Construction method and prediction system of frailty risk prediction model for elderly hospitalized patients

Abstract

A method for constructing a frailty risk prediction model for elderly hospitalized patients and a prediction system relate to the field of medical health technology. The method selects a plurality of factors influencing frailty in elderly hospitalized patients as candidate factors, first uses a univariate analysis method to screen out statistically significant candidate factors as initial screening factors; then constructs a multivariate logistic regression model as a frailty risk prediction model, using whether the elderly hospitalized patient experiences frailty as the dependent variable and the screened initial screening factors as independent variables. The frailty risk prediction model is used to screen out statistically significant initial screening factors as frailty factors, and the screened frailty factors are set as input factors of the model; the frailty risk prediction model is used to analyze the data of each frailty factor of the subject, and outputs a prediction result of whether the subject will experience frailty based on the analysis results. The method and system of the present invention are used to predict frailty risk in elderly hospitalized patients.

Description

Technical Field

[0001] The present invention relates to medical health technology, and in particular to a method for constructing a frailty risk prediction model and a prediction system for elderly hospitalized patients. Background Art

[0002] Frailty (senile frailty syndrome) is a common clinical syndrome among the elderly, characterized by decreased physiological reserve and functional impairment, which makes individuals vulnerable to external stimuli and increases the risk of adverse health outcomes (such as disability, hospitalization, and death). Frailty is more common in elderly hospitalized patients due to various factors, including illness and environmental factors. Progressive frailty can worsen the condition, prolong hospitalization, and increase medical costs. Therefore, accurately predicting the risk of frailty in elderly hospitalized patients is crucial for timely intervention and improved prognosis.

[0003] While current research on frailty in the elderly has revealed some relevant risk factors, such as age, gender, and chronic diseases, most use traditional logistic regression methods, which can only indicate the strength of risk factors and cannot combine multiple factors into an intuitive and convenient risk prediction model, making it difficult to directly apply in clinical practice. Furthermore, existing prediction models are mostly targeted at community-dwelling elderly people. Frailty risk prediction models for the specific population of elderly hospitalized patients are relatively rare, and they lack consideration of factors unique to hospitalized patients (such as a history of stress, polypharmacy, and assisted walking devices), failing to meet clinical needs. Summary of the Invention

[0004] In response to the defects existing in the above-mentioned prior art, the technical problem to be solved by the present invention is to provide a method for constructing a frailty risk prediction model and a prediction system for elderly hospitalized patients, which incorporates the unique factors of elderly hospitalized patients and can provide scientific and convenient frailty risk assessment for elderly hospitalized patients.

[0005] In order to solve the above technical problems, the present invention provides a method for constructing a frailty risk prediction model for elderly hospitalized patients, which specifically comprises the following steps:

[0006] 1) Select multiple factors that affect frailty in elderly hospitalized patients and define them as candidate factors;

[0007] Obtain sample data of multiple elderly hospitalized patients containing various candidate factors. Take whether the elderly hospitalized patients develop frailty as the outcome event. Use univariate analysis to analyze the sample data of each elderly hospitalized patient, and screen out statistically significant candidate factors to define as initial screening factors.

[0008] 2) Construct a multivariate logistic regression model as a frailty risk prediction model, with the occurrence of frailty in elderly hospitalized patients as the dependent variable and the initial screening factors as independent variables. The frailty risk prediction model was used to analyze the sample data of elderly hospitalized patients, and statistically significant initial screening factors were screened out and defined as frailty factors. There are a total of six frailty factors: stress history, walking device, polypharmacy, PSQI, MNA, and SAS.

[0009] The stress history factor is used to indicate whether the subject has experienced stressful events within a year;

[0010] The walking device factor is used to indicate whether the subject uses an assistive walking device;

[0011] The polypharmacy factor is used to indicate whether the subject uses at least 5 prescription drugs at the same time;

[0012] The PSQI factor is used to indicate the subject's sleep quality, the MNA factor is used to indicate the subject's nutritional status, and the SAS factor is used to indicate the subject's anxiety level;

[0013] 3) The outcome event of the frailty risk prediction model is set as whether the elderly hospitalized patients develop frailty, and the input factors of the frailty risk prediction model are set as the various frailty factors screened in step 2).

