Precision case management intervention system based on user portrait stratification and joint com-b behavior model

Abstract

The present application relates to the technical field of medical health management, and discloses a precise case management intervention system and method based on user portrait layered joint COM-B behavior model, and the technical points are as follows: the system collects clinical data, behavior data and three-dimensional evaluation data of ability, opportunity and motivation of a target object, constructs a multi-dimensional user portrait, determines a case management layered type and a behavior weak target point, identifies compliance, laxity, risk and intervention gaps in combination with multiple behavior indexes, matches a personalized intervention strategy, and converts it into a game-based nursing task to push for execution, and dynamically updates the user portrait, the layered type and the intervention scheme according to the task result, follow-up data, scale data and clinical indexes. The present application can improve the pertinence, continuity and adaptability of case management intervention.

Description

Technical Field

[0001] This invention relates to the field of medical and health management technology, and more specifically, to a precision case management and intervention system based on user profile stratification combined with COM-B behavioral model. Background Technology

[0002] With the increasing number of patients with malignant tumors and other chronic diseases, long-term follow-up, symptom monitoring, treatment compliance, and lifestyle management are often required after treatment. Existing case management systems primarily manage patients through electronic medical records, health check-ins, follow-up records, and scale data. However, these systems typically categorize patients based only on disease stage, risk indicators, or task completion rates, failing to comprehensively reflect their actual management status. For example, some patients may check in frequently but provide simple feedback and incomplete health records, potentially exhibiting false compliance; others may experience periods of apathy due to treatment fatigue, work commitments, or insufficient family support.

[0003] The COM-B behavioral model can analyze the reasons why patients fail to complete health management behaviors from three dimensions: ability, opportunity, and motivation. However, current technologies often only save COM-B assessment results as independent scale data, failing to conduct joint analysis with clinical data, behavioral data, and patient stratification results. Therefore, it is difficult to determine whether the patient's problem is caused by insufficient knowledge or operational ability, insufficient external support, or decreased management motivation. Existing systems also have limited analysis of the combined relationships between multiple behavioral indicators, making it difficult to identify pseudo-compliance, high-risk states, and intervention gaps where existing intervention content does not match the patient's actual obstacles in a timely manner.

[0004] Furthermore, existing intervention programs often use fixed templates, with patients in the same category typically receiving the same health education content, reminder frequency, and follow-up plans, lacking dynamic adjustments based on the causes of behavioral disorders. While some systems offer rewards such as points, check-ins, or achievements, gamified tasks usually cannot be adjusted in real time based on task completion, number of consecutive interruptions, abnormal indicators, and follow-up feedback. Moreover, post-intervention data is not continuously used to update user profiles and management stratification.

[0005] Therefore, there is an urgent need to provide a precise case management and intervention system based on user profile layering and COM-B behavior model to solve these problems. Summary of the Invention

[0006] The purpose of this invention is to solve the technical problems mentioned in the background section and to provide a precise case management and intervention system based on user profile layering and COM-B behavior model.

[0007] The above-mentioned objective of the present invention is achieved as follows: A precise case management and intervention system based on user profiling and COM-B behavioral models includes: The data acquisition module is used to collect clinical data, behavioral data, and COM-B three-dimensional assessment data of the target subjects. The COM-B three-dimensional assessment data includes capability dimension data, opportunity dimension data, and motivation dimension data. The user profile building module is used to preprocess and extract features from the clinical data, behavioral data, and COM-B three-dimensional assessment data, generate a multi-dimensional profile feature vector of the target object, and generate user profile tags based on the multi-dimensional profile feature vector. The profile layering module is used to input the multi-dimensional profile feature vector and user profile tags into a preset profile layering model and output the case management layering type corresponding to the target object. The COM-B target identification module is used to identify the target's behavioral weaknesses based on the assessment results of the target in terms of ability, opportunity, and motivation. The combined behavior prediction module is used to identify at least one of the target object's compliance status, apathy status, risk status, and intervention gap based on the combined relationship between multiple behavioral indicators of the target object. The intervention decision module is used to match intervention strategies from the intervention strategy library based on the case management stratification type, behavioral vulnerability targets, and combined behavior prediction results, and generate personalized case management intervention plans. The gamified intervention execution module is used to convert the personalized case management intervention plan into a gamified nursing task that can be executed on the terminal, and to adjust the subsequent nursing tasks according to the task execution results of the target object; The follow-up feedback optimization module is used to collect at least one of the following: follow-up data, scale data, and clinical indicators after intervention, and to update user profile tags, case management stratification types, and / or personalized case management intervention plans based on the collection results.

[0008] Furthermore, the clinical data includes at least one of the following: disease type, disease stage, treatment stage, treatment plan, complication status, symptom information, and follow-up indicators; the behavioral data includes at least one of the following: health check-in records, medication or treatment cooperation records, follow-up response records, symptom reporting records, health education learning records, and lifestyle records.

[0009] Furthermore, the user profile construction module is used to process the collected data for missing value processing, outlier processing, standardization processing, and encoding processing, and to fuse the processed clinical features, behavioral features, and COM-B three-dimensional features to form a multi-dimensional profile feature vector for profile stratification and intervention decision-making.

[0010] Furthermore, the profile stratification model is a machine learning model, a rule model, or a hybrid model formed by combining machine learning and rule models; the case management stratification type includes at least one of the following: routine maintenance type, pseudo-compliance type, phased slackness type, and high-risk key management type.

[0011] Furthermore, the COM-B target identification module calculates the weakness scores corresponding to the capability dimension, opportunity dimension, and motivation dimension, and determines the dimensions with weakness scores reaching a preset threshold as priority intervention targets, so that the intervention content corresponds to the behavioral disorder causes of the target object.

