Method and system for predicting intestinal preparation failure risk of old hospitalized patient

Abstract

The invention relates to an old inpatient intestinal preparation failure risk prediction method and system, and the method comprises the steps: collecting the multi-dimensional electronic medical data of a patient from a hospital information system (HIS), a laboratory information system (LIS) and a pharmacy information system (PIS), carrying out the cleaning, missing value filling and standardization preprocessing of the original data, so as to construct a high-quality data set, and carrying out the prediction of the intestinal preparation failure risk of the old inpatient. Generating three core risk characteristics based on clinical knowledge: an age-weighted complication index, a specific drug use mark and a key physiological index anomaly mark; inputting the features into a preset rule engine, performing weighted calculation according to a fixed weight coefficient to obtain a comprehensive risk score, finally dividing patients into low, medium and high risk levels according to a risk threshold value determined by historical data, and outputting a result and storing the result into a database for clinical retrieval and early warning. The method can effectively evaluate the intestinal preparation failure risk of the elderly patient, and provides an objective basis for clinical intervention.

Description

Technical Field

[0001] This application relates to the field of medical information technology, and in particular to a method and system for predicting the risk of bowel preparation failure in elderly hospitalized patients. Background Technology

[0002] Colonoscopy is a crucial tool for diagnosing and treating colorectal diseases, and its effectiveness heavily relies on adequate bowel preparation. However, in clinical practice, the incidence of inadequate bowel preparation is significantly higher in elderly hospitalized patients, often leading to a series of adverse consequences such as discontinuation of the procedure, missed diagnoses, repeated examinations, and prolonged hospital stays. Traditionally, healthcare professionals have relied primarily on limited clinical experience and personal intuition to assess patient preparation risks, lacking objective and systematic assessment tools, resulting in a lack of targeted and timely preventative interventions.

[0003] Current risk assessment methods are mostly limited to isolated analyses of single or limited risk factors, such as focusing only on a history of constipation or the use of a specific medication. These methods fail to comprehensively consider the complex and interdependent factors that often coexist in elderly patients, such as multiple comorbidities, combination drug use, and declining physiological function. Therefore, their predictive ability is limited, their generalization is insufficient, and they are difficult to provide reliable risk stratification in complex real-world clinical settings.

[0004] With the development of hospital information technology, electronic medical record systems have accumulated massive amounts of clinical diagnosis and treatment data, which makes it possible to use data-driven methods for accurate prediction. However, how to effectively extract key features from this multi-dimensional, high-quality data and build a risk assessment framework that does not require complex algorithm training, is easy to deploy clinically, and has strong interpretability, so as to achieve early identification and precise intervention for high-risk patients, remains a significant challenge and a gap in current understanding. Summary of the Invention

[0005] To achieve the above objectives, this application provides the following technical solution: According to a first aspect of the present invention, the present invention claims protection for a method for predicting the risk of bowel preparation failure in elderly hospitalized patients, comprising the following steps: S1: Collect multi-dimensional electronic data of the target elderly inpatients from the hospital information system. The multi-dimensional electronic data includes the patient's basic information, clinical diagnosis information, medication records and laboratory test indicators. S2: Preprocess the multi-dimensional electronic data, including data cleaning, missing value imputation and data standardization, to form a standardized patient feature dataset; S3: Based on predefined feature engineering rules, extract key risk features related to bowel preparation failure from the standardized patient feature dataset. The key risk features include at least the number of age-adjusted comorbidities, specific drug use markers, and key physiological indicator abnormalities. S4: Input the key risk features into the pre-built risk prediction rule engine, which has built-in feature weights and risk thresholds derived from clinical knowledge base and historical data statistical analysis. S5: The risk prediction rule engine performs weighted calculations based on the key risk characteristics and their corresponding weights, and outputs a quantified comprehensive risk score; S6: Compare the comprehensive risk score with the preset risk level threshold, generate and output the risk level prediction result of the patient's bowel preparation failure, and the risk level includes at least low risk, medium risk and high risk.

[0006] Furthermore, step S1 also includes: The patient's basic information includes age, gender, and body mass index; The clinical diagnostic information includes diagnostic records for constipation, diabetes, cerebrovascular disease, and Parkinson's disease; The medication records include the use of narcotic analgesics, anticholinergic drugs, and tricyclic antidepressants within the past week; The laboratory test indicators include serum albumin and hemoglobin levels.