[0014] The present invention provides a method for predicting frailty risk in elderly hospitalized patients, which comprises the following specific steps:

[0015] The frailty risk prediction model constructed using the above method is used to analyze the frailty factor data of the subject and output a prediction result of whether the subject will develop frailty based on the analysis results;

[0016] There are 6 frailty factors, namely stress history factor, walking device factor, multiple medication factor, PSQI factor, MNA factor, and SAS factor.

[0017] Furthermore, the frailty risk prediction model analyzes the received data on various frailty factors and constructs a frailty risk nomogram for the subject based on the analysis results.

[0018] The present invention provides a system for predicting frailty risk in elderly hospitalized patients, comprising:

[0019] A variable input module is used to collect data on various frailty factors of the subjects, including stress history factors, walking equipment factors, multiple medication factors, PSQI factors, MNA factors, and SAS factors;

[0020] The frailty prediction module has a built-in frailty risk prediction model constructed by the above method. The frailty risk prediction model is used to analyze the various frailty factor data of the subjects collected by the variable input module, and output a prediction result of whether the subjects will suffer from frailty based on the analysis results.

[0021] Furthermore, it also includes a nomogram construction module, which is used to construct a frailty risk nomogram based on the analysis results of the frailty risk prediction model on the various frailty factor data of the subject, and output the constructed frailty risk nomogram to a display device for display.

[0022] The present invention provides a method for constructing a frailty risk prediction model for elderly hospitalized patients, a prediction method, and a prediction system. The method uses univariate analysis combined with a stepwise regression method to screen out specific factors for elderly hospitalized patients and incorporates them into a multivariate logistic regression model to predict the frailty risk of elderly hospitalized patients. This method can provide scientific and convenient frailty risk assessment for elderly hospitalized patients, helping clinicians to identify high-risk groups early and formulate intervention strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a frailty risk nomogram constructed using the six frailty factor data of a subject in the frailty risk prediction method for elderly hospitalized patients in an embodiment of the present invention;

[0024] Figure 2 The ROC curve diagram drawn by inputting the sample data of the modeling group into the frailty risk prediction model in an embodiment of the present invention;

[0025] Figure 3 The ROC curve diagram drawn by inputting the validation group sample data into the frailty risk prediction model according to an embodiment of the present invention;

[0026] Figure 4 This is a calibration curve diagram drawn by inputting the validation group sample data into the frailty risk prediction model in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The following is a further detailed description of an embodiment of the present invention in conjunction with the accompanying drawings, but this embodiment is not intended to limit the present invention. All similar structures and similar variations of the present invention should be included in the scope of protection of the present invention. The semicolons in the present invention represent the relationship of and, and the English letters in the present invention are case-sensitive.

[0028] The embodiment of the present invention provides a method for constructing a frailty risk prediction model for elderly hospitalized patients, and the specific steps are as follows:

[0029] 1) Select multiple factors that affect frailty in elderly hospitalized patients and define them as candidate factors;

[0030] Obtain sample data from multiple elderly hospitalized patients containing various candidate factors. Using whether the elderly hospitalized patients developed frailty as the outcome event, the sample data of each elderly hospitalized patient were analyzed using univariate analysis. Statistically significant candidate factors (P value less than 0.05) were screened out and defined as initial screening factors.

[0031] 2) Construct a multivariate logistic regression model as a frailty risk prediction model, with whether elderly hospitalized patients develop frailty as the dependent variable and each initial screening factor as the independent variable. Analyze the sample data of each elderly hospitalized patient using the frailty risk prediction model, and screen out statistically significant initial screening factors (P value less than 0.05) as frailty factors;

[0032] 3) The outcome event of the frailty risk prediction model is set as whether the elderly hospitalized patients develop frailty, and the input factors of the frailty risk prediction model are set as the various frailty factors screened in step 2).

