Personalized intelligent management system for cleaning preparation of intestinal tracts of old people

Abstract

The invention provides a personalized intelligent management system for intestinal tract cleaning preparation of old people, and the system comprises a personalized risk assessment module which is used for obtaining the multi-dimensional individual data of an old patient, and outputting the risk level of insufficient intestinal tract preparation through a machine learning prediction model; the dynamic scheme generation module is used for generating a personalized intestinal tract cleaning scheme for the patient according to the risk level; the real-time monitoring and feedback module is used for collecting compliance data, defecation characters and untoward effect information of a patient in the process of executing the cleaning scheme of the patient, and performing dynamic evaluation and risk early warning on the cleaning quality; the user interaction module is used for providing health education content, operation guidance and psychological support for patients and family members, receiving data input by the patients and the family members and providing response service; and the medical collaboration platform is used for carrying out data connection with the hospital information system. Through data-driven intelligent evaluation and personalized scheme generation, effective management of the whole intestinal preparation process of the elderly patient is realized.

Description

Technical Field

[0001] This invention relates to the field of intestinal cleansing preparation technology, and in particular to a personalized intelligent management system for intestinal cleansing preparation in the elderly. Background Technology

[0002] Bowel preparation refers to the crucial clinical procedure of clearing feces and residue from the colon before a colonoscopy through dietary control, laxatives, and other means to ensure a clear view and accurate diagnosis. This process typically involves starting a low-residue diet several days before the examination, switching to a liquid diet the day before the examination, and taking cleansing agents such as polyethylene glycol electrolyte powder at prescribed times and dosages until the stool becomes watery.

[0003] Currently, standardized procedures are widely used in clinical practice for bowel preparation before colonoscopy. Healthcare professionals typically prescribe fixed doses and types of cleansing agents for all patients according to standardized medication guidelines, and provide patients with general dietary and medication instructions via paper or simple electronic documents. Some medical institutions use basic electronic health record systems or simple mobile applications to record patients' preparation status, but these systems are mostly limited to information recording functions and lack truly intelligent decision support. Regarding risk assessment, clinicians mainly rely on personal experience and a limited number of clinical indicators for subjective judgment, and a systematic and quantitative assessment system has not yet been formed. Secondly, traditional preparation processes lack effective off-site monitoring methods, making it difficult for healthcare professionals to promptly grasp patients' compliance and adverse reactions, and thus unable to intervene effectively during the preparation process. Furthermore, elderly patients have difficulty understanding the complex preparation procedures and have poor compliance, while family members lack professional assistance and guidance tools. Summary of the Invention

[0004] This invention provides a personalized intelligent management system for bowel preparation in the elderly. Through data-driven intelligent assessment and personalized plan generation, it achieves effective management of the entire bowel preparation process for elderly patients, thereby improving bowel preparation quality, ensuring patient safety, optimizing medical resource allocation, and enhancing the patient experience. The technical solution is as follows: A personalized intelligent management system for bowel preparation in the elderly includes: The personalized risk assessment module is used to obtain multi-dimensional individual data of elderly patients and output the risk level of inadequate bowel preparation through machine learning prediction models. The dynamic protocol generation module is used to generate personalized bowel cleansing protocols for patients based on their risk level. The real-time monitoring and feedback module is used to collect data on patient compliance, bowel characteristics and adverse reactions during the patient's implementation of the cleaning protocol, and to dynamically assess the cleaning quality and provide risk warnings. The user interaction module is used to provide health education content, operation guidance and psychological support to patients and their families, receive data input by patients and their families and provide response services. The medical collaboration platform is used to connect with the hospital information system and provide medical staff with a remote monitoring view of the patient's readiness.

[0005] Furthermore, the step of outputting the risk level of inadequate gut preparation through a machine learning prediction model includes the following steps: Acquire the patient’s individual characteristic data, including demographic characteristics, bowel function status, comorbidities, and functional status assessment. The individual feature data is standardized and preprocessed to form the model input feature vector; The Boruta algorithm based on random forest was used to evaluate the importance of the input feature vector and screen out risk factors related to the quality of bowel preparation. The selected risk factors are input into a pre-trained support vector machine classification model to obtain the probability prediction value of inadequate gut preparation. Based on the probability threshold range of the obtained probability prediction values, patients are classified into low-risk, medium-risk, or high-risk levels.