[0007] Further, step S3 involves extracting key risk features associated with bowel preparation failure from the standardized patient feature dataset, specifically including: S31: Calculate the age-adjusted number of comorbidities, which is the total number of chronic diseases in a predefined list that the patient has, including at least diabetes, chronic kidney disease, heart failure, and chronic obstructive pulmonary disease, and weight the total number according to the patient's age group. S32: Generate the specific drug use flag. If the patient's medication record contains records of the use of any one or more of the narcotic analgesics, anticholinergic drugs, or tricyclic antidepressants, then set the flag to true; otherwise, set it to false. S33: Generate the abnormal markers of the key physiological indicators by comparing the patient's serum albumin value with a first preset threshold. If the value is lower than the first preset threshold, the patient is determined to have an abnormal nutritional status. At the same time, compare the patient's hemoglobin value with a second preset threshold. If the value is lower than the second preset threshold, the patient is determined to have an abnormal anemia status. The abnormal markers of the key physiological indicators comprehensively reflect the presence or absence of abnormal conditions.

[0008] Furthermore, the preprocessing of the multi-dimensional electronic data in step S2 specifically includes: S21: Data cleaning, identifying and removing logically incorrect values ​​or values ​​that exceed the physiologically reasonable range from the laboratory test indicators; S22: Missing value imputation. For missing serum albumin values, imputation is performed using the population mean of the patient's age group or the median of the normal range for the patient's gender. S23: Data standardization. The imputed serum albumin and hemoglobin values ​​are Z-score standardized to convert them into standardized values ​​with a mean of 0 and a standard deviation of 1, so as to facilitate subsequent feature integration and calculation.

[0009] Furthermore, the pre-built risk prediction rule engine described in step S4 has built-in feature weights that are fixed coefficients determined by performing univariate and multivariate statistical analyses on the key risk features of patients with failed and successful bowel preparation in the historical population.

[0010] Furthermore, the operational logic of the risk prediction rule engine is as follows: The overall risk score is calculated as follows: (number of comorbidities adjusted for age * W1) + (specific drug use markers * W2) + (abnormal markers of key physiological indicators * W3), where W1, W2, and W3 are fixed weights pre-assigned to each key risk characteristic.

[0011] Furthermore, the preset risk level threshold in step S6 includes a first threshold value and a second threshold value, wherein the comprehensive risk score is lower than the first threshold value corresponding to a low risk level, between the first threshold value and the second threshold value corresponding to a medium risk level, and higher than the second threshold value corresponding to a high risk level.

[0012] Furthermore, after outputting the risk level prediction result in step S6, step S7 is also included: The prediction results are associated with the corresponding patient identifiers and then written into the specific risk warning database table of the hospital information system.

[0013] Furthermore, the method also includes step S8: Authorized clinical terminals can read a list of high-risk patients from the specific risk warning database table and prioritize clinical intervention or develop personalized bowel preparation plans. The data processed by the method is limited to elderly hospitalized patients undergoing elective colonoscopy for the first time.

[0014] According to a second aspect of the present invention, the present invention claims protection for a system for predicting the risk of bowel preparation failure in elderly hospitalized patients, comprising: One or more processors; A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method for predicting the risk of bowel preparation failure in elderly hospitalized patients.