[0033] The embodiments of the present invention collect factors that affect frailty in elderly hospitalized patients based on literature reports and clinical consensus. On the basis of combining a large amount of literature on frailty risk factors, after consulting geriatric medicine experts and geriatric nursing experts, discussions and combined with implementation feasibility, potential risk factors covering sociodemographics, lifestyle, disease medication, geriatric syndromes and psychosocial factors are selected as candidate factors that affect frailty in elderly hospitalized patients.

[0034] Sociodemographic factors include the patient's age, gender, marital status, living status, and income and expenditure status; among them, age is divided into four levels: 60-69 years old, 70-79 years old, 80-89 years old, and ≥90 years old; marital status is divided into four levels: single, married, widowed, and divorced; living status is divided into five levels: living alone, nursing home, living with children, living with spouse, and living with spouse and children; income and expenditure status is divided into three levels: low income, middle income, and high income. In this embodiment, the income and expenditure status is divided according to monthly income, and the currency unit is RMB, among which: monthly income ≤ 3,000 yuan is divided into the low income level, 3,000 yuan < monthly income < 6,000 yuan is divided into the middle income level, and monthly income ≥ 6,000 yuan is divided into the high income level;

[0035] Lifestyle factors included the patient's smoking history (yes or no), alcohol consumption history (yes or no);

[0036] Disease medication factors include multiple medication factors, stress history, BMI, walking device factors, fall history, surgical history, and disease type, among which:

[0037] Polypharmacy was determined by the patient's self-report of prescription medications used in the past year. Taking five or more prescription medications at the same time was defined as polypharmacy, and vice versa was defined as no polypharmacy.

[0038] Stress history is determined by the patient's self-report of whether they have experienced stressful events in the past year, with a value of yes or no. A stressful event refers to an unexpected event that can trigger a strong psychological or physiological reaction in the subject. The definition is manually pre-defined. In this embodiment, widowhood, patients receiving medical treatment due to illness, and patients being hospitalized due to illness are defined as stressful events.

[0039] BMI value is equal to weight (kg) / height (cm), BMI < 18.5 is defined as underweight, 18.5 ≤ BMI < 24 is defined as normal weight, 24 ≤ BMI < 28 is defined as overweight, and BMI ≥ 28 is defined as obesity;

[0040] The walking device factor is used to indicate whether the patient uses an assistive walking device (such as a cane or walker), with a value of yes or no;

[0041] Fall history is used to indicate whether the patient has had a fall event in the past year, with a value of yes or no. A fall event refers to the patient falling to the ground involuntarily without external force;

[0042] Surgical history was used to indicate whether the patient had undergone surgical operation in the past year, with a value of yes or no;

[0043] Disease type is used to indicate the number of diseases the patient currently suffers from;

[0044] Geriatric syndrome factors include PSQI (Pittsburgh Sleep Quality Index), ADL (Activities of Daily Living), and MNA (Mini Nutritional Assessment Brief);

[0045] PSQI is a self-assessment scale for sleep quality compiled by Buysee et al. in 1989. It includes 19 self-assessment items and 5 other-assessment items. Each component is scored from 0 to 3 points. The total score is the sum of the 7 component scores, ranging from 0 to 21 points. A PSQI score of > 7 indicates sleep problems.

