.001; as shown in Table 8).

[0071] Table 5. Decomposition of the effect of PNB on postoperative hospital stay mediated by postoperative pulmonary complications (PPC).

[0072] PPC = Postoperative pulmonary complications; CI = Confidence interval. * indicates P < 0.05. Bold values ​​indicate statistically significant results.

[0073] a Adjusted variables: age, sex, BMI, CCI, COPD, hemiplegia, asthma, long-term smoking, hemoglobin, alanine aminotransferase, albumin, ASA classification, hs-cTnI, surgical site (lung / kidney / prostate vs. gastrointestinal tract), surgical method (minimally invasive vs. open), operation time, ICU admission, reoperation, sepsis, unplanned ICU transfer from ward, year of surgery, and analgesia supplementation within 72 hours postoperatively.

[0074] Note: The adjusted indirect effect β=-0.278 (mediation rate 27.9%) is the calculation basis for the full sample default λ=0.279 in the patent claims; the total effect β=-0.987 and the direct effect β=-0.709 are the core parameters for evaluating the effect of PNB intervention.

[0075] Table 6. Mediation effect analysis of PPC on the relationship between PNB and hospitalization time in each subgroup.

[0076] PPC = Postoperative pulmonary complications; CI = Confidence interval; TAPB = Transversus abdominis plane block; TPVB = Thoracic paravertebral block. * indicates P < 0.05 (subgroups with statistically significant indirect effects are marked in bold).

[0077] Note: The sources of the invasiveness moderating coefficient λ are: minimally invasive surgery subgroup: λ=0.295 (mediation ratio point estimate 29.5%, n=1614); open surgery subgroup: λ=0.40 (mediation ratio point estimate 40.0%, n=226, with a wide confidence interval [12%, 104%], interpretation should be cautious).

[0078] Each subgroup underwent the same confounding adjustment as the main analysis (see notes in Table 5 for details). a ) Table 7 Baseline and intraoperative data after propensity score matching

[0079] a An ASD ≥ 0.121 is the unbalance criterion (based on Austin's formula: 1.96 × √[(n1 + n2) / (n1 × n2)]); all variables have an ASD < 0.121, indicating that the covariates are well balanced after matching.

[0080] ASD = Absolute Standardized Difference; PNB = Peripheral Nerve Block; the meanings of each abbreviation are the same as in Table 1. 1:1 nearest neighbor matching, caliper width = 0.2, sampling without replacement.

[0081] Table 8 Sensitivity analysis of the PPC mediation effect in the propensity score matched cohort.

[0082] The indirect effect β = -0.173 (mediation rate 18.9%) of the post-matching adjustment analysis is consistent with the main analysis results (β = -0.278, 27.9%), further verifying the mediating role of PPC and supporting the robustness of the algorithm of this invention.

[0083] This study demonstrates that in elderly patients undergoing major thoracic and abdominal surgery under general anesthesia, the association between preoperative nerve block (PNB) and shortened hospital stay was statistically significantly mediated by postoperative pulmonary complications. Despite improvements in postoperative pain awareness and management, a significant proportion of patients still experience severe postoperative pain, with 48.7% to 78.4% reporting moderate to severe pain in the early postoperative period. Severe pain is associated with an increased risk of postoperative complications, which are a major cause of prolonged hospital stays and increased medical costs. Although the incidence of postoperative pulmonary complications is relatively low, it remains a serious problem after non-cardiac surgery, leading to significant perioperative morbidity and mortality, with an incidence ranging from 15% to 37.5%. These complications require careful consideration in perioperative management.

[0084] Inadequate postoperative pain management is associated with an increased risk of postoperative complications and delayed time to walking and discharge. Relieving pain intensity through multimodal (primarily pharmacological) or epidural analgesia has been shown to shorten hospital stays, possibly by improving functional recovery and reducing postoperative complications. PNB (nerve block), as an important component of postoperative analgesia, may also have similar effects. Postoperative pain often impairs mobility and cough reflexes, thereby increasing the risk of respiratory complications such as atelectasis and pneumonia. Therefore, the observed mediating effect may be attributed to the enhanced analgesic effect provided by PNB.

[0085] Compared to general anesthesia alone, combined peripheral nerve block (PNB) and general anesthesia can shorten hospital stays and reduce the risk of postoperative pulmonary complications in elderly patients undergoing major thoracic and abdominal surgery. The effect of PNB in ​​reducing postoperative hospital stays is partly achieved by lowering the incidence of postoperative pulmonary complications. PNB and postoperative pulmonary complications should be considered core components of perioperative management strategies aimed at shortening postoperative hospital stays.

