Prediction system and prediction method for postoperative delirium of crowds living in plateau for long time and storage medium

Abstract

The invention provides a prediction system and a prediction method for postoperative delirium of people living in plateau for a long time and a storage medium, and belongs to the technical field of artificial intelligence target detection. According to the method, dangerous factors of delirium occurring after general anesthesia of people living in the plateau for a long time are collected, and a prediction model of delirium occurring after general anesthesia is established according to the number of the dangerous factors obtained through multi-factor Logistic regression analysis and correlation coefficients of the dangerous factors. According to the method, a prediction scoring system for delirium and cognitive function impairment after general anesthesia of plateau patients is established, and quantitative reference is provided for clinicians; accurate intervention can be started before an operation, postoperative cognitive impairment is reduced to the maximum extent, plateau crowds are assisted to rapidly recover, and life and working posts are returned as soon as possible.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence target detection technology, specifically relating to a prediction system, prediction method, and storage medium for postoperative delirium in people living in high-altitude areas. Background Technology

[0002] Postoperative delirium (POD) is the most common neurological complication following anesthesia and surgery, clinically manifested as inattention, fluctuating levels of consciousness, and cognitive impairment, with an incidence rate as high as 21.1% to 54%. POD can lead to prolonged hospital stays, increased hospitalization costs, postoperative functional decline, delayed return to work, and increased mortality, and has become one of the most pressing public health issues to be addressed in the perioperative period.

[0003] Currently, the prevention and treatment strategies for post-operative hip dysplasia (POD) mainly revolve around four stages: risk stratification, risk reduction, early identification, and treatment. Since there is no specific treatment for POD that has already occurred, clinical efforts tend to focus on identifying high-risk individuals preoperatively to facilitate close monitoring and rapid intervention. Existing research suggests that factors such as renal impairment, preoperative comorbidities, prolonged surgical time, and advanced age are associated with increased POD risk; anesthesia regimens and the choice of certain medications are also considered to potentially influence its incidence. Based on these risk factors, various POD risk prediction scores have been proposed both domestically and internationally, such as a nine-item scoring system for hip fracture surgery, and a weighted scoring model that incorporates dozens of variables based on hospital information systems. The latter, after internal validation in a subset of surgical populations, showed a positive predictive value of approximately 70%.

[0004] However, the aforementioned scoring systems primarily model internal medicine patients or single surgical procedures, limiting their applicability to the broader surgical population. Furthermore, existing models are all built upon data from populations in plains areas, failing to consider the potential impact of high-altitude environments on physiological and postoperative neurocognitive outcomes. Long-term residents of high-altitude regions in my country may exhibit different pathophysiological characteristics compared to those in plains areas due to chronic hypoxia, yet there are currently no dedicated tools for predicting postoperative disease (POD) risk in this population. With improved transportation and economic development, a large number of long-term high-altitude residents are undergoing surgery in hospitals in plains areas. How to achieve preoperative risk stratification and early identification of POD in this special population has become a crucial gap that urgently needs to be filled in the field of perioperative medicine. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention aims to construct a prediction system and scoring system for postoperative delirium in long-term high-altitude populations through a prospective, cohort study. This will enable early identification of at-risk individuals and the provision of appropriate health guidance and early disease prevention, thereby reducing the incidence of cognitive impairment after anesthesia and surgery in the vast high-altitude population.

[0006] This invention provides a predictive system for postoperative delirium in people living at high altitudes, comprising: The data acquisition module is configured to collect multiple predictive factor information of the target patient, including: whether the patient has high altitude hemoglobinopathies (HAPC), the patient's usual altitude information, and the time information of the patient's altitude acclimatization time. The risk assessment module, connected to the data acquisition module, is configured to input the predictor information into the prediction model, which is a logistic regression model trained on a dataset containing plateau-specific risk factors, and is used to output the risk assessment result of the target patient developing postoperative delirium.

[0007] Furthermore, the predictive factors also include: age information, American Society of Anesthesiologists (ASA) classification information, body mass index (BMI) information, history of diabetes, and history of stroke.