[0046] The Barthel index scale was used to assess the ADL (activities of daily living) ability of elderly hospitalized patients. The Barthel index was scored on a scale of 0-100, with a score of 100 indicating that the patient had good basic ADL function and could control urination and defecation, eat independently, dress, transfer from bed to chair, bathe, walk, and go up and down stairs without help from others. A score of 0 indicated very poor function, with no independent ability and requiring help from others for all daily activities. The Barthel index score was divided into three levels: good, moderate, and poor. A Barthel index score >60 was defined as good ADL, with mild functional impairment, and the patient could complete some ADL independently but required some help. A Barthel index score of 41 ≤ ≤60 was defined as moderate ADL, with moderate functional impairment, and requiring great help to complete ADL. A Barthel index score ≤40 was defined as poor ADL, with severe functional impairment, and the patient could not complete most ADL or required help from others.

[0047] MNA includes six items: changes in food intake in the past three months, changes in weight in the past three months, activity level, acute illness or psychological trauma in the past three months, mental and psychological problems, and body mass index. The total score is 14 points. A score greater than or equal to 11 indicates normal nutritional status, and a score less than 11 indicates malnutrition.

[0048] Psychosocial factors include SAS (Self-Rating Anxiety Scale) and SSRS (Social Support Rating Scale);

[0049] SAS is a self-assessment scale compiled by Zung in 1976. The scale consists of 20 self-assessment items reflecting subjective anxiety feelings, including 15 positive items and 5 negative items (items 5, 9, 13, 17, and 19). Patients answer "never or rarely", "a small part of the time", "quite a lot of the time", and "most or all of the time" according to the frequency of each item in the past week, with scores of 1-4 respectively. The negative items are scored from 4 to 1. The sum of the 20 items is the crude score, which is multiplied by 1.25 and rounded to the standard score. A standard score ≥50 points indicates the presence of anxiety symptoms.

[0050] SSRS was designed by Xiao Shuiyuan in 1986. It includes 10 items in three dimensions: objective support (3 items), subjective support (4 items) and utilization of social support (3 items). The assessment results can be analyzed according to the total score and the score of each dimension.

[0051] In this embodiment of the present invention, elderly inpatients who were admitted to Renji Hospital affiliated to Shanghai Jiao Tong University School of Medicine between March 2022 and April 2023 and who met the following conditions and voluntarily participated in the survey were selected to obtain sample data containing various candidate factors of the selected patients;

[0052] The inclusion criteria for the selected patients were: age ≥ 60 years, and bedridden patients with dementia, patients in the terminal stage of illness, patients who could not walk with assistive devices, and patients with mental illness were excluded.

[0053] A total of 725 patients were finally included, and these patients were divided into two groups: modeling group and validation group;

[0054] The modeling group included 525 patients, including 224 male patients, 363 frail patients, and a frailty positive rate of 69.14%;

[0055] The validation group included 200 patients, including 97 male patients, 142 frail patients, and a frailty positive rate of 71%;

[0056] There was no statistical difference in frailty detection between the modeling group and the validation group;

[0057] All research subjects in the examples of the present invention were approved by the ethics review committee and complied with the provisions of the national ethics review regulations.

[0058] Taking whether elderly hospitalized patients developed frailty as the outcome event, the candidate factors of the sample data of each patient in the modeling group were analyzed by univariate analysis. The univariate analysis used the X-ray diffraction analysis of SPSS 26.0 software. 2 PSQI (Pittsburgh Sleep Quality Index), ADL (Activities of Daily Living), MNA (Mini Nutritional Assessment), NRS (Numerical Rating Scale), SAS (Self-Rating Anxiety Scale), SSRS (Social Support Rating Scale) were tested using the z-test method, and other candidate factors were tested using the X-test method. 2 Inspection methods;

[0059] Table 1 shows the analysis results of the candidate factors of the sample data of each patient in the modeling group using the univariate analysis method. N in the table represents the number of people;

[0060] In the second column of data for the six variables PSQI, ADL, MNA, NRS, SAS, and SSRS, the data outside the brackets are the means of the non-frail patients in the sample, the left value in the brackets are the minimum values ​​of the non-frail patients in the sample, and the right value in the brackets are the maximum values ​​of the non-frail patients in the sample;