[0012] Furthermore, the combined behavior prediction module is used to perform combined analysis on at least two of the following: task completion frequency, number of consecutive interruptions, follow-up feedback quality, health record completeness, occurrence of abnormal indicators, and symptom reporting. Based on the combined analysis results, it determines whether the target subject has false compliance, periodic apathy, high relapse risk, or intervention gap.

[0013] Furthermore, the intervention decision module is used to call the corresponding intervention strategy template according to different case management hierarchical types, and adjust at least one of the following in the intervention strategy template: nursing education content, follow-up frequency, reminder method, task intensity, risk warning level, and case management focus, based on the behavioral weakness target.

[0014] Furthermore, the gamified nursing tasks include at least one of the following: challenge tasks, points tasks, level tasks, achievement tasks, and feedback tasks; the gamified intervention execution module dynamically adjusts the difficulty of subsequent tasks, the frequency of task pushes, or the content of reminders based on at least one of the target object's task completion rate, number of consecutive non-completions, abnormal indicator feedback, and follow-up response.

[0015] This invention also provides a precise case management and intervention method based on user profile layering and COM-B behavior model, comprising the following steps: Collect clinical data, behavioral data, and COM-B three-dimensional assessment data from the target subjects; The clinical data, behavioral data, and COM-B three-dimensional assessment data are preprocessed and feature extracted to generate multi-dimensional profile feature vectors and user profile tags. Based on the multi-dimensional profile feature vector and user profile tags, determine the case management hierarchical type of the target object; The behavioral weaknesses of the target individuals are identified based on the assessment results of the capability, opportunity, and motivation dimensions. Identify at least one of the following in the target object’s compliance status, apathy status, risk status, and intervention gap based on the combined relationships between multiple behavioral indicators; Personalized case management intervention plans are generated based on the case management stratification type, behavioral vulnerability targets, and combined behavioral prediction results. The personalized case management intervention plan is converted into gamified nursing tasks and pushed to the terminal for execution; Collect at least one of the following data, scale data, and clinical indicators after intervention, and update user profile tags, case management stratification types, and / or personalized case management intervention plans based on the collection results.

[0016] Furthermore, the target population is colorectal cancer patients; or, the target population is other disease populations that require long-term follow-up, self-management, or case management intervention, and the system adapts to different disease populations by adjusting clinical data fields, profile stratification model parameters, and intervention strategy library.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention integrates clinical data, behavioral data, and COM-B (Combined Combination of Business and Innovation) assessment data across three dimensions—capability, opportunity, and motivation—to construct a multidimensional user profile of the target population. By combining this profile with a stratified model and combined behavioral prediction results, it identifies the target population's compliance status, periods of apathy, high-risk states, and intervention gaps. Compared to management methods based solely on disease stage, single indicators, or task completion rates, this invention can more accurately distinguish between apparent task completion and actual management quality, further identifying the main obstacles affecting target behavior and improving the accuracy of case management stratification and behavioral risk identification.

[0018] 2. This invention matches intervention strategies based on case management stratification types, COM-B behavioral weaknesses, and combined behavioral prediction results. It personalizes nursing education content, follow-up frequency, reminder methods, task intensity, and risk warning levels, and enhances target audience participation and sustained commitment through gamified nursing tasks. Simultaneously, the system continuously updates user profiles, stratification types, and intervention plans based on task execution results, follow-up data, scale data, and clinical indicators, thus forming a dynamic closed-loop management system. This reduces ineffective reminders or over-intervention caused by fixed template interventions, improving the targeting, continuity, and adaptability of case management interventions. Attached Figure Description

[0019] Figure 1 A system block diagram of a precision case management and intervention system based on user profiling and COM-B behavioral model; Figure 2 This is a flowchart of a precision case management and intervention system based on user profile stratification and combined with the COM-B behavioral model. Detailed Implementation

[0020] The technical solution of the present invention will be further described below with reference to specific embodiments. It should be understood that the following embodiments are only used to illustrate the present invention and are not intended to limit the scope of protection of the present invention. Equivalent substitutions or conventional adjustments made by those skilled in the art to data fields, model types, parameter values, threshold settings, terminal forms, and intervention strategies without departing from the technical concept of the present invention are all within the scope of implementation of the present invention.

[0021] Example 1: This example uses the long-term follow-up and case management of colorectal cancer patients as an example to illustrate a precision case management intervention system based on user profile stratification combined with a COM-B behavioral model. The system is deployed on a server or cloud server in a medical institution and communicates with the hospital's electronic medical record system, follow-up management system, medical and nursing management terminal, and patient terminal. The patient terminal can be a smartphone, tablet, smartwatch, or other electronic terminal capable of receiving nursing tasks and uploading execution results.

[0022] The system comprises a data acquisition module, a user profile construction module, a profile stratification module, a COM-B target identification module, a combined behavior prediction module, an intervention decision-making module, a gamified intervention execution module, and a follow-up feedback optimization module. Each module establishes a data correspondence through a unique patient identifier, enabling continuous correlation between the patient's clinical data, behavioral data, assessment data, intervention tasks, and follow-up results.

[0023] The data acquisition module collects information from the hospital's electronic medical record system, including the target patient's disease type, disease stage, treatment phase, treatment plan, complications, symptoms, and follow-up indicators. In this embodiment, the disease type is colorectal cancer, the disease stage is the clinically confirmed stage, the treatment phase may include preoperative preparation, postoperative recovery, adjuvant therapy, and regular follow-up, and the treatment plan may include surgical treatment, drug therapy, radiotherapy, or a combination of multiple treatment methods.

[0024] The data acquisition module also collects patients' health check-in records, medication or treatment compliance records, follow-up response records, symptom reporting records, health education learning records, and lifestyle records through patient terminals and the follow-up management system. Lifestyle records can include data on diet, exercise, sleep, bowel movements, and weight changes. Each behavioral data item records the time of data generation, task type, completion status, and data source to determine changes in the patient's behavior across different management cycles.