[0008] Furthermore, the logistic regression model is calculated based on logistic regression of the following variables: age; American College of Anesthesiologists classification; body mass index; history of diabetes; history of stroke; history of high altitude hemoglobinemia; usual altitude; and time to acclimatization to high altitude.

[0009] Furthermore, the logistic regression model uses the following binary values ​​as input: age ≥ 65 years; American College of Anesthesiologists Class III; body mass index ≥ 30 kg / m²; history of diabetes; history of stroke; history of high altitude hemoglobinemia; habitual altitude > 4500m; high altitude acclimatization time ≤ 7 days.

[0010] Furthermore, the risk assessment results output by the risk assessment module include the probability value of the target patient developing postoperative delirium, and / or the risk level classified based on the probability value.

[0011] This invention also provides a method for predicting postoperative delirium in people living at high altitudes for extended periods, comprising the following steps: Obtain information on multiple predictive factors for the target patient, including at least: whether the patient has high altitude hemoglobinemia, the patient's usual altitude, and the time information of the patient's altitude acclimatization period. The acquired predictor information is input into the prediction model, which is a logistic regression model trained on a dataset containing plateau-specific risk factors. Based on the output of the prediction model, a risk assessment result for postoperative delirium in the target patient is generated.

[0012] Furthermore, the predictive factors also include: age information, American College of Anesthesiologists classification information, body mass index information, history of diabetes, and history of stroke.

[0013] Furthermore, the prediction model maps the predictor information to the probability of postoperative delirium through a logistic regression function.

[0014] Furthermore, the logistic regression model calculates the risk score using the following formula. F And further obtain the probability P : in, The variable weights for age, American College of Anesthesiologists classification, body mass index, diabetes, stroke, high altitude hemoglobinemia, usual altitude, and high altitude acclimatization time were 0.547, 0.500, 0.727, 1.101, 1.624, 1.744, 0.745, and 0.467, respectively. Choose from 0 or 1.

[0015] The present invention also provides a computer storage medium storing a computer program for implementing the above-mentioned prediction system for postoperative delirium in people living at high altitudes, or a computer program for implementing the above-mentioned prediction method for postoperative delirium in people living at high altitudes.

[0016] The present invention has achieved the following beneficial effects: (1) This invention has constructed the first prediction model for postoperative delirium (POD) after general anesthesia for people who have lived in high-altitude areas for a long time, filling a gap in this field; (2) This invention transforms complex physiological characteristics into a simple scoring system, forming a scoring system for the risk of postoperative cognitive impairment in high-altitude patients; (3) This invention can achieve the purpose of early identification of high-risk groups and early prevention of diseases, and provide clinical data support and practical prevention and control measures for the prevention of postoperative cognitive impairment in people living in high-altitude areas.

[0017] This invention collects risk factors for postoperative delirium after general anesthesia in individuals who have lived in high-altitude areas (≥2500 m) for ≥20 years. Based on the number of risk factors and their correlation coefficients obtained from multivariate logistic regression analysis, a predictive model for postoperative delirium after general anesthesia is established. This invention establishes a predictive scoring system for postoperative delirium and cognitive impairment in high-altitude patients, providing quantitative references for clinicians. Precise intervention can be initiated preoperatively to minimize postoperative cognitive impairment, facilitating rapid recovery and early return to life and work for high-altitude populations.

[0018] Obviously, based on the above description of the present invention, and according to common technical knowledge and conventional methods in the field, various other modifications, substitutions or alterations can be made without departing from the basic technical concept of the present invention.

[0019] The following detailed embodiments further illustrate the above-described content of the present invention. However, this should not be construed as limiting the scope of the present invention to the following embodiments. All technologies implemented based on the above-described content of the present invention fall within the scope of the present invention. Attached Figure Description

[0020] Figure 1 Construct a flowchart for the prediction model.

[0021] Figure 2 This is a flowchart for the diagnosis of postoperative delirium (POD).

[0022] Figure 3 This is a flowchart for the Numerical Rating Scale (NRS) diagnostic process for pain.