[0061] In the third column of data for the six variables PSQI, ADL, MNA, NRS, SAS, and SSRS, the data outside the brackets are the mean values ​​of the frail patients in the sample, the left value in the brackets are the minimum values ​​of the frail patients in the sample, and the right value in the brackets are the maximum values ​​of the frail patients in the sample;

[0062] The fourth column of data for the six variables, PSQI, ADL, MNA, NRS, SAS, and SSRS, is the z-value of the z-test;

[0063] In the data of other variables except PSQI, ADL, MNA, NRS, SAS, and SSRS, the data outside the brackets in the second column are the number of non-frail patients, and the data in the brackets are the proportion of the number of the variable in the non-frail patients in the sample. The data outside the brackets in the third column are the number of frail patients, and the data in the brackets are the proportion of the number of the variable in the frail patients in the sample. The data in the fourth column are X 2 Test X 2 value;

[0064] As can be seen from Table 1 , age, gender, stress history, marital status, fall history, surgical history, disease type, walking device, polypharmacy, PSQI, ADL, MNA, NRS, SAS, and SSRS are candidate factors with statistical significance ( P value less than 0.05), so these candidate factors were defined as primary screening factors;

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071] After screening out statistically significant initial screening factors through univariate analysis, a multivariate logistic regression model was constructed as a frailty risk prediction model. The frailty risk prediction model was then used to analyze the initial screening factors of each patient sample data in the modeling group, with whether the elderly hospitalized patients developed frailty as the dependent variable and each initial screening factor as the independent variable.

[0072] The results showed that age, sex, marital status, history of falls, history of surgery, type of illness, ADL, NRS, and SSRS were excluded in the stepwise backward selection of variables;

[0073] Finally, six independent risk factors were screened out (the AIC value of the regression equation was the smallest at this time), namely, stress history (OR=1.878), walking device (OR=2.476), polypharmacy (OR=2.541), PSQI (OR=1.041), MNA (OR=0.888), and SAS (OR=1.065). The P values ​​of these six independent risk factors were all less than 0.05, and these six independent risk factors were defined as frailty factors.

[0074] In the multivariate logistic regression analysis, the values ​​of each discrete independent variable are shown in Table 2 , and the results of the multivariate logistic regression analysis are shown in Table 3 . In Table 3 , β is the multivariate logistic regression coefficient, se is the standard error, OR is the odds ratio, and logit_CI95 is the confidence interval.

[0075] Table 2. Assignment of discrete independent variables in multivariate logistic regression analysis

[0076] Independent variable Assignment age 60-69 years = 0; 70-79 years = 1; 80-89 years = 2; 90 years or older = 3 gender Male=0; Female=1 Marital status Single = 0; Married = 1; Widowed = 2; Divorced = 3 History of stress No = 0; Yes = 1 History of falls No = 0; Yes = 1 Surgical history No = 0; Yes = 1 Walking equipment No = 0; Yes = 1 Polypharmacy No = 0; Yes = 1

[0077] Table 3 Results of multivariate logistic regression analysis of frailty in elderly hospitalized patients in the modeling group (N=525)

[0078] variable β se Z value P-value OR logit_CI95 History of stress 0.630 0.216 2.924 0.003 1.878 1.233,2.874 Walking equipment 0.908 0.390 2.327 0.020 2.479 1.204,5.648 Polypharmacy 0.933 0.440 2.119 0.034 2.541 1.133,6.516 PSQI 0.041 0.019 2.136 0.033 1.041 1.003,1.081 MNA -0.119 0.050 -2.401 0.016 0.888 0.804,0.976 SAS 0.063 0.018 3.494 0.000 1.065 1.029,1.105

[0079] After screening out six statistically significant frailty factors using a frailty risk prediction model (multivariate logistic regression analysis), the outcome event of the frailty risk prediction model was set as whether elderly hospitalized patients developed frailty, and the input factors of the frailty risk prediction model were set as the six screened frailty factors.