[0025] Simultaneously, the data acquisition module collects COM-B three-dimensional assessment data through patient self-assessment scales, healthcare staff evaluation forms, or assessment questionnaires jointly completed by patients and healthcare staff. The competence dimension data reflects whether patients possess the necessary disease knowledge, symptom recognition ability, treatment cooperation ability, operational ability, and information comprehension ability to complete health management behaviors. The opportunity dimension data reflects whether patients have sufficient family support, healthcare support, time availability, terminal usage conditions, transportation access, and medical resources. The motivation dimension data reflects patients' health beliefs, risk perception, self-efficacy, treatment willingness, behavioral habits, and willingness to continue management.

[0026] When a patient is first included in case management, the data acquisition module collects a set of initial clinical data and COM-B assessment data. During subsequent management, the system continuously collects behavioral data according to a preset management cycle. This management cycle can be set to one day, one week, two weeks, or one month, depending on the patient's treatment stage. For example, a shorter management cycle is used during the early postoperative recovery stage, while a longer management cycle is used during the stable, regular follow-up stage.

[0027] The user profile building module preprocesses the collected data. For missing data, it prioritizes retrieving the most recent valid record for the patient; if no valid historical records exist, it can supplement the data using reference data from patients with the same disease type and treatment stage, or mark the data as temporarily unknown. For data that clearly exceeds a reasonable range, the system distinguishes between data sources and clinical significance. If the data is due to duplicate uploads from the terminal, incorrect formatting, or incorrect patient input, it is corrected or removed; if it represents a genuine clinical abnormality in the patient, the abnormal data is retained, and an abnormal indicator label is generated.

[0028] The user profile building module also standardizes data from different sources, units, and formats, converting data such as disease staging, treatment stage, treatment plan, and complication status into codes that the system can recognize, and converting review indicators, task completion status, and COM-B evaluation results into a unified data format.

[0029] After preprocessing, the user profile building module extracts clinical features from clinical data, such as disease severity, treatment stage, complication burden, symptom changes, and follow-up risk; behavioral features from behavioral data, such as task completion frequency, continuous interruption, follow-up response, completeness of health records, level of health education, and symptom reporting; and features from COM-B assessment data, including three dimensions: ability, opportunity, and motivation.

[0030] The user profile building module integrates the aforementioned clinical characteristics, behavioral characteristics, and COM-B three-dimensional characteristics to form a multi-dimensional profile feature vector for the patient, and generates user profile tags based on this multi-dimensional profile feature vector. These user profile tags may include tags for treatment stage, disease risk, treatment cooperation, follow-up participation, symptom reporting, family support, digital terminal usage ability, skill deficiencies, lack of opportunity, and lack of motivation.

[0031] For example, a colorectal cancer patient in the postoperative adjuvant therapy phase may have a high health check-in completion rate on their device, but the patient may have provided limited symptom information, only responding "normal" or "completed" during multiple follow-ups, and the health record may contain incomplete information on diet, bowel movements, and adverse drug reactions. Based on this, the user profile building module generates user profile tags such as "high task completion rate," "low feedback quality," "incomplete health record," and "postoperative adjuvant therapy phase."

[0032] The profile stratification module inputs the patient's multidimensional profile feature vector and user profile labels into a preset profile stratification model. The profile stratification model can be a rule-based model, a machine learning model, or a hybrid model combining rule-based and machine learning models.

[0033] When using a hybrid model, the machine learning model outputs the probability that a patient belongs to each stratification type based on previous case management data, while the rule-based model determines whether the patient meets the corresponding stratification requirements based on management conditions pre-set by medical staff. The system then combines the results of the two assessments to determine the final stratification type. The case management stratification types include routine maintenance, pseudo-compliance, intermittent apathy, and high-risk intensive management.

[0034] Routine maintenance refers to patients whose clinical condition is relatively stable, who complete health management tasks well, provide relatively complete feedback, and for whom no obvious behavioral disorders or risk factors have been found. For these patients, the system's primary goal is to maintain their existing health management behaviors.

[0035] Pseudo-compliance refers to patients who exhibit high task completion rates or check-in frequency on the terminal, but whose task feedback is simplistic, health records are incomplete, symptom reporting is insufficient, or task completion results are inconsistent with their actual clinical status. For such patients, the system cannot determine their compliance status solely based on the number of clicks to complete tasks.

[0036] Phased apathy refers to a condition where a patient performs well in the previous phase but experiences a decline in task completion rate, multiple interruptions, delayed follow-up responses, or reduced health education learning during the current management cycle.

[0037] High-risk, high-priority management refers to patients who exhibit significant clinical abnormalities, risk of complications, changes in symptoms, require continuous follow-up, or exhibit multiple adverse behaviors, necessitating increased priority for medical and nursing follow-up and intensified case management.

[0038] For the aforementioned postoperative adjuvant therapy patients, although their health check-in completion rate was high, their feedback content and health records were incomplete, and their follow-up indicators showed abnormal changes. Therefore, the profiling and stratification module can identify these patients as high-risk key management patients and retain their "false compliance tendency" label for subsequent intervention decisions.

[0039] The COM-B target identification module determines the degree of restriction on patient behavior execution based on the assessment results of the ability, opportunity, and motivation dimensions.

[0040] In terms of capabilities, the system determines whether the patient understands the treatment cooperation requirements, can identify the symptoms that need to be reported, has mastered the operation of the patient terminal, and can correctly complete the health record.

[0041] In terms of opportunity, the system determines whether the patient has family support, the conditions for stable terminal use, the time to complete nursing tasks, the ability to obtain timely guidance from medical staff, and whether there are external limitations such as transportation, work, or family care.