[0006] Furthermore, the method of evaluating the importance of the input feature vector using the Boruta algorithm based on random forest includes the following steps: Obtain the input feature vector and create a randomly rearranged version of the original features as a shadow feature; The importance scores of the original features and shadow features are calculated using the random forest algorithm. Compare the importance of each original feature with the importance of its corresponding shadow feature; Features with significantly higher importance than shadow features are retained, while features with insignificant importance are removed.

[0007] Furthermore, the step of inputting the screened risk factors into a pre-trained support vector machine classification model includes the following steps: The k-fold cross-validation method is used to optimize the model parameters on the training set; The model's performance metrics, including accuracy, sensitivity, and specificity, were evaluated using an independent validation set. Test the model's generalization ability on unknown data using an external validation set.

[0008] Furthermore, the process of generating a personalized bowel cleansing plan for patients based on their risk level includes the following steps: Establish a solution knowledge base that includes standard solutions, enhanced solutions, and strengthened solutions; Based on the risk level output by the risk assessment module, select the corresponding basic solution template from the solution knowledge base; Based on the patient’s individual characteristics data, the parameters of the cleaning agent type, dosage and timing of medication in the basic protocol template are optimized. Based on the patient's examination appointment time and personal routine, a personalized medication schedule is generated; a complete cleaning plan including cleaning agent information, medication time, dietary guidance, and exercise suggestions is output.

[0009] Furthermore, the collection of patient compliance data, bowel characteristics, and adverse reaction information during the implementation of the cleaning protocol includes the following steps: The medication record interface records the actual dosage and time of each medication taken by the patient; Provides a graphical stool characteristics assessment tool to guide patients in recording their bowel movements using the Bristol Stool Characteristics Scale; The severity and timing of adverse reactions such as abdominal distension, abdominal pain, and nausea were collected from patients using structured forms. The system synchronizes patients' water intake records and exercise data in real time.

[0010] Furthermore, the dynamic assessment and risk warning of cleaning quality includes the following steps: Based on real-time collected compliance data, bowel characteristics, and adverse reaction information, a multi-dimensional assessment index was constructed. The various indicators are weighted and calculated using preset scoring rules to obtain a comprehensive cleaning quality score. When the overall score falls below the threshold or a serious adverse reaction is detected, a tiered early warning mechanism is activated. Based on the severity of the risk, corresponding levels of warning information are sent to both patients and medical staff.

[0011] Furthermore, the provision of health education content, operational guidance, and psychological support to patients and their families includes the following steps: Based on the patient's personalized cleaning plan, generate corresponding illustrated and video operation guides; At the corresponding time points during the implementation of the plan, automatic push notifications will be sent for diet and medication. It provides psychological counseling content, including relaxation training guidance and sharing of success stories; it also provides collaborative supervision functions, including progress tracking, reminders, and assistance guidance.

[0012] Furthermore, the user interaction module adopts a microservice architecture, which separates user management and message push functions into independent service units. Each service unit communicates and exchanges data through a defined API interface.

[0013] Furthermore, the SHAP interpretability algorithm is used to analyze the prediction results of the machine learning model. By calculating the Shapley value of each feature parameter, the contribution of risk factors to the prediction results is shown.

[0014] As can be seen from the above technical solutions, the present invention has the following advantages: This invention achieves data-driven precision medicine through the collaborative work of a personalized risk assessment module and a dynamic protocol generation module. It also uses machine learning technology to comprehensively analyze the multi-dimensional individual characteristics of patients, which can accurately identify high-risk patients with inadequate bowel preparation and generate customized cleansing protocols for them. This improves the pass rate of bowel preparation, provides reliable assurance for the accuracy and safety of colonoscopy, and effectively reduces the probability of missed diagnoses and repeated examinations.

[0015] In terms of optimizing patient experience and ensuring safety, the close integration of the real-time monitoring and feedback module with the user interaction module constructs a comprehensive management loop. The system can continuously monitor patient adherence during the several-day preparation process, promptly identifying deviations and adverse reactions, and initiating corresponding interventions through a tiered early warning mechanism. This proactive monitoring model significantly improves the safety of elderly patients, especially high-risk patients, during the outpatient preparation phase. Simultaneously, the system's age-friendly interactive design and psychological support functions effectively reduce patients' cognitive burden and anxiety, improving treatment adherence.