[0023] Figure 4 This is the ROC curve for this invention.

[0024] Figure 5 This is the AUC value of the ROC curve of this invention.

[0025] Figure 6 The ROC curve is for HAPC as a predictor.

[0026] Figure 7 The ROC curve is the one with the average altitude as a predictor.

[0027] Figure 8 ROC curve with altitude acclimatization time as a predictor.

[0028] Figure 9 The ROC curve is for a conventional factor as a predictor.

[0029] Figure 10 The ROC curve for the existing model.

[0030] Figure 11 ROC curves for external datasets.

[0031] Figure 12 This is a nomogram.

[0032] Figure 13 This is a Decision Curve Analysis (DCA) diagram. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] The term "long-term residence at high altitude" in this invention refers to continuous residence in areas at an altitude of ≥2500 meters for ≥20 years.

[0035] The "plateau deacclimation" described in this invention refers to a specific symptom in which long-term plateau residents and migrants who have become accustomed to the plateau environment undergo a series of morphological, structural, metabolic, and functional changes in order to adapt to the normoxic environment after moving down to the plains.

[0036] Example 1: Construction and validation of a predictive model for postoperative delirium in a population residing at high altitudes. This invention provides a predictive model, its construction method, and its application specifically for assessing the risk of postoperative delirium in people living in high-altitude areas after general anesthesia for non-cardiac surgery. "Living in high-altitude areas" refers to residing continuously at an altitude of ≥2500 meters for ≥20 years. The flowchart is as follows. Figure 1 As shown.

[0037] 1. Research subjects and data sources This invention selects patients who underwent non-cardiac surgery under elective intubation general anesthesia at the Chengdu Office of the People's Government of Tibet Autonomous Region from September 1, 2024 to August 31, 2025.

[0038] 2. Diagnostic and Assessment Criteria 2.1 Diagnosis of Postoperative Delirium (POD) The Confusion Assessment Method (CAM) was used for diagnosis. Diagnosis requires all of the following criteria: ① Feature 1: Acute alteration or fluctuation in consciousness: This mainly observes whether the patient's state of consciousness differs from baseline or whether there has been any fluctuation in the patient's state of consciousness over the past 24 hours. ② Feature 2: Attention deficit: This mainly observes whether the patient has difficulty concentrating or a weakening of attention shifting. ③ Feature 3: Altered clarity of consciousness: In addition to wakefulness, alertness, drowsiness, stupor, and coma are all considered alterations in consciousness. ④ Feature 4: Disordered thinking: With the patient awake and cooperative, four questions that can be answered with "yes" or "no," and a gesture of raising two fingers on both hands as instructed by the assessor, are used to evaluate disordered thinking. A total of 5 points are used; a correct score ≤3 points indicates disordered thinking. Diagnostic criteria: When both Feature 1 and Feature 2 are positive, and at least one of Feature 3 or Feature 4 is positive, a diagnosis of POD can be made (the specific assessment procedure is as follows...). Figure 2 (As shown).

[0039] 2.2 Diagnosis of Cognitive Impairment The Mini-mental State Examination (MMSE) was used for assessment. The total score is 30 points. Those with a score ≤24 points were diagnosed with cognitive impairment.

[0040] 2.3 Pain Assessment The Numerical Rating Scale (NRS) was used for pain assessment: scores ranged from 0 to 10, with 0 indicating no pain, 1-3 indicating mild pain (pain not affecting sleep), 4-6 indicating moderate pain (pain affecting sleep to some extent), 7-9 indicating severe pain (inability to fall asleep or waking up in pain during sleep), and 10 indicating excruciating pain. Figure 3 ).

[0041] 2.4 Anesthesia Risk Classification The assessment was conducted using the American Society of Anesthesiologists (ASA) anesthesia classification criteria, detailed in Table 1.

[0042] Table 1. ASA Classification Diagnostic Criteria 3. Research Plan 3.1 Inclusion criteria: ① Age 18-80 years; ② ASA classification I-III; ③ Elective non-cardiac surgery under general anesthesia with intubation; ④ Surgery time ≥1 hour; ⑤ Long-term residence (≥20 years) in high-altitude areas (≥2500 m); ⑥ Signed informed consent form.