[0015] This invention relates to a method and system for predicting the risk of bowel preparation failure in elderly hospitalized patients. It collects multi-dimensional electronic medical data from patients' Hospital Information System (HIS), Laboratory Information System (LIS), and Pharmacy Information System (PIS). The raw data undergoes cleaning, missing value imputation, and standardization preprocessing to construct a high-quality dataset. Based on clinical knowledge, three core risk features are generated: an age-weighted comorbidity index, specific drug use markers, and key physiological indicator abnormalities. These features are input into a pre-set rule engine, and weighted calculations are performed according to fixed weight coefficients to obtain a comprehensive risk score. Finally, based on risk thresholds determined from historical data, patients are classified into low, medium, and high risk levels. The results are output and stored in a database for clinical review and early warning. This invention can effectively assess the risk of bowel preparation failure in elderly patients, providing an objective basis for clinical intervention. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the workflow of a method for predicting the risk of bowel preparation failure in elderly hospitalized patients, as claimed in this application. Figure 2 The flowchart of the key risk feature extraction process for the method for predicting the risk of bowel preparation failure in elderly hospitalized patients claimed in this application is shown in the embodiment of the present application. Figure 3 The flowchart shows a multi-dimensional electronic data preprocessing process for a method to predict the risk of bowel preparation failure in elderly hospitalized patients, as claimed in this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0018] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the figures). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0019] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0020] According to the first embodiment of the present invention, referring to Figure 1 This invention claims protection for a method for predicting the risk of bowel preparation failure in elderly hospitalized patients, comprising the following steps: S1: Collect multi-dimensional electronic data of the target elderly inpatients from the hospital information system. The multi-dimensional electronic data includes the patient's basic information, clinical diagnosis information, medication records and laboratory test indicators. S2: Preprocess the multi-dimensional electronic data, including data cleaning, missing value imputation and data standardization, to form a standardized patient feature dataset; S3: Based on predefined feature engineering rules, extract key risk features related to bowel preparation failure from the standardized patient feature dataset. The key risk features include at least the number of age-adjusted comorbidities, specific drug use markers, and key physiological indicator abnormalities. S4: Input the key risk features into the pre-built risk prediction rule engine, which has built-in feature weights and risk thresholds derived from clinical knowledge base and historical data statistical analysis. S5: The risk prediction rule engine performs weighted calculations based on the key risk characteristics and their corresponding weights, and outputs a quantified comprehensive risk score; S6: Compare the comprehensive risk score with the preset risk level threshold, generate and output the risk level prediction result of the patient's bowel preparation failure, and the risk level includes at least low risk, medium risk and high risk.

[0021] In this embodiment, each step further includes: Electronic data of target elderly inpatients were retrieved from the hospital's integrated information systems, including the Hospital Information System (HIS), Laboratory Information System (LIS), and Pharmacy Information System (PIS). The data collected included: (a) Basic patient information: age, gender, height, and weight recorded in the HIS (for calculating body mass index); (b) Clinical diagnostic information: all current and past diagnoses as indicated by the International Classification of Diseases, 10th Revision (ICD-10) were extracted from the inpatient medical record homepage and diagnostic records in the HIS; (c) Medication records: all long-term and temporary medical orders from one week prior to admission to the present were extracted from the PIS, with a focus on analgesics, nervous system drugs, and digestive system drugs; (d) Laboratory test results: the most recent serum albumin and hemoglobin test results after admission were extracted from the LIS. The collected raw electronic data is standardized. First, data cleaning is performed, and reasonable numerical ranges for each indicator are set based on medical common sense. Outliers outside the range are automatically identified and removed. Then, missing values ​​are handled. For missing laboratory indicators, they are filled in based on the central tendency of historical test results of the patient's age group and gender. Finally, data standardization is performed, converting numerical laboratory indicators into standard scores with a mean of zero and a standard deviation of one to eliminate the influence of dimensions. Based on clinical knowledge, the preprocessed data is transformed and combined to construct specific features for risk prediction. This step generates at least three core features: (i) a comorbidity index calculated based on the diagnosis list and weighted by age groups, (ii) a specific drug use binary marker generated based on medication records, and (iii) a physiological state abnormality binary marker generated based on laboratory indicators. The constructed specific feature values ​​are input one by one into a predefined computational logic device, which stores the fixed contribution coefficients of each feature item determined by clinical expert consensus and retrospective analysis of historical data. In the rule application step, the computational logic device performs a deterministic calculation, multiplying each input feature value by its corresponding fixed contribution coefficient, and summing all the product results to obtain a final comprehensive risk score. The calculated comprehensive risk score is compared with two pre-set cutoff values, which are determined by analyzing the risk score distribution of historical patient populations. Based on the score's range (below the first cutoff value, between the first and second cutoff values, or above the second cutoff value), the patient is classified as "low risk," "medium risk," or "high risk." Finally, this risk classification result is output to a computer monitor or written to a specified database field.