[0016] In terms of medical collaboration efficiency, the deep integration of the medical collaboration platform enables seamless connection between in-hospital and out-of-hospital medical processes. This platform allows medical staff to remotely monitor the preparedness of a large number of patients and precisely target medical resources to the high-risk groups most in need. This model not only significantly reduces the workload of medical staff and improves management efficiency but also builds a continuous medical service system, promoting the optimal allocation of medical resources.

[0017] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from an examination of the following, or may be learned from the practice of the invention. Attached Figure Description

[0018] Figure 1 This is a system block diagram provided for the present invention. Detailed Implementation

[0019] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0020] Example 1 The implementation method in this embodiment can be implemented in a system, on a server, or on a terminal; no specific limitation is made. The method in this application will be described below from the perspective of system implementation. As shown in the figure, a personalized intelligent management system for intestinal cleansing preparation for the elderly includes: The personalized risk assessment module is used to obtain multi-dimensional individual data of elderly patients and output the risk level of inadequate bowel preparation through machine learning prediction models. The dynamic protocol generation module is used to generate personalized bowel cleansing protocols for patients based on their risk level. The real-time monitoring and feedback module is used to collect data on patient compliance, bowel characteristics and adverse reactions during the patient's implementation of the cleaning protocol, and to dynamically assess the cleaning quality and provide risk warnings. The user interaction module is used to provide health education content, operation guidance and psychological support to patients and their families, receive data input by patients and their families and provide response services. The medical collaboration platform is used to connect with the hospital information system and provide medical staff with a remote monitoring view of the patient's readiness.

[0021] The user interaction module serves as the direct window for interaction between the system and the user. Through its interface design, it guides patients or their families to input multi-dimensional individual data, including demographic characteristics, medical history, and lifestyle habits. Simultaneously, this module also acts as a channel for the system to transmit information to the user, responsible for displaying personalized plans, providing health education content, offering psychological support, and receiving real-time feedback from the user.

[0022] The personalized risk assessment module performs in-depth analysis and calculations on this data, using a built-in machine learning model to output a quantified risk level of inadequate bowel preparation. The dynamic protocol generation module connects to the personalized risk assessment module, generating a customized bowel cleansing plan based on the provided risk level and the patient's specific situation. This plan determines the type and dosage of cleansing agents, establishes a personalized medication schedule, and provides accompanying dietary and exercise guidance.

[0023] The real-time monitoring and feedback module establishes data connections with the user interaction module, personalized risk assessment module, and dynamic protocol generation module. The user interaction module collects patient compliance data, bowel characteristics, and adverse reaction information during the preparation process; based on the initial predictions from the risk assessment module, high-risk patients are monitored in a focused manner; and the implementation effect is dynamically evaluated according to the expected goals of the protocol generation module.

[0024] The aforementioned medical collaboration platform serves as a bridge between the system and medical institutions. Connected to the real-time monitoring and feedback module, it transmits basic patient information and examination schedules synchronized from the hospital information system to various modules within the system. Simultaneously, it feeds back preparation process data, early warning information, and evaluation results generated by each module from the real-time monitoring module to the hospital information system, updating the patient's electronic medical record. Furthermore, the platform provides medical staff with a centralized remote monitoring view, enabling them to comprehensively understand the patient's preparation status and provide timely remote intervention for high-risk patients.

[0025] Example 2 The difference between this embodiment and Embodiment 1 is that the step of outputting the risk level of inadequate gut preparation through a machine learning prediction model includes the following steps: Acquire the patient’s individual characteristic data, including demographic characteristics, bowel function status, comorbidities, and functional status assessment. The individual feature data is standardized and preprocessed to form the model input feature vector; The Boruta algorithm based on random forest was used to evaluate the importance of the input feature vector and screen out risk factors related to the quality of bowel preparation. The selected risk factors are input into a pre-trained support vector machine classification model to obtain the probability prediction value of inadequate gut preparation. Based on the probability threshold range of the obtained probability prediction values, patients are classified into low-risk, medium-risk, or high-risk levels.

[0026] The collected raw data, encompassing demographic characteristics, bowel function status, comorbidities, and functional status, was cleaned and standardized to generate feature vectors suitable for machine learning models, laying the foundation for subsequent precise analysis. Subsequently, the system employed the Boruta algorithm based on random forest to evaluate the importance of the preprocessed feature vectors. This algorithm uses a randomly rearranged version of the original features as a benchmark, and systematically filters out all key risk factors truly and significantly related to bowel preparation quality by comparing the importance scores of the original features and the shadow features.