[0043] 3.2 Exclusion criteria: ① History of active brain disease or neuropsychiatric disease, such as stroke, Alzheimer's disease, epilepsy, Parkinson's disease, mental illness, etc.; ② Cognitive impairment, i.e., MMSE score ≤ 24; ③ Severe hearing or vision impairment; ④ Pregnant or lactating women; ⑤ Those who have received general anesthesia in the past 30 days; ⑥ Those who have received other clinical studies during the same period.

[0044] 3.3 Case screening and data collection Patients meeting all inclusion criteria and without any exclusion criteria were screened, with the screening period not exceeding one week between the surgery date and the date of surgery. After obtaining informed consent, basic patient information was recorded on the Case Report Form (CRF). MMSE, CAM assessments, and routine laboratory tests were completed within 3 days preoperatively. From the postoperative day to the 7th postoperative day, CAM assessments were performed twice daily, once in the morning and once in the afternoon; if POD was diagnosed, assessments continued until the 7th postoperative day. All assessments were performed by uniformly trained nurses. A telephone follow-up was conducted 30 days postoperatively to assess CAM and MMSE.

[0045] 3.4 Anesthesia and Perioperative Management Combined intravenous and inhalation anesthesia was used, and the depth of anesthesia was maintained by monitoring the bispectral index (BIS) of the electroencephalogram (EEG) (40-55 during induction and 35-45 during maintenance). Muscle relaxation monitoring was used to maintain core body temperature ≥36℃. Multimodal analgesia was used postoperatively, and NRS scores were recorded twice daily.

[0046] 3.5 Database Establishment Establish a perioperative database and enter the following data: Preoperative data: ① General demographic data: height, weight, sex, ethnicity, education level, occupation, altitude of residence, length of time spent at high altitudes, time spent at low altitudes, past medical history, and past anesthesia / surgery history. ② Preoperative complete blood count, preoperative ferritin, biochemical and coagulation indicators. ③ Preoperative pain NRS score. ④ Preoperative CAM and MMSE scores.

[0047] Intraoperative data: ① Anesthesia data: ASA score, nerve block method, mechanical ventilation time, dosage of intravenous sedatives, sevoflurane, opioids and muscle relaxants, postoperative analgesia method, NRS score during PACU; ② Surgical data: Surgical name, surgical time, intraoperative bleeding, intraoperative fluid infusion volume, intraoperative blood transfusion volume, intraoperative urine output, circulatory and respiratory adverse events.

[0048] Postoperative data: ① Time to ambulation after surgery; ② CAM score and NRS score within 7 days after surgery; ③ Complication record; ④ Follow-up data after discharge: CAM and MMSE at 30 days after surgery, and patient's postoperative recovery status; ⑤ Other events that the investigators deemed necessary to record.

[0049] 4. Statistical Analysis Statistical analysis was performed using IBM SPSS Statistics 26.0 and R 4.5 software. Continuous variables conforming to a normal distribution were expressed as mean ± standard deviation and analyzed using t-tests; variables not normally distributed were expressed as median (interquartile range) and analyzed using Mann-Whitney U tests. Categorical variables were expressed as percentages and analyzed using chi-square tests or Fisher's exact tests. All tests were two-tailed. P <0.05 indicates a statistically significant difference.

[0050] 5. Study population characteristics A total of 2059 patients were screened during the study, and 1230 patients ultimately met the inclusion criteria for analysis. The incidence of POD was 21.1% (259 / 1230). The mean age of the entire group of patients was (51.3±14.8) years, of which 22.1% (272 / 1230) were ≥65 years old. Patient baseline characteristics are shown in Table 2.

[0051] Table 2 Baseline levels of included patients Key intraoperative indicators—including surgical type, anesthesia and operative duration, intraoperative blood transfusion, total fluid infusion volume, and blood loss—showed no statistically significant differences between the POD and non-POD groups (all indicators were comparable). P >0.05).