[0022] Furthermore, step S1 also includes: The patient's basic information includes age, gender, and body mass index; The clinical diagnostic information includes diagnostic records for constipation, diabetes, cerebrovascular disease, and Parkinson's disease; The medication records include the use of narcotic analgesics, anticholinergic drugs, and tricyclic antidepressants within the past week; The laboratory test indicators include serum albumin and hemoglobin levels.

[0023] In this embodiment, the patient's basic information includes the date of birth (used to calculate age), gender, and height and weight measured and recorded upon admission, directly obtained from the HIS patient master index; clinical diagnostic information includes diagnostic records extracted from the HIS diagnostic module for all ICD-10 code prefixes E10-E14 (diabetes), I60-I69 (cerebrovascular disease), G20 (Parkinson's disease), and K59.0 (constipation); medication records include drug usage records retrieved from the PIS for chemical drug names containing active ingredients such as "tramadol," "OxyContin," "Atropine," "Benztropine," "Amitriptyline," and "Doxepin"; and laboratory test indicators refer to the latest quantitative test results of "serum albumin" and "hemoglobin" read from the LIS test report.

[0024] Furthermore, referring to Figure 2 Step S3 involves extracting key risk features associated with bowel preparation failure from the standardized patient feature dataset, specifically including: S31: Calculate the age-adjusted number of comorbidities, which is the total number of chronic diseases in a predefined list that the patient has, including at least diabetes, chronic kidney disease, heart failure, and chronic obstructive pulmonary disease, and weight the total number according to the patient's age group. S32: Generate the specific drug use flag. If the patient's medication record contains records of the use of any one or more of the narcotic analgesics, anticholinergic drugs, or tricyclic antidepressants, then set the flag to true; otherwise, set it to false. S33: Generate the abnormal markers of the key physiological indicators by comparing the patient's serum albumin value with a first preset threshold. If the value is lower than the first preset threshold, the patient is determined to have an abnormal nutritional status. At the same time, compare the patient's hemoglobin value with a second preset threshold. If the value is lower than the second preset threshold, the patient is determined to have an abnormal anemia status. The abnormal markers of the key physiological indicators comprehensively reflect the presence or absence of abnormal conditions.

[0025] In this embodiment, each step further includes: First, a target chronic disease list is defined, which includes four diseases: diabetes, chronic kidney disease, heart failure, and chronic obstructive pulmonary disease. Then, the patient's diagnostic information is traversed to count the total number of diseases that match the list, resulting in a baseline comorbidity count. Next, the patient is categorized into three age groups based on their actual age: 65-74 years, 75-84 years, and 85 years and above. Finally, the baseline comorbidity count is multiplied by a preset adjustment coefficient corresponding to the age group to obtain the final feature value. Create a logical judgment process that sequentially checks whether any pre-specified drug has appeared in the patient's medication record. The list of specified drugs includes three major categories: narcotic analgesics, anticholinergic drugs, and tricyclic antidepressants. If a usage record is found in any drug category, regardless of dosage or frequency, the characteristic flag is set to "yes" or logical "true", otherwise it is set to "no" or logical "false". A key physiological indicator abnormality marker feature was constructed, which was formed by combining the results of two independent physiological indicator judgments. The first judgment was for serum albumin level, which was compared with a fixed clinical threshold (3.5 g / dL). If it was below this threshold, it was marked as abnormal. The second judgment was for hemoglobin level, which was compared with different clinical thresholds based on the patient's gender (13 g / dL for males and 12 g / dL for females). If it was below the corresponding threshold, it was marked as abnormal. Finally, if either of the above two judgments was abnormal, the feature marker was set to "yes" or logical "true", indicating that the patient had a significant physiological indicator abnormality.

[0026] Furthermore, referring to Figure 3 The preprocessing of the multi-dimensional electronic data in step S2 specifically includes: S21: Data cleaning, identifying and removing logically incorrect values ​​or values ​​that exceed the physiologically reasonable range from the laboratory test indicators; S22: Missing value imputation. For missing serum albumin values, imputation is performed using the population mean of the patient's age group or the median of the normal range for the patient's gender. S23: Data standardization. The imputed serum albumin and hemoglobin values ​​are Z-score standardized to convert them into standardized values ​​with a mean of 0 and a standard deviation of 1, so as to facilitate subsequent feature integration and calculation.