[0027] After feature selection, the system inputs the selected key risk factors into a pre-trained support vector machine classification model for calculation. This model calculates a predicted probability value representing the likelihood of inadequate bowel preparation in the patient by finding the optimal classification hyperplane in a multidimensional feature space. Finally, the system compares this predicted probability value with a preset risk threshold range to accurately classify the patient into low-risk, medium-risk, or high-risk categories.

[0028] Example 3 The difference between this embodiment and embodiment two is that the Boruta algorithm based on random forest is used to evaluate the importance of the input feature vector, which includes the following steps: Obtain the input feature vector and create a randomly rearranged version of the original features as a shadow feature; The importance scores of the original features and shadow features are calculated using the random forest algorithm. Compare the importance of each original feature with the importance of its corresponding shadow feature; Features with significantly higher importance than shadow features are retained, while features with insignificant importance are removed.

[0029] The shadow features are generated by randomly rearranging the original feature values. They retain the statistical distribution of the original data but destroy the actual correlation with the target variable. Subsequently, the original features and shadow features are merged to form an expanded feature set, and the importance score of all features is calculated based on the random forest algorithm. During this process, the importance of each original feature is repeatedly and statistically compared with the importance of its corresponding shadow feature. If the importance of an original feature is consistently and significantly higher than that of its shadow feature, it proves that the feature's predictive ability for the target variable surpasses random factors, and therefore it is judged as an important feature and retained; otherwise, it is considered a redundant or noisy feature and is discarded.

[0030] Example 4 The difference between this embodiment and Embodiment 3 is that the step of inputting the screened risk factors into the pre-trained support vector machine classification model includes the following steps: The k-fold cross-validation method is used to optimize the model parameters on the training set; The model's performance metrics, including accuracy, sensitivity, and specificity, were evaluated using an independent validation set. Test the model's generalization ability on unknown data using an external validation set.

[0031] By employing k-fold cross-validation to optimize model hyperparameters on the training dataset, the training data is randomly divided into k complementary subsets. Through k iterations of training and validation, one subset is used as the validation set each time, while the remaining subsets serve as the training set. This ensures that model parameters are sufficiently and robustly tuned on limited data, effectively preventing overfitting. After parameter optimization, the model's performance is evaluated on an independent validation set not used during training. The system comprehensively assesses the model's discriminative ability under known data distributions by calculating a series of metrics, including accuracy, sensitivity, and specificity. Finally, to verify the model's applicability in novel clinical settings, the system also incorporates external validation sets from different times or medical centers for testing.

[0032] Example 5 The difference between this embodiment and embodiment four is that the step of generating a personalized bowel cleansing plan for the patient based on the risk level includes the following steps: Establish a solution knowledge base that includes standard solutions, enhanced solutions, and strengthened solutions; Based on the risk level output by the risk assessment module, select the corresponding basic solution template from the solution knowledge base; Based on the patient’s individual characteristics data, the parameters of the cleaning agent type, dosage and timing of medication in the basic protocol template are optimized. Based on the patient's examination appointment time and personal routine, a personalized medication schedule is generated; a complete cleaning plan including cleaning agent information, medication time, dietary guidance, and exercise suggestions is output.

[0033] After obtaining the patient's risk level from the risk assessment module, the module selects a corresponding basic treatment template from the knowledge base and refines the key parameters in the template by combining the patient's specific individual characteristics. For example, for patients with chronic constipation, prokinetic drugs or osmotic laxatives are intelligently added as adjuncts to the cleansing agent; the dosage of polyethylene glycol electrolytes is precisely calculated and adjusted based on the patient's weight and renal function; and for patients with limited mobility, an abdominal massage and leg-raising exercise program that can be performed in bed is designed. After parameter optimization, the module further combines the patient's colonoscopy appointment time with their daily routine and uses an intelligent scheduling algorithm to generate a personalized medication schedule with reminders, such as reminding the patient to drink a cup of cleansing solution every 20 minutes to ensure proper medication adherence. Finally, the system integrates all information and outputs a complete cleansing plan that includes cleansing agent information, time-segmented medication times, a phased dietary list, and personalized exercise suggestions.