[0080] The values ​​of the six frailty factors in the sample data of each patient in the modeling group were input into the frailty risk prediction model to draw the ROC curve. The sensitivity of the model was measured to be 73.4%, the specificity was 77.8%, the maximum Youden index was 0.411, the AUC value was 0.741, and the confidence interval (95% CI) was 0.697-0.785, indicating that the model has good predictive efficiency; Figure 2 The ROC curve is drawn by inputting the values ​​of the six frailty factors in the sample data of each patient in the modeling group into the frailty risk prediction model. Figure 2 The red line in the figure is the ROC curve of the frailty risk prediction model;

[0081] The frailty risk prediction model was externally validated using sample data from the validation group. The values ​​of the six frailty factors in the validation group were input into the frailty risk prediction model to plot the receiver operating characteristic (ROC) curve. The model showed a sensitivity of 71.9%, a specificity of 82.5%, a maximum Youden index of 0.444, an area under the curve (AUC) of 0.783, and a confidence interval (95% CI) of 0.707-0.859, indicating that the model has high sensitivity and can effectively identify high-risk individuals. Figure 3 The ROC curve is drawn by inputting the values ​​of the six frailty factors in the sample data of each patient in the validation group into the frailty risk prediction model. Figure 3 The red line in the figure is the ROC curve of the frailty risk prediction model;

[0082] To examine the reproducibility and generalizability of frailty risk prediction models, model performance must be evaluated after model establishment. The primary evaluation metric is calibration, which focuses on the accuracy of the model's absolute risk prediction values—that is, the degree of consistency between the model's predicted risk of an event and the actual risk of that event. The most common presentation method is a calibration plot, a scatter plot of the actual and predicted rates for all individuals. Models with good calibration have their scatter plots aligned along a 45-degree angle.

[0083] Figure 4 The calibration curve is drawn by inputting the values ​​of the six frailty factors in the sample data of each patient in the validation group into the frailty risk prediction model. Figure 4 The X-axis represents the predicted probability of the outcome event by the frailty risk prediction model, and the Y-axis represents the actual probability of the outcome event. The diagonal dotted line (Ideal) represents that the predicted probability is always equal to the actual probability under ideal conditions. The red solid line is the Apparent line, and the green solid line is the Bias-corrected line. It can be seen that the green solid line is always arranged along a 45° diagonal line near the diagonal dotted line, indicating that the model calibration is good.

[0084] The embodiment of the present invention provides a method for predicting frailty risk in elderly hospitalized patients, which specifically comprises the following steps: using the frailty risk prediction model constructed by the above method to analyze the data of various frailty factors of the subject, and outputting a prediction result of whether the subject will develop frailty based on the analysis results;

[0085] There are 6 frailty factors screened out by the above method, namely stress history factor, walking device factor, multiple medication factor, PSQI factor, MNA factor, and SAS factor.

[0086] In the frailty risk prediction method for elderly hospitalized patients provided in an embodiment of the present invention, the frailty risk prediction model analyzes the received data on various frailty factors, and then constructs a frailty risk nomogram for the subject based on the analysis results to make it visual and convenient for clinical use by medical staff. The constructed frailty risk nomogram includes 6 frailty factor score segments, as well as nomogram score segments, nomogram total score segments, and frailty risk segments.

[0087] Figure 1 This is a frailty risk nomogram constructed using data on six frailty factors of a particular subject. The stress history, presence of a walking device, polypharmacy, PSQI, MNA, and SAS in the figure are the six frailty factor score segments, the Points segment is the nomogram score segment, the Total Points segment is the nomogram total score segment, and the Risk segment is the frailty risk segment.

[0088] The length of the frailty factor score segment reflects the contribution of the frailty factor to the outcome event (frailty occurrence). The scale marks on the score segment of the nomogram are used to represent the score, and the scale marks on the total score segment of the nomogram are used to represent the total score. The scale marks on the frailty risk segment correspond to the scale marks on the total score segment of the nomogram and are used to represent the predicted frailty risk value.