[0027] In this embodiment, each step further includes: In the data cleaning sub-step, reasonable upper and lower limits are set for serum albumin and hemoglobin, respectively; for serum albumin, the lower limit is 2.0 g / dL and the upper limit is 5.5 g / dL; for hemoglobin, the lower limit is 7.0 g / dL and the upper limit is 20.0 g / dL. The system automatically scans all collected test values ​​and identifies any values ​​below the lower limit or above the upper limit as erroneous data and isolates them from subsequent analysis. In the missing value imputation sub-step, patient records with missing serum albumin or hemoglobin test results are identified. For these missing items, a replacement value strategy is used for imputation. When imputing missing serum albumin values, the average serum albumin value of a group of patients of the same sex whose age difference is no more than 5 years from that patient is calculated and used as the value. When imputing missing hemoglobin values, the median value of the normal reference range of hemoglobin corresponding to the patient's sex is directly used. In the data standardization sub-step, the cleaned and imputed serum albumin and hemoglobin values ​​are transformed; the overall mean and standard deviation of these two indicators for all elderly patients in the hospital's historical database are calculated; then, the original test value of each patient is subtracted from the overall mean of the indicator and divided by the overall standard deviation of the indicator to obtain the standardized score of this indicator for that patient.

[0028] Furthermore, the pre-built risk prediction rule engine described in step S4 has built-in feature weights that are fixed coefficients determined by performing univariate and multivariate statistical analyses on the key risk features of patients with failed and successful bowel preparation in the historical population.

[0029] In this embodiment, the fixed contribution coefficients of each feature stored in the predefined computational logic device in step S4 are predetermined by the following method: collecting complete data from a historical patient cohort in which the actual effect of bowel preparation for each patient is known; analyzing the cohort data using statistical software to calculate the correlation strength between each specific feature constructed in S3 and the bowel preparation failure result; and assigning a specific value as its fixed contribution coefficient to each feature based on the magnitude of its correlation strength. Once determined, this coefficient is fixed in the device and will not be changed during routine prediction.

[0030] Furthermore, the operational logic of the risk prediction rule engine is as follows: The overall risk score is calculated as follows: (number of comorbidities adjusted for age * W1) + (specific drug use markers * W2) + (abnormal markers of key physiological indicators * W3), where W1, W2, and W3 are fixed weights pre-assigned to each key risk characteristic.

[0031] In this embodiment, the deterministic calculation performed by the computational logic device is as follows: the device receives three feature values ​​output from step S3; first, it multiplies the "age-adjusted number of comorbidities" feature value by a first fixed coefficient; second, it determines whether the "specific drug use marker" feature is "true," and if true, it adds a second fixed coefficient, otherwise it does not; third, it determines whether the "key physiological indicator abnormality marker" feature is "true," and if true, it adds a third fixed coefficient, otherwise it does not; finally, it adds the results of the first three steps, and the sum is the comprehensive risk score value.

[0032] Furthermore, the preset risk level threshold in step S6 includes a first threshold value and a second threshold value, wherein the comprehensive risk score is lower than the first threshold value corresponding to a low risk level, between the first threshold value and the second threshold value corresponding to a medium risk level, and higher than the second threshold value corresponding to a high risk level.

[0033] In this embodiment, the two pre-set boundary values ​​in step S6 are determined as follows: after constructing the historical patient queue, the comprehensive risk score value of each patient in the queue is calculated using a computing logic device; after sorting all scores from smallest to largest, the value near the 25th percentile is selected as the first boundary value to distinguish between low risk and medium risk; the value near the 75th percentile is selected as the second boundary value to distinguish between medium risk and high risk.

[0034] Furthermore, after outputting the risk level prediction result in step S6, step S7 is also included: The prediction results are associated with the corresponding patient identifiers and then written into the specific risk warning database table of the hospital information system.

[0035] In this embodiment, after outputting the risk level prediction result in step S6, the method further includes step S7: data storage step: the prediction result, including the patient's medical record number, prediction date and time, calculated comprehensive risk score, and final determined risk level (low / medium / high), is written as a new data record row into a specially created "Intestinal Preparation Risk Prediction Result" data table in the HIS database.

[0036] Furthermore, the method also includes step S8: Authorized clinical terminals can read a list of high-risk patients from the specific risk warning database table and prioritize clinical intervention or develop personalized bowel preparation plans. The data processed by the method is limited to elderly hospitalized patients undergoing elective colonoscopy for the first time.