[0034] Example 6 The difference between this embodiment and embodiment five is that, during the patient's implementation of the cleaning protocol, the collection of compliance data, bowel characteristics, and adverse reaction information includes the following steps: The medication record interface records the actual dosage and time of each medication taken by the patient; Provides a graphical stool characteristics assessment tool to guide patients in recording their bowel movements using the Bristol Stool Characteristics Scale; The severity and timing of adverse reactions such as abdominal distension, abdominal pain, and nausea were collected from patients using structured forms. The system synchronizes patients' water intake records and exercise data in real time.

[0035] In terms of medication adherence monitoring, the module provides an intuitive medication record interface through a user interaction module, guiding patients or their families to accurately record the actual dosage and specific time after each dose, forming a continuous medication adherence curve. For stool characteristics monitoring, a graphical stool characteristics assessment tool, based on the Bristol Stool Characteristics Scale, converts typical stool forms into color-coded illustrations, allowing patients to record stool characteristics simply by selecting options. The system also adds a function for collecting and recognizing stool images; patients can take pictures of their stool, and the system uses image recognition algorithms to compare them with standard bowel cleansing samples to objectively assess the cleansing effect. For adverse reaction monitoring, the module collects data through structured forms, allowing patients to select symptoms such as bloating, abdominal pain, and nausea from a preset list. Specific monitoring is set up for abdominal pain, nausea, and vomiting that may occur during laxative administration, noting the severity and timing of occurrence. In addition, the module simultaneously tracks auxiliary patient data, including fluid intake records and exercise completion status.

[0036] Example 7 The difference between this embodiment and Embodiment Six is ​​that the dynamic assessment and risk warning of cleaning quality includes the following steps: Based on real-time collected compliance data, bowel characteristics, and adverse reaction information, a multi-dimensional assessment index was constructed. The various indicators are weighted and calculated using preset scoring rules to obtain a comprehensive cleaning quality score. When the overall score falls below the threshold or a serious adverse reaction is detected, a tiered early warning mechanism is activated. Based on the severity of the risk, corresponding levels of warning information are sent to both patients and medical staff.

[0037] By establishing an assessment index system that includes patient medication adherence, changes in bowel characteristics, severity of adverse reactions, and fluid intake, and using scoring rules based on clinical evidence, different weights are assigned to each index, and a comprehensive bowel cleanliness score is obtained through weighted calculation, which reflects the actual effect of the patient's current bowel preparation.

[0038] The system compares the overall score with preset quality thresholds in real time. When the score falls below the acceptable threshold, or when a specific serious adverse reaction (such as severe abdominal pain or repeated vomiting) is detected, the module immediately activates an intelligent early warning mechanism. This mechanism adopts a tiered processing principle: for a "Level 1 warning" indicating a slight deviation from the protocol, the system automatically pushes personalized guidance information to the patient through the user interaction module, such as suggesting increased water intake or appropriate movement; for a "Level 2 warning" indicating moderate risk, while pushing information, the system will also notify the responsible medical staff through the medical collaboration platform, prompting them to pay attention to the patient's condition; for a "Level 3 warning" indicating high risk, the system will send an emergency notification to the medical staff and suggest that they immediately take proactive intervention via telephone or video to prevent possible serious complications.

[0039] Example 8 The difference between this embodiment and embodiment seven is that providing health education content, operational guidance, and psychological support to patients and their families includes the following steps: Based on the patient's personalized cleaning plan, generate corresponding illustrated and video operation guides; At the corresponding time points during the implementation of the plan, automatic push notifications will be sent for diet and medication. It provides psychological counseling content, including relaxation training guidance and sharing of success stories; it also provides collaborative supervision functions, including progress tracking, reminders, and assistance guidance.

[0040] Based on the personalized cleaning plan output by the dynamic plan generation module, the module calls upon the media resource library to generate matching text-based and video-based operation guides. On the psychological support front, the module doesn't passively wait for queries but proactively pushes psychological guidance content to patients at key points that could trigger anxiety. This content may include professional relaxation training audio, mindfulness meditation tutorials, and successful experience sharing from other patients, aiming to effectively alleviate their tension. Furthermore, the module provides a dedicated collaborative monitoring interface for family members, allowing them to remotely view the patient's preparation progress, receive system reminders synchronously, and learn how to assist and encourage the patient.