[0042] In terms of motivation, the system assesses whether patients recognize the importance of long-term follow-up, whether they have the willingness to continue completing tasks, and whether their behavioral motivation is reduced due to treatment fatigue, anxiety, negative emotions, or lack of obvious short-term effects.

[0043] The COM-B target identification module identifies dimensions with weaknesses meeting preset criteria as priority intervention targets. If significant weaknesses exist in multiple dimensions, the intervention sequence is determined based on the degree of weakness and the current clinical risk.

[0044] In this embodiment, the patient was able to understand basic postoperative management requirements and operate the patient terminal; therefore, the ability dimension was not a major obstacle. However, the patient's daytime work hours were irregular, family member involvement was low, and there was a lack of continuous reminders. Furthermore, the patient believed that "completing the check-in is sufficient," indicating insufficient understanding of the importance of detailed symptom recording. Therefore, the system identified the opportunity and motivation dimensions as the main behavioral weaknesses.

[0045] The combined behavior prediction module analyzes multiple behavioral indicators of patients in combination, rather than making judgments based on a single indicator. The analyzed behavioral indicators include at least two of the following: task completion frequency, number of consecutive interruptions, follow-up feedback quality, health record completeness, occurrence of abnormal indicators, and symptom reporting.

[0046] For example, when a patient completes tasks frequently but has low follow-up feedback quality and health record completeness, the system judges that they may have false compliance; when a patient's task completion frequency decreases over several consecutive management cycles, accompanied by continuous interruptions and follow-up delays, the system judges that they may be in a state of temporary apathy; when a patient has abnormal re-examination indicators, worsening symptoms, or fails to report important symptoms in a timely manner, the system judges that they are at high risk; when the intervention content already pushed by the system fails to resolve the patient's main COM-B behavioral disorders, the system judges that there is an intervention gap in the existing intervention program.

[0047] For the patient in this embodiment, the combined behavior prediction module found that although the number of check-ins was high, the symptom descriptions, bowel movement records, and adverse drug reaction records were significantly insufficient, and the patient did not proactively report any abnormal indicators. Therefore, pseudo-compliance and high-risk states were identified. Meanwhile, existing intervention programs mainly involved repeatedly pushing health education articles, failing to address the patient's lack of time and sustained motivation, thus further identifying intervention gaps.

[0048] The intervention decision module matches intervention strategies from the intervention strategy library based on the patient's case management stratification type, COM-B behavioral weak targets, and combined behavioral prediction results. The intervention strategy library stores intervention strategy templates corresponding to different stratification types, different COM-B targets, and different behavioral states.

[0049] For routine maintenance patients, the intervention strategy template mainly consists of low-frequency follow-up, routine health education, and periodic check-up reminders; for patients with pseudo-compliance, the strategy adds verification of feedback authenticity, detailed symptom inquiry, and health record integrity checks; for patients with periodic apathy, the strategy breaks down complex tasks into multiple easy-to-complete short tasks, and adds periodic reminders and feedback on the reasons for interruption; for high-risk key management patients, the strategy increases the frequency of follow-up, risk warning level, and priority of review by medical staff.

[0050] For patients with weak competence, the system adds content on knowledge explanation, operation demonstration, step guidance, and error correction; for patients with weak opportunity, the system adds content on family collaboration, time window adjustment, terminal adaptation, and alternative feedback methods; for patients with weak motivation, the system adds content on goal feedback, progress feedback, points rewards, level rewards, achievement rewards, and self-efficacy reinforcement.

[0051] In this embodiment, the intervention decision module calls the high-risk, key management intervention strategy template and adjusts it based on weaknesses in the opportunity and motivation dimensions. The adjusted personalized case management intervention plan includes: changing the original fixed-time push to two task time windows that patients can choose; breaking down longer health education content into shorter learning tasks; changing the simple "completed?" to a feedback task that includes symptom options and necessary explanations; pushing stage-based collaborative reminders to family members after obtaining patient authorization; increasing the warning level of abnormal symptoms and re-examination indicators in the medical care management terminal; increasing the frequency of proactive follow-up by case management personnel; and enhancing patients' willingness to continue recording through points and stage achievements.

[0052] The gamified intervention execution module transforms personalized case management intervention plans into gamified nursing tasks that patients can perform on their devices. These tasks include at least one of the following: challenge tasks, point-based tasks, level-based tasks, achievement tasks, and feedback tasks.

[0053] In this embodiment, the system divides a management cycle into multiple stages. The first stage is a symptom recognition learning task, where patients complete a brief learning session and a corresponding test before moving on to the next stage. The second stage is a continuous symptom recording task, where patients complete records of bowel movements, diet, and discomfort symptoms as required and earn points. The third stage is a treatment cooperation feedback task, where patients provide feedback on treatment execution time, execution status, and adverse reactions. The fourth stage is a stage summary task, where patients view the completion status of this cycle, changes in points, and items that need improvement.

[0054] When a patient completes a task consecutively, the system generates a consecutive completion achievement; when the quality of the patient's feedback and the completeness of their health records improve, the system upgrades the patient's task level; when a patient fails to complete a task consecutively, the system does not directly and repeatedly push the same content, but first asks the patient to select the reason for not completing it, such as being too busy at work, feeling unwell, not knowing how to operate it, forgetting to perform it, or thinking that the task is unnecessary, and then adjusts the subsequent tasks according to the reason.

[0055] If a patient fails to complete a task due to irregular working hours, the system expands the executable time window and reduces the content of a single task; if a patient fails to complete a task because they do not know how to fill in symptom records, the system adds examples and operation guidance; if a patient fails to complete a task due to a lack of management willingness, the system adds stage goals, progress feedback, and achievement incentives; if a patient reports physical discomfort or the system detects abnormal indicators, the system stops increasing the difficulty of the game task and sends a risk warning to the medical management terminal.