[0037] In this embodiment, when a risk warning is issued, clinicians or nurses can query the "Gut Preparation Risk Prediction Results" data table through the software interface on their workstation computer. The software interface provides a filtering function, allowing users to quickly filter out a list of all patients marked as "high-risk". This list displays the basic identification information of the patients to remind medical staff to pay special attention to these high-risk patients and prioritize intervention measures.

[0038] Before performing data collection in step S1, there is also an initial filtering step S0: the method is triggered only for hospitalized patients who meet the following conditions: age greater than or equal to 65 years and whose medical plan includes an item to be performed called "colonoscopy".

[0039] According to a second embodiment of the present invention, the present invention claims protection for a system for predicting the risk of bowel preparation failure in elderly hospitalized patients, comprising: One or more processors; A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method for predicting the risk of bowel preparation failure in elderly hospitalized patients.

[0040] The following is a specific example: This embodiment was implemented in the gastroenterology ward of a large tertiary-level hospital to systematically predict the likelihood of bowel preparation failure in each elderly inpatient scheduled for colonoscopy. The hospital possesses a highly information-based infrastructure, including a Hospital Information System (HIS), a Laboratory Information System (LIS), and a Pharmacy Management System (PIS), which interact and integrate data through an intra-hospital secure network. The implementation of this method relies entirely on existing hardware servers and clinical terminal computers, achieving automated prediction through the development of dedicated software logic modules.

[0041] Detailed implementation steps: S0: Initial filtering and triggering; The system automatically performs a scanning task once a day at a set time. This task queries the list of hospitalized patients in the HIS database, first calculating the patient's current age based on the patient's date of birth field, and automatically filtering out all patient records that meet the age criteria for elderly patients.

[0042] The system then checks each elderly patient's current treatment plan and medical order list, precisely searching for items containing specific keywords. Once a colonoscopy is identified in a patient's plan, the system immediately marks that patient as the target and automatically initializes the risk prediction process, creating a separate analysis task queue for that patient.

[0043] S1: Multi-source data acquisition and extraction; The system extracts comprehensive electronic medical data of target patients from different information systems through predefined database interfaces and access permissions.

[0044] Data collected from the HIS: This involves retrieving the patient's date of birth, gender, and objectively measured height and weight at admission from the patient's basic information module. Simultaneously, the medical record management module is accessed to extract all entered diagnostic information, all coded according to the International Classification of Diseases (ICD).

[0045] Data collected from PIS: A query request is sent to the pharmacy management system to retrieve all medication usage records for the patient from a specific time window prior to admission to the present, including long-term and temporary prescriptions. The system has a built-in list of key drug categories, and the program automatically matches and filters the retrieved medications against this list.

[0046] Acquisition from LIS: Send a query request to the laboratory information system to locate the patient's most recent complete metabolomics and routine blood test report, and accurately read the quantitative test results of serum albumin and hemoglobin from it.

[0047] S2: Data preprocessing and normalization; Data cleaning sub-step: The system has a built-in knowledge base containing medical common sense rules, which defines the theoretically reasonable range of various physiological indicators. The program automatically compares the collected raw laboratory data with this range. Any outliers identified as significantly exceeding biologically possible limits (e.g., theoretically impossible extremely low hemoglobin values) are automatically marked and isolated by the system, temporarily stored in the pending review area, and not involved in subsequent calculations, while an error log is generated.

[0048] Missing value imputation sub-step: For patient records with missing data for test indicators, the system does not simply remove them. The program identifies the missing items and intelligently imputes them according to a preset strategy. For example, for missing serum albumin values, the system calculates the central tendency value of this indicator for similar populations from the historical database based on the patient's age group and gender, and uses this value to impute the missing data, ensuring the integrity of the dataset.

[0049] Data standardization sub-step: To address the issue of inconsistent dimensions among different test indicators, the program transforms the imputed raw values ​​of serum albumin and hemoglobin. This transformation is based on a large historical data pool, calculating the overall average level and dispersion statistics of each indicator for the elderly population across the entire hospital. Subsequently, each patient's raw value is converted into a standardized score that reflects its relative position to the hospital-wide average level.