[0086] S2: Calculate the respiratory risk probability using the respiratory clinical data and predict the risk probability after PNB intervention; calculate the coupling factor using the respiratory risk probability and the risk probability after PNB intervention; In one embodiment, the respiratory risk probability is calculated using a second regression model (multivariate logistic regression model) to determine the probability of postoperative pulmonary complications based on respiratory clinical data. The respiratory clinical data includes: history of asthma (multivariate OR=4.745, P=0.016), presence of hyperlipidemia (multivariate OR=2.727, P=0.035), gender (multivariate OR=1.633, P=0.031), surgical site (lung surgery OR=8.075, kidney / ureter OR=0.229, P<0.05), pulmonary embolism (multivariate OR=71.531, P=0.004), and reoperation (multivariate OR=5.976, P=0.003); all OR values ​​are derived from a multivariate logistic regression model of 1840 elderly patients undergoing thoracic and abdominal surgery. The model only includes variables available preoperatively. Postoperative variables (deep vein thrombosis of the lower extremities, unplanned transfer from the ward to the ICU), although statistically significant in the multivariate analysis, were excluded from the predictive model because they were not available preoperatively. Surgical invasiveness parameters (open / minimally invasive) were not used as input variables for this logistic regression.

[0087] The risk probability after PNB intervention is simulated by the ratio of respiratory clinical data to the predicted respiratory risk after PNB intervention. The specific calculation formula is as follows: P_PNB = P_base × 0.395 / (1 Pbase + Pbase × 0.395) Where Pbasic is the basic probability of PPC occurrence output by the logistic regression model; 0.395 is the multifactor odds ratio of PNB (OR_PNB=0.395, 95% CI 0.261~0.597, P<0.001), derived from the above multifactor logistic regression model, indicating that the risk of PPC occurrence is reduced to 0.395 times the original risk after using PNB.

[0088] In one embodiment, the coupling factor calculation is obtained by first calculating the difference between the respiratory risk probability and the risk probability after PNB intervention, and then performing a weighted product on the difference.

[0089] The weights for the weighted product calculation are as follows: A respiratory clinical dataset of patients with pulmonary complications is obtained; this dataset is then input into a first regression prediction model (a multivariate generalized linear regression prediction model) to calculate the regression coefficients. The independent variable of the first regression prediction model is postoperative pulmonary complications, and the dependent variable is length of hospital stay. Specifically, after adjusting for confounding factors such as age, gender, BMI, CCI, COPD, hemiplegia, asthma, long-term smoking, hemoglobin, alanine aminotransferase (ALT), albumin, ASA classification, hs-cTnI, surgical site, surgical type, surgical duration, ICU admission, reoperation, sepsis, unplanned ICU transfer from the ward, year of surgery, and analgesia administered within 72 hours postoperatively, the non-standardized regression coefficient β for PPC is 4.634 days (95% CI 3.642–5.627, P < 0.001). This means that, all other things being equal, PPC extends the average length of hospital stay by 4.634 days.

[0090] S3: Calculate the invasiveness adjustment coefficient using the surgical invasiveness parameters; In one embodiment, the invasiveness moderating coefficient is dynamically selected based on the surgical invasiveness parameters of respiratory clinical data. The mediating proportion of the reduction in hospital stay due to prevention of pulmonary complications within the total reduction in hospital stay resulting from PNB intervention is calculated when any type of surgical invasiveness parameter is implemented. The surgical invasiveness parameters include open surgery and minimally invasive surgery. The mediating proportion is calculated using multivariate mediation analysis based on the Baron-Kenny framework and Bootstrap sampling (5000 iterations) in different surgical subgroups: the mediation proportion for the minimally invasive surgery subgroup (n=1614) is 29.5%, corresponding to λ=0.295; the mediation proportion for the open surgery subgroup (n=226) is 40.0%, corresponding to λ=0.40 (these are all point estimates with wide confidence intervals, see subgroup analysis table 6); when the surgical type cannot be clearly classified, λ is taken as the full sample default value of 0.279.

[0091] S4: Calculate the expected number of bed-saving days using the coupling factor and the invasive adjustment coefficient.