[0041] Example 9 The difference between this embodiment and embodiment eight is that the user interaction module adopts a microservice architecture, which separates user management and message push functions into independent service units. Each service unit communicates and exchanges data through a defined API interface.

[0042] The service units are built around specific business capabilities, such as user management, risk assessment, solution generation, and message push, and each has its own independent codebase, database (if required), and deployment process. These service units communicate and exchange data through well-defined API interfaces, typically based on HTTP / REST or gRPC.

[0043] Example 10 The difference between this embodiment and Embodiment Nine is that the system uses the SHAP interpretability algorithm to analyze the prediction results of the machine learning model, and demonstrates the contribution of risk factors to the prediction results by calculating the Shapley values ​​of each feature parameter.

[0044] Once the Support Vector Machine (SVM) model makes a "high-risk" prediction for a patient, the SHAP interpretability algorithm is activated. Based on Shapley value theory in cooperative game theory, this algorithm aims to fairly allocate the contribution of each feature to the specific prediction outcome. By calculating the difference in model output with and without the feature, the algorithm calculates a SHAP value for each input feature (e.g., age = 80 years, history of diabetes, moderate constipation), generating an intuitive visual interpretation report for the prediction. For example, the report clearly shows that in this patient's case, "advanced age" is the primary factor driving the prediction towards "high risk," followed by "history of diabetes" and "moderate constipation." This ability to make the decision-making logic of the "black box model" transparent allows clinicians to understand and trust the model's judgment, knowing which specific clinical factors led to the high-risk warning. This allows them to more confidently adopt the system's recommendations and take appropriate clinical interventions, achieving intelligent clinical decision-making through human-machine collaboration.

[0045] In summary, this invention achieves data-driven precision medicine through the collaborative work of a personalized risk assessment module and a dynamic protocol generation module. By employing machine learning technology to comprehensively analyze the multi-dimensional individual characteristics of patients, it can accurately identify high-risk patients with inadequate bowel preparation and generate customized cleansing protocols for them, thereby improving the pass rate of bowel preparation and providing reliable assurance for the accuracy and safety of colonoscopy. At the same time, it effectively reduces the probability of missed diagnoses and duplicate examinations.

[0046] In terms of optimizing patient experience and ensuring safety, the close integration of the real-time monitoring and feedback module with the user interaction module constructs a comprehensive management loop. The system can continuously monitor patient adherence during the several-day preparation process, promptly identifying deviations and adverse reactions, and initiating corresponding interventions through a tiered early warning mechanism. This proactive monitoring model significantly improves the safety of elderly patients, especially high-risk patients, during the outpatient preparation phase. Simultaneously, the system's age-friendly interactive design and psychological support functions effectively reduce patients' cognitive burden and anxiety, improving treatment adherence.

[0047] In terms of medical collaboration efficiency, the deep integration of the medical collaboration platform enables seamless connection between in-hospital and out-of-hospital medical processes. This platform allows medical staff to remotely monitor the preparedness of a large number of patients and precisely target medical resources to the high-risk groups most in need. This model not only significantly reduces the workload of medical staff and improves management efficiency but also builds a continuous medical service system, promoting the optimal allocation of medical resources.

[0048] It is understood that those skilled in the art can combine various implementation methods in the above embodiments under the guidance of the above examples to obtain technical solutions with multiple implementation methods.

[0049] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0050] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0051] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0052] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0053] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A personalized intelligent management system for intestinal cleansing preparation in the elderly, characterized in that, include: The personalized risk assessment module is used to obtain multi-dimensional individual data of elderly patients and output the risk level of inadequate bowel preparation through machine learning prediction models. The dynamic protocol generation module is used to generate personalized bowel cleansing protocols for patients based on their risk level. The real-time monitoring and feedback module is used to collect data on patient compliance, bowel characteristics and adverse reactions during the patient's implementation of the cleaning protocol, and to dynamically assess the cleaning quality and provide risk warnings. The user interaction module is used to provide health education content, operation guidance and psychological support to patients and their families, receive data input by patients and their families and provide response services. The medical collaboration platform is used to connect with the hospital information system and provide medical staff with a remote monitoring view of the patient's readiness.