[0050] S3: Construction and conversion of clinical features; The core of this step is to transform the preprocessed raw data into three quantifiable risk characteristics with clear clinical significance.

[0051] Constructing an age-adjusted comorbidity index: The system pre-defines a finite list of chronic diseases, including several clinically validated comorbidities associated with bowel dysfunction. The program parses the patient's diagnostic code, calculates the total number of diseases on the list, and obtains a baseline number of comorbidities. Subsequently, based on the patient's specific age and a predefined age bracket, the program selects a corresponding adjustment multiplier, weights the baseline number, and ultimately generates a composite feature value that comprehensively reflects disease burden and the degree of aging.

[0052] Constructing a drug use flag: The system pre-defines a list of key drugs, all of which may inhibit intestinal motility due to their pharmacological effects. The program scans the patient's medication records using a combination of string matching and classification code matching. If any drug from the list is found in the record, the binary feature flag is set to a positive state.

[0053] Constructing physiological abnormality markers: The program separately determines whether the patient's serum albumin and hemoglobin levels are below the lower limit of the clinical normal range corresponding to their gender. These two determinations are performed independently; if either indicator is found to be abnormal, the overall physiological abnormality marker is set to a positive state.

[0054] S4: Rule Engine Application; After preprocessing and feature construction, the three feature values ​​are passed to a pre-built rule engine module. This engine internally stores a fixed set of parameters determined by historical data analysis and clinical expert consensus, which includes the contribution coefficient for each feature. Once set, these coefficients remain fixed values ​​in routine predictions.

[0055] S5: Risk score calculation; After receiving the feature values, the rule engine performs a deterministic, linear weighted calculation process. It multiplies the age-adjusted comorbidity index feature value by its corresponding fixed coefficient to obtain the first product; it then determines the status of the medication use marker; if it is positive, it adds its corresponding fixed coefficient, otherwise it adds zero; similarly, it determines the status of the physiological abnormality marker; if it is positive, it adds its corresponding fixed coefficient, otherwise it adds zero. Finally, these three values ​​are summed to obtain a final comprehensive risk score.

[0056] S6: Risk Level Classification and Result Output; The calculated overall risk score will be sent to the classification module. This module has two preset thresholds, which are determined by percentile analysis of the risk score distribution of historical patient populations. The program compares the patient's score with these two thresholds and automatically classifies the patient into one of three categories: "low risk," "medium risk," or "high risk," based on the range in which the patient falls.

[0057] Ultimately, the prediction results (including risk score and grade category) will be linked to the patient's unique identifier (such as medical record number) and output in two ways: first, displayed in real time on the risk warning monitoring screen at the nurse's workstation; second, automatically written as a new record into a dedicated data table in the HIS central database.

[0058] S7: Prediction results storage; The system creates a dedicated table in the HIS database to store the results of this prediction method. This table contains the following fields: patient medical record number, prediction calculation date and time, calculated raw risk score, and final risk level (low, medium, high). Each time a prediction is completed, the program automatically performs a database insert operation to persistently save the result.

[0059] S8: Clinical risk warning; The clinician workstation integrates a customized software plugin interface. Authorized healthcare personnel can use this interface to query and access the aforementioned dedicated data tables at any time. The interface provides filtering and sorting functions, allowing users to generate a list of all currently hospitalized patients assessed as "high-risk" with a single click. This list clearly displays the patients' basic information, serving a significant early warning function and reminding healthcare personnel of the need for immediate and more proactive clinical interventions for these key populations.

[0060] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only 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 coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0061] Furthermore, the functional units in the various embodiments of this application 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. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0062] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A method for predicting the risk of bowel preparation failure in elderly hospitalized patients, characterized in that, Includes the following steps: S1: Collect multi-dimensional electronic data of the target elderly inpatients from the hospital information system. The multi-dimensional electronic data includes the patient's basic information, clinical diagnosis information, medication records and laboratory test indicators. S2: Preprocess the multi-dimensional electronic data, including data cleaning, missing value imputation and data standardization, to form a standardized patient feature dataset; S3: Based on predefined feature engineering rules, extract key risk features related to bowel preparation failure from the standardized patient feature dataset. The key risk features include at least the number of age-adjusted comorbidities, specific drug use markers, and key physiological indicator abnormalities. S4: Input the key risk features into the pre-built risk prediction rule engine, which has built-in feature weights and risk thresholds derived from clinical knowledge base and historical data statistical analysis. S5: The risk prediction rule engine performs weighted calculations based on the key risk characteristics and their corresponding weights, and outputs a quantified comprehensive risk score; S6: Compare the comprehensive risk score with the preset risk level threshold, generate and output the risk level prediction result of the patient's bowel preparation failure, and the risk level includes at least low risk, medium risk and high risk.