[0089] Through the correspondence between the weight segments of the six frailty factors in the frailty risk nomogram and the scales marked on the score segments of the nomogram, the scores corresponding to each frailty factor can be calculated. The scores corresponding to each frailty factor are added together to obtain the total score. Based on the correspondence between the scales of the frailty risk segment and the scales of the total score segment of the nomogram, the risk prediction value corresponding to the calculated total score is obtained from the frailty risk segment.

[0090] The data on the six frailty factors in the embodiment of the present invention are easy to obtain. Clinicians can obtain predictive factors by asking short questions and using assessment scales, conduct rapid and accurate assessments, and make reasonable management decisions, thereby facilitating the promotion and application of frailty risk prediction models in clinical practice and improving the standardization of frailty risk health management for hospitalized elderly patients.

[0091] An embodiment of the present invention provides a system for predicting frailty risk in elderly hospitalized patients, comprising:

[0092] A variable input module is used to collect data on various frailty factors of the subjects, including stress history factors, walking equipment factors, multiple medication factors, PSQI factors, MNA factors, and SAS factors;

[0093] A frailty prediction module, which has a built-in frailty risk prediction model constructed using the above method. The frailty risk prediction model is used to analyze the frailty factor data of the subject collected by the variable input module and output a prediction result of whether the subject will become frail based on the analysis results;

[0094] The nomogram construction module is used to construct a frailty risk nomogram based on the analysis results of the frailty risk prediction model on the data of various frailty factors of the subject, and output the constructed frailty risk nomogram to a display device for display.

Claims

1. A method for constructing a frailty risk prediction model for elderly hospitalized patients, characterized by: The specific steps are as follows: 1) Select multiple factors that affect frailty in elderly hospitalized patients and define them as candidate factors; Obtain sample data of multiple elderly hospitalized patients containing various candidate factors. Take whether the elderly hospitalized patients develop frailty as the outcome event. Use univariate analysis to analyze the sample data of each elderly hospitalized patient, and screen out statistically significant candidate factors to define as initial screening factors. 2) Construct a multivariate logistic regression model as a frailty risk prediction model, with the occurrence of frailty in elderly hospitalized patients as the dependent variable and the initial screening factors as independent variables. The frailty risk prediction model was used to analyze the sample data of elderly hospitalized patients, and statistically significant initial screening factors were screened out and defined as frailty factors. There are a total of six frailty factors: stress history, walking device, polypharmacy, PSQI, MNA, and SAS. The stress history factor is used to indicate whether the subject has experienced stressful events within a year; The walking device factor is used to indicate whether the subject uses an assistive walking device; The polypharmacy factor is used to indicate whether the subject uses at least 5 prescription drugs at the same time; The PSQI factor is used to indicate the subject's sleep quality, the MNA factor is used to indicate the subject's nutritional status, and the SAS factor is used to indicate the subject's anxiety level; 3) The outcome event of the frailty risk prediction model is set as whether the elderly hospitalized patients develop frailty, and the input factors of the frailty risk prediction model are set as the various frailty factors screened in step 2).

2. A system for predicting frailty risk in elderly hospitalized patients, characterized by: include: A variable input module is used to collect data on various frailty factors of the subjects, including stress history factors, walking equipment factors, multiple medication factors, PSQI factors, MNA factors, and SAS factors; The frailty prediction module has a built-in frailty risk prediction model constructed by the method of claim 1, and the frailty risk prediction model is used to analyze the various frailty factor data of the subject collected by the variable input module, and output a prediction result of whether the subject will suffer from frailty based on the analysis result.

3. The system for predicting frailty risk in elderly hospitalized patients according to claim 2, characterized in that: It also includes a nomogram construction module, which is used to construct a frailty risk nomogram based on the analysis results of the frailty risk prediction model on the various frailty factor data of the subject, and output the constructed frailty risk nomogram to a display device for display.

Images