2. The personalized intelligent management system for intestinal cleansing preparation in the elderly according to claim 1, characterized in that, The step of outputting the risk level of inadequate bowel preparation through a machine learning prediction model includes the following steps: Acquire the patient’s individual characteristic data, including demographic characteristics, bowel function status, comorbidities, and functional status assessment. The individual feature data is standardized and preprocessed to form the model input feature vector; The Boruta algorithm based on random forest was used to evaluate the importance of the input feature vector and screen out risk factors related to the quality of bowel preparation. The selected risk factors are input into a pre-trained support vector machine classification model to obtain the probability prediction value of inadequate gut preparation. Based on the probability threshold range of the obtained probability prediction values, patients are classified into low-risk, medium-risk, or high-risk levels.

3. The personalized intelligent management system for intestinal cleansing preparation in the elderly according to claim 1, characterized in that, The Boruta algorithm based on random forest is used to evaluate the importance of the input feature vector, which includes the following steps: Obtain the input feature vector and create a randomly rearranged version of the original features as a shadow feature; The importance scores of the original features and shadow features are calculated using the random forest algorithm. Compare the importance of each original feature with the importance of its corresponding shadow feature; Features with significantly higher importance than shadow features are retained, while features with insignificant importance are removed.

4. The personalized intelligent management system for intestinal cleansing preparation in the elderly according to claim 1, characterized in that, The process of inputting the screened risk factors into a pre-trained support vector machine classification model includes the following steps: The k-fold cross-validation method is used to optimize the model parameters on the training set; The model's performance metrics, including accuracy, sensitivity, and specificity, were evaluated using an independent validation set. Test the model's generalization ability on unknown data using an external validation set.

5. The personalized intelligent management system for intestinal cleansing preparation for the elderly according to claim 1, characterized in that, The process of generating a personalized bowel cleansing plan for patients based on their risk level includes the following steps: Establish a solution knowledge base that includes standard solutions, enhanced solutions, and strengthened solutions; Based on the risk level output by the risk assessment module, select the corresponding basic solution template from the solution knowledge base; Based on the patient’s individual characteristics data, the parameters of the cleaning agent type, dosage and timing of medication in the basic protocol template are optimized. Based on the patient's examination appointment time and personal routine, a personalized medication schedule is generated; a complete cleaning plan including cleaning agent information, medication time, dietary guidance, and exercise suggestions is output.

6. The personalized intelligent management system for intestinal cleansing preparation for the elderly according to claim 1, characterized in that, The process of collecting patient compliance data, bowel characteristics, and adverse reaction information during the implementation of the cleaning protocol includes the following steps: The medication record interface records the actual dosage and time of each medication taken by the patient; Provides a graphical stool characteristics assessment tool to guide patients in recording their bowel movements using the Bristol Stool Characteristics Scale; The severity and timing of adverse reactions such as abdominal distension, abdominal pain, and nausea were collected from patients using structured forms. The system synchronizes patients' water intake records and exercise data in real time.

7. The personalized intelligent management system for intestinal cleansing preparation for the elderly according to claim 1, characterized in that, The dynamic assessment and risk warning of cleaning quality includes the following steps: Based on real-time collected compliance data, bowel characteristics, and adverse reaction information, a multi-dimensional assessment index was constructed. The various indicators are weighted and calculated using preset scoring rules to obtain a comprehensive cleaning quality score. When the overall score falls below the threshold or a serious adverse reaction is detected, a tiered early warning mechanism is activated. Based on the severity of the risk, corresponding levels of warning information are sent to both patients and medical staff.

8. The personalized intelligent management system for intestinal cleansing preparation for the elderly according to claim 1, characterized in that, Providing health education content, operational guidance, and psychological support to patients and their families includes the following steps: Based on the patient's personalized cleaning plan, generate corresponding illustrated and video operation guides; At the corresponding time points during the implementation of the plan, automatic push notifications will be sent for diet and medication. It provides psychological counseling content, including relaxation training guidance and sharing of success stories; it also provides collaborative supervision functions, including progress tracking, reminders, and assistance guidance.

9. A personalized intelligent management system for intestinal cleansing preparation in the elderly according to claim 1, characterized in that, The user interaction module adopts a microservice architecture, which separates user management and message push functions into independent service units. Each service unit communicates and exchanges data through a defined API interface.

10. A personalized intelligent management system for intestinal cleansing preparation in the elderly according to claim 1, characterized in that, The SHAP interpretability algorithm is used to analyze the prediction results of the machine learning model. By calculating the Shapley value of each feature parameter, the contribution of risk factors to the prediction results is shown.

Images