[0092] In one embodiment, the method further includes bed turnover early warning, which calculates the expected extension of hospital stay due to PPC by using the postoperative pulmonary complication PPC probability output by a multi-factor generalized linear regression model (β_PPC=4.634 days) and a logistic regression model; when patients are predicted to have a hospital stay of more than N days due to postoperative pulmonary complications (N is a natural number greater than 1, with a default N=3 days, which can be adjusted according to the actual needs of hospital bed management), they are marked as key patients and a mandatory preoperative nerve block list is generated.

[0093] In one specific embodiment, the patient's "respiratory vulnerability characteristics" and "surgical invasiveness parameters" are collected, and the calculation process is as follows: Step S1: Calculate the baseline respiratory risk probability.

[0094] Based on a multivariate logistic regression model (input variables: history of asthma [OR=4.75], hyperlipidemia [OR=2.73], gender [OR=1.63], reoperation [OR=5.98], pulmonary embolism [OR=71.53], surgical site), the baseline probability Pbaseline of postoperative pulmonary complications (PPC) was calculated. Note: Surgical invasiveness (open / minimally invasive) was not included in the logistic regression model of this step, but was only used for the selection of the λ coefficient in step S3.

[0095] Step S2 (Indicator Processing): Calculate the PPC-LOS coupling factor.

[0096] Algorithm: Coupling factor = (Respiratory risk - Post-PNB intervention risk) × 4.634.

[0097] Post-PNB intervention risk P_PNB = P_basic × 0.395 / (1 Pbase + Pbase × 0.395) 4.634 is the non-standardized regression coefficient (beta) for the independent variable "postoperative pulmonary complications (PPC)" in a multivariate linear regression model. After controlling for other factors such as age and type of surgery, once a patient develops postoperative pulmonary complications (PPC), their length of hospital stay (LOS) will be extended by an average of 4.634 days.

[0098] Step S3 (Hierarchical Weighting): Introduce an "invasive adjustment coefficient".

[0099] For open surgery, the invasiveness moderating coefficient is 0.40 (corresponding to the 40.0% mediating proportion point estimate, source: multivariate mediation analysis of the open surgery subgroup (n=226) in 1840 patients, Bootstrap 5000 iterations). For minimally invasive surgery, the invasiveness moderating coefficient is 0.295 (corresponding to the 29.5% mediating proportion point estimate, source: multivariate mediation analysis of the minimally invasive surgery subgroup n=1614). If the type of surgery is unclear, λ is set to the default value of 0.279 (estimated total median proportion of the whole sample).

[0100] The final calculation of "expected bed-saving days" = coupling factor / invasiveness adjustment coefficient.

[0101] Furthermore, the system outputs a "bed turnover warning": for patients whose hospital stay is expected to be extended by more than N days (default N=3 days) due to PPC, the system automatically marks them as "key concern" and generates a "mandatory preoperative nerve block request form".

[0102] The invasiveness moderating coefficient (λ) is a dimensionless correction parameter representing the proportion (mediating proportion) of the "hospital length reduction effect obtained through prevention of pulmonary complications (PPC)" in the "total hospital length reduction effect of PNB intervention" within a specific surgical invasiveness category (open or minimally invasive). Example: This invention calculates that PNB saves patients one day of hospital time due to "lung protection," but we know that "lung protection" only accounts for 40% of the total benefit of PNB. Therefore, by dividing by λ (0.4), the system can infer that the total benefit of PNB is actually a saving of 2.5 days (including other implicit benefits such as analgesia and early ambulation). The above λ values ​​are all point estimates based on retrospective cohorts, and the confidence intervals are wide when the sample size is limited (e.g., the open surgery subgroup n=226). It is recommended to validate and update these values ​​in larger sample studies.

[0103] The design of the coupling factor / invasiveness modulation coefficient overcomes the limitations of traditional assessments that only focus on overt complications (such as pneumonia). Especially in minimally invasive surgical settings, due to the small lambda coefficient (0.295), the algorithm can, through mathematical amplification, keenly capture the potential benefits of nerve blocks in non-respiratory pathways (such as analgesia and gastrointestinal function recovery). This allows the system to output hospitalization day predictions that are more consistent with the patient's actual recovery process than simple risk assessments, thereby guiding medical resources towards patients with significant 'hidden benefits'. The improved solution of this invention goes beyond the medical conclusion that 'PNB can reduce complications' and instead constructs a digital management system.

[0104] The present invention also discloses a computer program product or system, including a computer program that, when executed by a processor, implements the above-described method steps.