[0056] The gamified intervention execution module dynamically adjusts the difficulty, frequency, and reminder content of subsequent tasks based on the patient's task completion rate, number of consecutive non-completions, feedback on abnormal indicators, and follow-up response. When task performance is good and clinical risk is low, repeated reminders can be appropriately reduced or the task level can be increased; when task performance declines, complex tasks are broken down into shorter tasks; when abnormal indicators appear, safety follow-up and medical staff review are prioritized, and increasing the number of tasks is not the main intervention method.

[0057] The follow-up feedback optimization module continues to collect patient follow-up data, COM-B scale data, and clinical indicators after intervention. Follow-up data includes the patient's task completion status, reasons for non-completion, symptom feedback, treatment cooperation, and medical staff follow-up records; scale data includes the reassessment results of the three dimensions of ability, opportunity, and motivation; clinical indicators include relevant indicators generated during the patient's follow-up examination and changes in clinical status confirmed by medical staff.

[0058] After several management cycles, if the quality of patients' symptom feedback improves, health records become more complete, and abnormal symptoms are reported in a timely manner, while the weakness in the opportunity and motivation dimensions decreases, the follow-up feedback optimization module updates the patient's user profile tags, adjusting the original tags such as "low feedback quality," "insufficient opportunities," and "insufficient motivation" to the new status.

[0059] The patient profile stratification module redetermines the patient's case management stratification type based on the updated multidimensional profile features and user profile tags. If the patient's clinical condition tends to stabilize and task performance and feedback quality significantly improve, they can be adjusted from high-risk key management type to routine maintenance type or other lower management intensity type; if the patient's task performance improves but clinical abnormalities still exist, they should remain in high-risk key management type, while the patient's task burden should be adjusted appropriately.

[0060] The COM-B target identification module re-evaluates the weaknesses across the three dimensions of ability, opportunity, and motivation. If the opportunity dimension has significantly improved, but the motivation dimension remains insufficient, subsequent intervention programs reduce family collaboration and time adjustment content, and increase goal feedback and sustained behavioral incentives.

[0061] The combined behavior prediction module reassesses whether patients still exhibit pseudo-compliance, intermittent apathy, high-risk states, or intervention gaps based on new behavioral indicators. The intervention decision module updates the personalized case management intervention plan based on the redefined stratification type, weak targets, and combined behavior prediction results.

[0062] The aforementioned system is also capable of executing corresponding precise case management intervention methods. These methods include: collecting clinical data, behavioral data, and COM-B three-dimensional assessment data from target patients; preprocessing and extracting features from the collected data to generate multi-dimensional profile feature vectors and user profile tags; determining the case management stratification type based on the multi-dimensional profile feature vectors and user profile tags; identifying behavioral weaknesses based on the assessment results of the three dimensions of ability, opportunity, and motivation; identifying compliance status, apathy status, risk status, and intervention gaps based on the combined relationships between multiple behavioral indicators; generating personalized case management intervention plans based on the case management stratification type, behavioral weaknesses, and combined behavioral prediction results; converting the intervention plan into gamified nursing tasks and pushing them to the patient's terminal for execution; collecting follow-up data, scale data, and clinical indicators after the intervention, and updating the user profile tags, case management stratification type, and personalized case management intervention plan accordingly.

[0063] This embodiment targets colorectal cancer patients. In other embodiments, the target population can also be other disease populations requiring long-term follow-up, self-management, or case management interventions. The system can adjust clinical data fields, user profile tags, profile stratification model parameters, risk assessment conditions, and intervention strategy library according to different diseases, but the target identification method, combined behavior analysis method, intervention strategy matching method, gamified task execution method, and follow-up feedback optimization method for the three dimensions of ability, opportunity, and motivation remain consistent.

[0064] Example 2: Based on the system structure and business process in Example 1, this example focuses on patients in the adjuvant therapy stage after colorectal cancer surgery. It describes the calculation process of user profile construction, profile stratification, identification of weak targets in COM-B behavior, prediction of combined behaviors, personalized intervention decision-making, gamified task adjustment, and follow-up feedback optimization.

[0065] In this embodiment, the system uses a seven-day management cycle. The data acquisition module collects clinical data, behavioral data, and COM-B three-dimensional assessment data from the target patient within each management cycle. Clinical data includes disease stage, treatment phase, treatment plan, complications, symptom information, and follow-up indicators; behavioral data includes health check-in records, medication or treatment compliance records, follow-up response records, symptom reporting records, health education learning records, and lifestyle records; COM-B three-dimensional assessment data includes capability dimension data, opportunity dimension data, and motivation dimension data.

[0066] The user profile building module performs missing value processing, outlier processing, standardization, and encoding on the collected data. Data generated due to duplicate uploads, incorrect data formats, or obvious errors is corrected or deleted; abnormal data confirmed by medical personnel to have real clinical significance is retained and anomaly indicator labels are generated. Continuous data is converted into dimensionless feature values ​​in the range of 0 to 1, while categorical data is processed using one-hot encoding or ordinal encoding.

[0067] This embodiment selects clinical risk level, task completion rate, degree of continuous interruption, follow-up feedback quality, health record completeness, and degree of abnormal indicators.

[0068] The degree of symptom reporting deficiency and the scores of the three dimensions of COM-B constitute the target patient's [score / analysis / etc.] in the [specific context]. Multidimensional profile feature vector within a management cycle (1); In the formula, Indicates the first Multidimensional profile feature vectors within a management cycle; Indicates the degree of clinical risk; Indicates the task completion rate; Indicates the degree of continuous interruption; Indicates the quality of follow-up feedback; Indicates the completeness of health records; Indicates the degree of abnormality; Indicates the degree of deficiency in symptom reporting; , and These represent the assessment scores for the ability, opportunity, and motivation dimensions, respectively; the superscript T indicates matrix transpose.