[0052] Table 3 Intraoperative information of included patients 6. Dataset Partitioning The models were randomly divided into a training set (n=1230) and an internal validation set (n=527) in a 7:3 ratio. The training set was used to select variables and build the model, while the validation set was used to evaluate the model's predictive performance.

[0053] 7. Predictor selection and model building In the training set, using the occurrence of POD as the outcome variable, the univariate analysis... P Variables with a value <0.1 or those considered clinically significant were included in the multivariate logistic regression analysis.

[0054] The results identified the following eight factors as independent risk factors for POD (Table 4): age ≥ 65 years, ASA class III, BMI ≥ 30 kg / m², diabetes, history of stroke, HAPC, habitual altitude > 4500 m, and altitude acclimatization time ≤ 7 days.

[0055] Table 4. Analysis of influencing factors of delirium after non-cardiac surgery in high-altitude patients. Note: ASA: American Society of Anesthesiologists classification of anesthesia. HAPC: High Altitude Hemoglobinopathies.

[0056] 8. Predictive model building, expression, and validation 8.1 Model Building and Scoring System A risk prediction model is constructed based on the results of multivariate logistic regression analysis. The weights of each variable are as follows: The total risk score is determined by the regression coefficients. F The calculation formula is as follows: in, The variable weights are respectively for age (≥65 years), ASA classification (Level III), BMI (obesity ≥30 kg / m²), diabetes, stroke, high altitude hemoglobinopathies (HAPC), usual altitude (>4500m), and high altitude acclimatization time (≤7 days).

[0057] For the values ​​of each variable, age: ≥65 years = 1, <65 years = 0; ASA classification: Class III = 1, Class I-II = 0; BMI obesity: ≥30 kg / m² = 1, <30 kg / m² = 0; diabetes: present = 1, absent = 0; stroke: present = 1, absent = 0; HAPC: present = 1, absent = 0; long-term altitude: >4500m = 1, ≤4500m = 0; altitude acclimatization time: ≤7 days = 1, >7 days = 0.

[0058] The logistic regression coefficients of the prediction model are shown in Table 5.

[0059] Table 5. Logistic Regression Coefficients of Predictive Models Based on Table 5, the total risk score ( F The formula for calculating ) is: F = (0.547 × age ≥ 65 years) + (0.500 × ASA III) + (0.727 × BMI ≥ 30) + (1.101 × diabetes) + (1.624 × history of stroke) + (1.744 × HAPC) + (0.745 × usual altitude > 4500m) + (0.467 × time to get used to clothes ≤ 7 days).

[0060] The probability of a patient developing POD ( P ) is calculated using the following formula: 8.2 Model Validation In the validation set, the area under the receiver operating characteristic (AUC) of the predictive model was 0.787 (95% CI: 0.755–0.819). P <0.001 indicates that the model has good predictive power. Figure 4 , Figure 5 ).

[0061] 9. Model Implementation System The prediction model described in this invention can be implemented through a system comprising the following modules: The data acquisition module is configured to collect multiple predictive factor information of the target patient, including: whether the patient has high altitude hemoglobinemia, the patient's usual altitude, and the time information of the patient's altitude acclimatization period. The risk assessment module, connected to the data acquisition module, is configured to input the predictor information into the prediction model, which is a logistic regression model trained on a dataset containing plateau-specific risk factors, and is used to output the risk assessment result of the target patient developing postoperative delirium.

[0062] The following experimental examples demonstrate the beneficial effects of the present invention.

[0063] Experimental Example 1: Optimal Analysis of the Prediction Model of the Invention 1. Model Performance Comparison Analysis To verify the superior predictive performance of the "specific factor combination" (i.e., the model containing the above 8 factors) determined in this invention compared with other factor combinations, multiple control models were constructed and compared on the training set (n=861). The settings and purposes of the control models are shown in Table 6 below.