2. The method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to claim 1, characterized in that, Step S1 also includes: The patient's basic information includes age, gender, and body mass index; The clinical diagnostic information includes diagnostic records for constipation, diabetes, cerebrovascular disease, and Parkinson's disease; The medication records include the use of narcotic analgesics, anticholinergic drugs, and tricyclic antidepressants within the past week; The laboratory test indicators include serum albumin and hemoglobin levels.

3. The method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to claim 2, characterized in that, Step S3 involves extracting key risk features associated with bowel preparation failure from the standardized patient feature dataset, specifically including: S31: Calculate the age-adjusted number of comorbidities, which is the total number of chronic diseases in a predefined list that the patient has, including at least diabetes, chronic kidney disease, heart failure, and chronic obstructive pulmonary disease, and weight the total number according to the patient's age group. S32: Generate the specific drug use flag. If the patient's medication record contains records of the use of any one or more of the narcotic analgesics, anticholinergic drugs, or tricyclic antidepressants, then set the flag to true; otherwise, set it to false. S33: Generate the abnormal markers of the key physiological indicators by comparing the patient's serum albumin value with a first preset threshold. If the value is lower than the first preset threshold, the patient is determined to have an abnormal nutritional status. At the same time, compare the patient's hemoglobin value with a second preset threshold. If the value is lower than the second preset threshold, the patient is determined to have an abnormal anemia status. The abnormal markers of the key physiological indicators comprehensively reflect the presence or absence of abnormal conditions.

4. The method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to claim 1, characterized in that, The preprocessing of the multi-dimensional electronic data in step S2 specifically includes: S21: Data cleaning, identifying and removing logically incorrect values ​​or values ​​that exceed the physiologically reasonable range from the laboratory test indicators; S22: Missing value imputation. For missing serum albumin values, imputation is performed using the population mean of the patient's age group or the median of the normal range for the patient's gender. S23: Data standardization. The imputed serum albumin and hemoglobin values ​​are Z-score standardized to convert them into standardized values ​​with a mean of 0 and a standard deviation of 1, so as to facilitate subsequent feature integration and calculation.

5. The method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to claim 1, characterized in that, The pre-built risk prediction rule engine described in step S4 has built-in feature weights that are fixed coefficients determined by performing univariate and multivariate statistical analyses on the key risk features of patients with failed and successful bowel preparation in the historical population.

6. The method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to claim 5, characterized in that, The operational logic of the risk prediction rule engine is as follows: The overall risk score is calculated as follows: (number of comorbidities adjusted for age * W1) + (specific drug use markers * W2) + (abnormal markers of key physiological indicators * W3), where W1, W2, and W3 are fixed weights pre-assigned to each key risk characteristic.

7. The method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to claim 1, characterized in that, The preset risk level thresholds in step S6 include a first threshold and a second threshold. The comprehensive risk score below the first threshold corresponds to a low risk level, the score between the first threshold and the second threshold corresponds to a medium risk level, and the score above the second threshold corresponds to a high risk level.

8. The method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to claim 1, characterized in that, After outputting the risk level prediction result in step S6, step S7 is also included: The prediction results are associated with the corresponding patient identifiers and then written into the specific risk warning database table of the hospital information system.

9. A method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to claim 8, characterized in that, It also includes step S8: Authorized clinical terminals can read a list of high-risk patients from the specific risk warning database table and prioritize clinical intervention or develop personalized bowel preparation plans. The data processed by the method is limited to elderly hospitalized patients undergoing elective colonoscopy for the first time.

10. A system for predicting the risk of bowel preparation failure in elderly hospitalized patients, characterized in that, include: One or more processors; A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement a method for predicting the risk of bowel preparation failure in elderly hospitalized patients according to any one of claims 1 to 9.

Images