[0069] During the current management cycle, the system planned to push 20 nursing tasks to the target patient. The patient actually completed 16 tasks, so the task completion rate was 0.80. The patient experienced two consecutive interruptions during the management cycle, and the degree of consecutive interruptions was normalized to 0.40. Based on the completeness, relevance, and effectiveness of the follow-up content, the follow-up feedback quality was determined to be 0.45. Based on the ratio of items that should be recorded to items that are actually recorded, the health record completeness was determined to be 0.50.

[0070] The patient's follow-up indicators were abnormal, with an abnormality level of 0.60; the patient had delayed or missed reporting of symptoms, with a symptom reporting omission level of 0.40; and based on the disease stage, treatment phase, complications, and follow-up indicators, the clinical risk level was determined to be 0.55.

[0071] The COM-B assessment results showed that the patient possessed basic disease knowledge, symptom recognition ability, and patient terminal operation skills, but had insufficient available time, low family cooperation, and a decreased willingness to continue health management. Therefore, the ability dimension score was 0.65, the opportunity dimension score was 0.40, and the motivation dimension score was 0.35.

[0072] Therefore, the multi-dimensional profile feature vector for the current management cycle is: (2); The user profile building module generates user profile tags such as "postoperative adjuvant treatment stage", "high clinical risk", "high task completion rate", "low feedback quality", "incomplete health record", "insufficient opportunity dimension" and "insufficient motivation dimension" based on the multidimensional profile feature vector.

[0073] The profile stratification module inputs multi-dimensional profile feature vectors and user profile labels into a pre-defined profile stratification model. The profile stratification model includes a machine learning model and a rule-based model. The machine learning model outputs the predicted probability of the target patient belonging to the routine maintenance type, the pseudo-adherence type, the phased slackness type, or the high-risk key management type; the rule-based model outputs the corresponding rule compliance degree based on conditions such as clinical risk, task performance, feedback quality, and abnormal indicators.

[0074] No. The comprehensive stratification score for each case management stratification type is determined according to the following formula: (3); In the formula, Indicates the first The comprehensive stratification score for each case management stratification type; This represents the predicted probability output by the machine learning model; This indicates the degree of rule compliance output by the rule model; This represents the fusion coefficient between the machine learning model and the rule-based model, and .

[0075] The final case management stratification type for the target patients is: (4); In the formula, This indicates the case management stratification type with the highest overall stratification score.

[0076] In this embodiment, the predicted probabilities of the four hierarchical types output by the machine learning model are 0.10, 0.23, 0.20, and 0.47, respectively; the rule compliance degrees of the four hierarchical types output by the rule model are 0.05, 0.25, 0.20, and 0.50, respectively. Setting the fusion coefficient to 0.60, the comprehensive hierarchical scores of the four hierarchical types are as follows: ; Among them, the fourth stratification type had the highest comprehensive stratification score, so the profiling stratification module identified this patient as a high-risk, key management type. At the same time, the system retained the "false compliance tendency" user profile tag for subsequent combined behavior analysis and intervention strategy matching.

[0077] The COM-B target identification module calculates the weakness level in the corresponding dimension based on the assessment scores of the capability, opportunity, and motivation dimensions. (5); In the formula, Indicates the first The degree of weakness in each COM-B dimension; This represents the evaluation score for the corresponding dimension; when hour, ;when hour, ;when hour, Substituting the scores of the three dimensions in this embodiment into equation (5), we get: ; The COM-B weakness threshold was set to 0.45. Since the weaknesses in both the opportunity and motivation dimensions reached the threshold, these dimensions were identified as priority intervention targets.

[0078] For dimensions that reach the weakness threshold, the intervention priority weight is: (6); In the formula, Indicates the first Prioritize interventions in each weak dimension; This represents the set of COM-B dimensions that have reached the weakness detection threshold; Represents the set of elements. The degree of weakness in each dimension.

[0079] In this embodiment, Therefore, the intervention priority weights for the opportunity dimension and the motivation dimension are respectively: ; Therefore, both the opportunity dimension and the motivation dimension are the main intervention targets, with the intervention priority of the motivation dimension being slightly higher than that of the opportunity dimension.

[0080] The combined behavior prediction module calculates the predicted probabilities of pseudo-compliance, intermittent apathy, high-risk status, and intervention gaps based on task completion rate, degree of continuous interruption, follow-up feedback quality, health record completeness, degree of abnormal indicators, degree of symptom reporting omissions, and clinical risk level. The intermediate evaluation value for pseudo-compliance is: (7); In the formula, This represents the intermediate evaluation value for the state of pseudo-compliance. The first combined term is used to represent the state of high task completion rate but low follow-up feedback quality; the second combined term is used to represent the state of high task completion rate but incomplete health records; the third term is used to reflect the impact of abnormal indicators on the pseudo-compliance identification results. Substituting the data of the current management cycle into equation (7), we get: ; The probability of a false compliance state is: (8); Substitution ,get: ; The threshold for determining false compliance was set to 0.60, thus confirming that the patient was in a state of false compliance.

[0081] The intermediate evaluation value for a phase of slackness is: (9); In the formula, Intermediate evaluation value indicating a phase of slackness; This is used to reflect the simultaneous occurrence of a decrease in task completion rate and continuous interruptions.

[0082] Substituting the corresponding data into equation (9), we get: ; Therefore, the probability of a period of slackness is: (10); The threshold for determining a state of periodic apathy was set to 0.60, therefore, the current management cycle has not yet determined that the patient has a state of periodic apathy.

[0083] The intermediate rating for a high-risk status is: (11); In the formula, Intermediate evaluation value indicating a high-risk state; This is used to reflect the combined risk formed when abnormal indicators and symptom reporting are both missing.