[0064] Table 6. Comparison models to be constructed (all fitted on the training set n=1230) Control 1: Using HAPC as the predictor, the constructed prediction model has an AUC of 0.652, and the ROC curve is shown below. Figure 6 As shown; using the habitual altitude >4500m as the predictor, the constructed prediction model has an AUC of 0.588, and the ROC curve is shown in the figure. Figure 7 As shown; using altitude acclimatization time as a predictor, the constructed prediction model has an AUC of 0.542, and the ROC curve is shown in the figure. Figure 8 As shown.

[0065] Control 2: The predictive model that includes all conventional factors (i.e., the conventional model excluding HAPC, native altitude, and altitude acclimatization time) has an AUC of 0.690, and the ROC curve is shown below. Figure 9 As shown.

[0066] Control 3: Using an existing POD prediction model (reference “ROSSLER J, SHAH K, MEDELLIN S, et al. Development and validation of delirium prediction models for noncardiac surgery patients [J]. J Clin Anesth, 2024, 93: 111319.”, the prediction model is “Random Survival Forests: Delirium Prediction Model for Noncardiac Surgery Patients”), including conventional factors, the constructed prediction model has an AUC of 0.668, and the ROC curve is shown below. Figure 10 As shown.

[0067] The data from this invention and the various comparative models described above are compared as follows (Table 7): Table 7 Comparison of data between the present invention and related models (all fitted on training set n=1230) The results showed that the predictive model constructed in this invention, comprising eight specific factors, significantly outperformed all control models in terms of AUC (0.787), sensitivity (94.3%), and specificity (70.4%), and had the lowest information criterion (AIC, BIC) values, indicating that this combined model achieved optimal model fit and simplicity while maintaining high predictive accuracy. In particular, its sensitivity of 94.3% effectively identified the vast majority of high-risk patients for postoperative delirium, fully meeting the clinical need for priority prevention. This demonstrates that the "specific factor combination" selected in this invention produced a synergistic effect, achieving unexpected technical results superior to single plateau factors, conventional factor combinations, and existing general models.

[0068] 2. Model robustness verification To evaluate the generalization ability of the model, this invention underwent rigorous robustness verification.

[0069] (1) Internal validation The final model was applied to the internal validation set (n=369), and its AUC was 0.787 (95% CI: 0.755–0.819), which is highly consistent with the performance on the training set, indicating that the model did not show significant overfitting. Figure 4 ).

[0070] (2) External verification Further validation was conducted using a newly collected external patient cohort that was completely independent in time to provide a higher level of evidence for generalization ability.

[0071] Validation cohort: 523 long-term high-altitude patients who underwent non-cardiac surgery under general anesthesia were prospectively and continuously collected at our center between July 1, 2024 and May 31, 2025 (Chinese Clinical Trial Registry Number: ChiCTR2400085420). This cohort was not involved in any stage of model construction.

[0072] Validation method: The established final model formula (including specific coefficients) is directly applied to this cohort to calculate the predicted risk probability for each patient and compare it with the actual outcome.

[0073] Validation results: The model still exhibits excellent and stable predictive performance in this independent external cohort, with an AUC of 0.835 (95% CI: 0.794-0.877), a sensitivity of 0.981, and a specificity of 0.911. Figure 11 The comparison of internal and external validation results is shown in Table 8 below. This external validation set strongly demonstrates that the predictive model constructed in this invention is not an overfit to the development dataset, but rather possesses excellent generalization ability and temporal robustness, and can be effectively applied to future clinical patients, solving the technical challenge of the widespread application of models in this field. This further strengthens the inventiveness and practical value of the technical solution of this invention (Table 8).

[0074] Table 8. Data Comparison Between Internal and External Validation Sets 3. Model Visualization and Application To improve the clinical applicability and convenience of the model, a nomogram was generated using R 4.4.0 software based on the training set (n=1230) data. Figure 12 This graph transforms eight predictors into an intuitive graphical scoring tool, allowing clinicians to directly estimate the risk probability of an individual patient developing POD by simply accumulating points.

[0075] To further evaluate the net benefit of the model in clinical decision-making, decision curve analysis (DCA) was performed. Figure 13 As shown, within a wide range of threshold probabilities, the net benefit of using the predictive model of this invention for clinical decision-making is significantly higher than the strategies of "intervening in all patients" or "not intervening in any patients", demonstrating that the model has good clinical applicability.