[0105] Figure 3 A schematic diagram of the perioperative bed turnover day prediction system based on postoperative complications provided in this embodiment of the invention specifically includes: Acquisition Unit: Acquires respiratory clinical data of hospitalized patients awaiting surgery; Coupling unit: Calculates respiratory risk probability using the respiratory clinical data and predicts risk probability after PNB intervention; calculates coupling factor using the respiratory risk probability and the risk probability after PNB intervention; Invasive Unit: Calculates the invasive adjustment coefficient using the aforementioned respiratory clinical data; Days Unit: Calculates the expected bed-saving days using the coupling factor and the invasive adjustment coefficient.

[0106] Figure 4 An embodiment of the present invention provides a schematic diagram of a computer device, specifically including: A memory and a processor; the memory is used to store program instructions; the processor is used to invoke the program instructions, when any of the above-mentioned methods for predicting the number of days of perioperative bed turnover based on postoperative complications are executed.

[0107] The present invention also discloses a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, is any of the above-described methods for predicting the number of days of perioperative bed turnover based on postoperative complications.

[0108] The verification results of this verification embodiment show that assigning inherent weights to indications can improve the performance of this method compared to the default settings. Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. In the several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for example, the division of units is merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separated; the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of this embodiment. Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. This program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0109] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0110] The computer device provided by the present invention has been described in detail above. For those skilled in the art, there will be changes in the specific implementation and application scope based on the ideas of the embodiments of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications, characterized in that, include: Acquire respiratory clinical data and surgical invasive parameters of hospitalized patients awaiting surgery. The respiratory clinical data shall include at least the following: history of asthma, presence of hyperlipidemia, gender, surgical site, presence of pulmonary embolism, and need for reoperation. The respiratory clinical data are used to calculate the respiratory risk probability and predict the risk probability after PNB intervention. The coupling factor is calculated using the respiratory risk probability and the risk probability after PNB intervention; The invasiveness adjustment coefficient is calculated using surgical invasiveness parameters; The expected number of bed-saving days is calculated using the coupling factor and the invasive adjustment coefficient. Bed turnover will be dynamically adjusted based on the expected number of days saved per bed.

2. The method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications according to claim 1, characterized in that, The invasiveness adjustment coefficient is dynamically selected based on the surgical invasiveness parameters of respiratory clinical data. It is calculated as the mediating proportion of the reduction in hospital stay achieved by preventing pulmonary complications in the total reduction in hospital stay brought about by PNB intervention when any type of surgical invasiveness parameter is implemented. The types of surgical invasiveness parameters include open surgery or minimally invasive surgery.

3. The method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications according to claim 1, characterized in that, The coupling factor is calculated by first calculating the difference between the respiratory risk probability and the risk probability after PNB intervention, and then performing a weighted product on the difference.

4. The method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications according to claim 3, characterized in that, The weights for the weighted product calculation are as follows: obtain the respiratory clinical dataset of patients with pulmonary complications; input the respiratory clinical dataset into the first regression prediction model to calculate the regression coefficients, where the independent variable of the regression prediction model is postoperative pulmonary complications and the dependent variable is the length of hospital stay.

5. The method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications according to claim 1, characterized in that, The respiratory risk probability is calculated using a second regression model to determine the probability of postoperative pulmonary complications based on respiratory clinical data. The respiratory clinical data includes a history of asthma, hyperlipidemia, gender, surgical site, pulmonary embolism, and whether reoperation is required.

6. The method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications according to claim 5, characterized in that, The risk probability after PNB intervention is simulated by the ratio of respiratory clinical data to the predicted respiratory risk after PNB intervention. The specific calculation formula is as follows: P_PNB = P_base × OR_PNB / (1 P-based + P-based × OR_PNB) Where P is the probability of respiratory risk; OR_PNB is the odds ratio corresponding to PNB intervention.

7. The method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications according to claim 1, characterized in that, The method also includes bed turnover early warning, and the respiratory risk probability is calculated by the first regression prediction model to predict the additional hospitalization days due to postoperative pulmonary complications; when patients are predicted to have a hospitalization duration of more than N days due to postoperative pulmonary complications, they are marked as key patients and a mandatory preoperative nerve block list is generated.

8. A computer program product comprising a computer program or instructions, characterized in that, The computer program or instructions are executed by the processor to implement the method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications as described in any one of claims 1-7.

9. A computer device comprising a memory, a processor, and a computer program or instructions stored in the memory, characterized in that, The computer program or instructions are executed by the processor to implement the method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, The computer program or instructions are executed by the processor to implement the method for predicting the number of days of optimized perioperative bed turnover based on postoperative complications as described in any one of claims 1-7.

Images