[0084] Substituting the corresponding data into equation (11), we get: ; The probability of a high-risk state is: (12); The threshold for determining a high-risk status was set at 0.60, therefore this patient was identified as having a high-risk status. The intermediate evaluation value for the intervention gap was: (13); In the formula, The intermediate evaluation value representing the intervention gap; This indicates the greatest weakness among the capability, opportunity, and motivation dimensions. This represents the probability of a false compliance state. This indicates the probability of a high-risk state.

[0085] In this embodiment: ; Substituting the corresponding data into equation (13), we get: ; The probability of an intervention gap is: (14); The threshold for identifying the intervention gap was set at 0.60, thus confirming the existence of an intervention gap in the existing intervention program. The calculations above indicate that this patient belongs to the high-risk, high-priority management category, with opportunity and motivation dimensions being the priority intervention targets. Furthermore, the patient exhibits pseudo-compliance, a high-risk status, and an intervention gap.

[0086] Based on the above calculation results, the intervention decision-making module selects candidate intervention strategies from the intervention strategy library. The matching scores for the candidate intervention strategies are as follows: (15); In the formula, Indicates the first Matching scores of candidate intervention strategies; This indicates the degree to which the strategy is adapted to the current case management stratification type; This indicates that the strategy applies to the first... The degree of fit of each COM-B weak target; Indicates the first The predicted probability of a combination of behavioral states; This indicates that the strategy applies to the first... The degree of fit between various combinations of behavioral states; This indicates the historical intervention effects of the strategy; This indicates the executive burden that the strategy places on patients; , β, γ, η and These represent the weights of the corresponding evaluation factors.

[0087] The intervention decision module selects intervention strategies from high to low according to the matching score and excludes strategies that are not suitable for the patient's current treatment stage, complication status, or end-use conditions.

[0088] In this embodiment, intervention strategies with high matching scores include task time window adjustment strategy, short-term nursing education strategy, family collaboration reminder strategy and effective feedback verification strategy, symptom reporting reinforcement strategy, stage goal feedback strategy and medical staff key follow-up strategy.

[0089] The personalized case management intervention plan generated based on the above strategy includes: adjusting the original fixed-time push notifications to two task time windows that patients can choose; breaking down lengthy health education content into multiple short-term learning tasks; adjusting simple "completed" feedback to feedback tasks that include symptom options, execution time, and necessary explanations; pushing phased collaborative reminders to family members after obtaining patient authorization; increasing the warning level of abnormal re-examination indicators and missing symptom reporting information in the medical care management terminal; increasing the frequency of proactive follow-up visits by case management personnel; and strengthening patients' motivation for continuous management through points, levels, and phased achievements.

[0090] The gamified intervention execution module transforms the personalized case management intervention plan into challenge tasks, point-based tasks, level-based tasks, achievement tasks, and feedback tasks. The system adjusts the task intensity for the next management cycle based on the patient's performance in the current management cycle. (16); In the formula, Indicates the task intensity for the next management cycle; Indicates the task intensity of the current management cycle; This indicates the performance of tasks during the current management cycle; Indicates the performance of the preset target; Indicates the probability of a high-risk state; This represents the task performance adjustment factor; Indicates the risk inhibition coefficient; and These represent the minimum and maximum allowed task intensity, respectively; This indicates that the calculation results will be limited to a preset range.

[0091] When patients complete tasks well, follow-up feedback quality improves, and the probability of high-risk states decreases, the system appropriately increases the task level or reduces repetitive reminders. When patients fail to complete tasks consecutively or the probability of high-risk states increases, the system reduces the complexity of individual tasks on the patient's end, breaks longer tasks into multiple shorter tasks, and increases the follow-up priority for medical staff, rather than simply increasing the number of tasks for patients.

[0092] After three management cycles of intervention, the follow-up feedback optimization module re-collected patient follow-up data, COM-B scale data, and clinical indicators. Assuming the newly collected task completion rate is 0.90, the degree of continuous interruption is 0.15, the follow-up feedback quality is 0.75, the health record completeness is 0.82, the degree of abnormal indicators is 0.30, the degree of missing symptom reports is 0.10, the opportunity dimension score is 0.62, and the motivation dimension score is 0.68.

[0093] For profile features that change dynamically with the management process, update them according to the following formula: (17); In the formula, This indicates the updated dynamic profile features; Represents dynamic profile features for the current management cycle; This indicates the newly collected observation features for the next management cycle; This indicates that the coefficients are dynamically updated, and .

[0094] The updated multidimensional profile features are re-input into the profile stratification module, COM-B target identification module, and combined behavior prediction module. The system regenerates user profile tags, redetermines the case management stratification type, and reassesses whether patients exhibit false compliance, periodic apathy, high-risk status, or intervention gaps.

[0095] When the quality of patients' follow-up feedback, the completeness of health records, the scores of opportunity dimension and motivation dimension improve, and the degree of abnormal indicators and the degree of missing symptom reporting decrease, the system reduces the probability of false compliance, the probability of high-risk status and the probability of intervention gap, and adjusts the patient's case management stratification type from high-risk key management type to routine maintenance type or lower management intensity type.

[0096] The intervention decision-making module reduces high-frequency reminders and family collaboration tasks based on the updated calculation results, while retaining periodic follow-up reminders, symptom recording tasks, and low-intensity health education tasks, thereby avoiding continued over-intervention for patients with stable conditions and improved behavioral status.

[0097] In this embodiment, a multidimensional user profile is formed through equations (1) and (2), the case management stratification type is determined through equations (3) and (4), weak targets of COM-B behavior are identified through equations (5) and (6), pseudo-compliance, phased slackness, high-risk status and intervention gap are identified through equations (7) to (14), personalized intervention strategies are matched through equation (15), gamified nursing tasks are adjusted through equation (16), and the dynamic profile characteristics of the patient are updated through equation (17).