[0076] In summary, this invention provides a predictive system, predictive construction method, and storage medium for postoperative delirium in long-term high-altitude populations. This invention collects risk factors for postoperative delirium after general anesthesia in long-term high-altitude populations and establishes a predictive model for postoperative delirium based on the number of risk factors and their correlation coefficients obtained from multivariate logistic regression analysis. This invention establishes a predictive scoring system for postoperative delirium and cognitive impairment in high-altitude patients, providing quantitative references for clinicians; precise intervention can be initiated preoperatively to minimize postoperative cognitive impairment, helping high-altitude populations recover quickly and return to life and work as soon as possible. Through detailed model performance comparisons and multi-level validation, it is fully demonstrated that the "specific factor combination" selected in this invention has unexpected and significant advantages in predictive efficacy compared to existing technologies, possessing outstanding substantive characteristics, significant progress, and clear industrial application value.

Claims

1. A predictive system for postoperative delirium in a population residing at high altitudes, characterized in that, include: The data acquisition module is configured to collect multiple predictive factor information of the target patient, including: whether the patient has high altitude hemoglobinemia, the patient's usual altitude, and the time information of the patient's altitude acclimatization period. The risk assessment module, connected to the data acquisition module, is configured to input the predictor information into the prediction model, which is a logistic regression model trained on a dataset containing plateau-specific risk factors, and is used to output the risk assessment result of the target patient developing postoperative delirium.

2. The prediction system according to claim 1, characterized in that, The predictive factors also include: age information, American College of Anesthesiologists classification information, body mass index information, history of diabetes, and history of stroke.

3. The prediction system according to claim 1, characterized in that, The logistic regression model is calculated based on logistic regression of the following variables: age; American College of Anesthesiologists classification; body mass index; history of diabetes; history of stroke; history of high altitude hemoglobinemia; usual altitude; and time to acclimatization to high altitude.

4. The prediction system according to claim 3, characterized in that, The logistic regression model takes the following binary values ​​as input: age ≥ 65 years; American College of Anesthesiologists Class III; body mass index ≥ 30 kg / m²; history of diabetes; history of stroke; history of high altitude hemoglobinemia; habitual altitude > 4500m; high altitude acclimatization time ≤ 7 days.

5. The prediction system according to any one of claims 1 to 4, characterized in that, The risk assessment results output by the risk assessment module include the probability value of the target patient developing postoperative delirium, and / or the risk level based on the probability value.

6. A method for predicting postoperative delirium in a population residing at high altitudes for extended periods, characterized in that, Includes the following steps: Obtain information on multiple predictive factors for the target patient, including at least: whether the patient has high altitude hemoglobinemia, the patient's usual altitude, and the time information of the patient's altitude acclimatization period. The acquired predictor information is input into the prediction model, which is a logistic regression model trained on a dataset containing plateau-specific risk factors. Based on the output of the prediction model, a risk assessment result for postoperative delirium in the target patient is generated.

7. The prediction method according to claim 6, characterized in that, The predictive factors also include: age information, American College of Anesthesiologists classification information, body mass index information, history of diabetes, and history of stroke.

8. The prediction method according to claim 6, characterized in that, The prediction model uses a logistic regression function to map the predictor information to the probability of postoperative delirium.

9. The prediction method according to claim 8, characterized in that, The logistic regression model calculates the risk score using the following formula. F And further obtain the probability P : in, The variable weights for age, American College of Anesthesiologists classification, body mass index, diabetes, stroke, high altitude hemoglobinemia, usual altitude, and high altitude acclimatization time were 0.547, 0.500, 0.727, 1.101, 1.624, 1.744, 0.745, and 0.467, respectively. Choose from 0 or 1.

10. A computer storage medium, characterized in that, The storage medium stores a computer program for implementing the prediction system for postoperative delirium in long-term high-altitude populations as described in any one of claims 1 to 5, or a computer program for implementing the prediction method for postoperative delirium in long-term high-altitude populations as described in any one of claims 6 to 9.

Images