[0098] The above calculation process can also be used to execute the precise case management intervention method described above. When applied to other disease populations that require long-term follow-up, self-management or case management intervention, the clinical feature fields, profile stratification model parameters, risk judgment thresholds and intervention strategy library in formula (1) can be adjusted according to the disease characteristics, but the processing logic of user profile construction, COM-B target identification, combined behavior prediction, intervention strategy matching, gamified task adjustment and follow-up feedback update remains unchanged.

[0099] 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 precise case management and intervention system based on user profiling and COM-B behavioral model, characterized in that: include: The data acquisition module is used to collect clinical data, behavioral data, and COM-B three-dimensional assessment data of the target subjects. The COM-B three-dimensional assessment data includes capability dimension data, opportunity dimension data, and motivation dimension data. The user profile building module is used to preprocess and extract features from the clinical data, behavioral data, and COM-B three-dimensional assessment data, generate a multi-dimensional profile feature vector of the target object, and generate user profile tags based on the multi-dimensional profile feature vector. The profile layering module is used to input the multi-dimensional profile feature vector and user profile tags into a preset profile layering model and output the case management layering type corresponding to the target object. The COM-B target identification module is used to identify the target's behavioral weaknesses based on the assessment results of the target in terms of ability, opportunity, and motivation. The combined behavior prediction module is used to identify at least one of the target object's compliance status, apathy status, risk status, and intervention gap based on the combined relationship between multiple behavioral indicators of the target object. The intervention decision module is used to match intervention strategies from the intervention strategy library based on the case management stratification type, behavioral vulnerability targets, and combined behavior prediction results, and generate personalized case management intervention plans. The gamified intervention execution module is used to convert the personalized case management intervention plan into a gamified nursing task that can be executed on the terminal, and to adjust the subsequent nursing tasks according to the task execution results of the target object; The follow-up feedback optimization module is used to collect at least one of the following: follow-up data, scale data, and clinical indicators after intervention, and to update user profile tags, case management stratification types, and / or personalized case management intervention plans based on the collection results.

2. The precision case management and intervention system according to claim 1, characterized in that, The clinical data includes at least one of the following: disease type, disease stage, treatment stage, treatment plan, complication status, symptom information, and follow-up indicators; the behavioral data includes at least one of the following: health check-in records, medication or treatment cooperation records, follow-up response records, symptom reporting records, health education learning records, and lifestyle records.

3. The precision case management and intervention system according to claim 1, characterized in that, The user profile building module is used to process the collected data for missing value processing, outlier processing, standardization processing, and encoding processing. It then integrates the processed clinical features, behavioral features, and COM-B three-dimensional features to form a multi-dimensional profile feature vector for profile stratification and intervention decision-making.

4. The precision case management and intervention system according to claim 1, characterized in that, The profile stratification model is a machine learning model, a rule model, or a hybrid model formed by combining machine learning and rule models; the case management stratification type includes at least one of the following: routine maintenance type, pseudo-compliance type, phased slackness type, and high-risk key management type.

5. The precision case management and intervention system according to claim 1, characterized in that, The COM-B target identification module calculates the weakness scores corresponding to the capability dimension, opportunity dimension, and motivation dimension, and determines the dimensions with weakness scores reaching a preset threshold as priority intervention targets, so that the intervention content corresponds to the behavioral disorder causes of the target object.

6. The precision case management and intervention system according to claim 1, characterized in that, The combined behavior prediction module is used to perform combined analysis on at least two of the following: task completion frequency, number of consecutive interruptions, follow-up feedback quality, health record completeness, occurrence of abnormal indicators, and symptom reporting. Based on the combined analysis results, it determines whether the target subject has false compliance, periodic apathy, high relapse risk, or intervention gap.

7. The precision case management and intervention system according to claim 1, characterized in that, The intervention decision module is used to call the corresponding intervention strategy template according to different case management hierarchical types, and adjust at least one of the following in the intervention strategy template: nursing education content, follow-up frequency, reminder method, task intensity, risk warning level, and case management focus, based on the behavioral weakness target.

8. The precision case management and intervention system according to claim 1, characterized in that, The gamified nursing tasks include at least one of the following: challenge tasks, points tasks, level tasks, achievement tasks, and feedback tasks; the gamified intervention execution module dynamically adjusts the difficulty of subsequent tasks, the frequency of task pushes, or the content of reminders based on at least one of the target object's task completion rate, number of consecutive non-completions, abnormal indicator feedback, and follow-up response.

9. A precise case management and intervention method based on user profile layering and COM-B behavioral model, characterized in that, Includes the following steps: Collect clinical data, behavioral data, and COM-B three-dimensional assessment data from the target subjects; The clinical data, behavioral data, and COM-B three-dimensional assessment data are preprocessed and feature extracted to generate multi-dimensional profile feature vectors and user profile tags. Based on the multi-dimensional profile feature vector and user profile tags, determine the case management hierarchical type of the target object; The behavioral weaknesses of the target individuals are identified based on the assessment results of the capability, opportunity, and motivation dimensions. Identify at least one of the following in the target object’s compliance status, apathy status, risk status, and intervention gap based on the combined relationships between multiple behavioral indicators; Personalized case management intervention plans are generated based on the case management stratification type, behavioral vulnerability targets, and combined behavioral prediction results. The personalized case management intervention plan is converted into gamified nursing tasks and pushed to the terminal for execution; Collect at least one of the following data, scale data, and clinical indicators after intervention, and update user profile tags, case management stratification types, and / or personalized case management intervention plans based on the collection results.

10. The precision case management intervention system according to any one of claims 1-8 or the precision case management intervention method according to claim 9, characterized in that, The target population is colorectal cancer patients; or, the target population is other disease populations that require long-term follow-up, self-management, or case management intervention, and the system adapts to different disease populations by adjusting clinical data fields, profile stratification model parameters, and intervention strategy library.

Images