Intelligent follow-up visit robot system for transitional period after liver transplantation of children

Abstract

The invention relates to the technical field of medical software, and discloses an intelligent follow-up visit robot system for a posthepatic transplantation transition period of children, which comprises a robot body, a communication module, and a parent terminal and a transplantation center follow-up visit system which are in communication connection with the robot body, a compliance prediction module; an ecological instantaneous evaluation module; and the ecological instantaneous intervention module is used for automatically triggering intervention measures of corresponding levels according to the follow-up state evaluation result and a preset grading intervention strategy. The ecological instantaneous evaluation module and the ecological instantaneous intervention module work cooperatively based on the same real-time data stream, so that the intervention content, the intervention opportunity and the intervention intensity can be adaptively adjusted along with the dynamic change of the medication compliance and the psychological behavior state of the child patient.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical software, in particular to an intelligent follow-up robot system for children in the transition period after liver transplantation. BACKGROUND

[0002] Liver transplantation is currently recognized as an effective treatment for children with end-stage liver disease, which can significantly improve patient survival and quality of life. Due to the continuous breakthroughs in surgical technology and the optimization and upgrading of immunosuppressive treatment programs, the 1-year and 5-year survival rates of children after liver transplantation have significantly improved, and are currently at the world's leading position. For the management of children after liver transplantation, the focus of clinical attention is no longer limited to extending the patient's life cycle, but rather to improving the long-term quality of life of patients through systematic management and intervention.

[0003] Due to the dynamic nature of children's growth and development, the management of children after liver transplantation is already facing unique and complex challenges, especially when children enter puberty, they will experience rapid physiological changes and significant changes in self-awareness and emotional state on the psychological level. This double change greatly increases the difficulty of follow-up management at this stage. At the same time, puberty is a critical period for patients to transition from childhood to adulthood. During childhood, the parents are the main body of postoperative follow-up management, and the children do not need to bear much responsibility, but after entering adulthood, patients must gradually take on the responsibility of self-management, so the follow-up management in puberty not only needs to deal with the current management difficulties brought about by physiological and psychological changes, but also needs to help patients gradually establish self-management awareness and ability during this process, laying a good foundation for their long-term health in the future. The transition of liver transplant patients after surgery refers to helping patients and their parents change the follow-up management mode from parent responsibility management to patient self-management in a safe, simple and cooperative manner. This period is a high-risk period for liver transplant patients to develop adverse health outcomes, and is also an important period for caregivers to change roles. The entire transition period is divided into three stages. Adherence management is the core of transition follow-up management, and a decrease in adherence will increase the risk of acute rejection, loss of graft function, and even death in children.

[0004] The current traditional postoperative follow-up mode cannot reflect the dynamic changes in the daily health status of children in real time and lacks dynamic monitoring and timely intervention for changes in adherence during the transition period, so diversified, intelligent and systematic efforts are needed to ensure that the quality of follow-up management for this population meets the standards. SUMMARY

[0005] The main purpose of the present application is to solve the technical problem of low follow-up management quality in the prior art. An intelligent follow-up robot system for children in the transition period after liver transplantation, comprising: a robot body, a communication module, a parent terminal and a transplant center follow-up system in communication with the robot body. The robot body includes a processing unit, a sensing unit, and an interaction output unit. The system further includes a follow-up management software module running on the processing unit, comprising at least: The adherence prediction module is used to build a medication adherence prediction model based on the child's baseline scale data, adherence interview coding data and historical clinical follow-up data, and output the child's adherence risk assessment results. The ecological instantaneous assessment module is used to call the sensing unit to collect the child's medication-related behavioral data, emotional state data and / or physiological state data in real time when the compliance risk assessment result meets the preset triggering conditions, and to integrate and analyze the real-time collected data with the clinical indicators in the transplant center follow-up system to generate the child's current follow-up status assessment result. The ecological instantaneous intervention module is used to automatically trigger corresponding level intervention measures according to the follow-up status assessment results and a preset hierarchical intervention strategy, including: The primary intervention involves sending personalized medication reminders to the child through the interactive output unit. If primary intervention is ineffective, secondary intervention, such as monitoring and reminders, is initiated by linking the communication module to the parent's terminal. If secondary intervention is ineffective, an early warning message is sent to the transplant center's follow-up system to trigger tertiary intervention with manual intervention; The ecological instantaneous assessment module and the ecological instantaneous intervention module work together based on the same real-time data stream, enabling the intervention content, timing, and intensity to be adaptively adjusted according to the dynamic changes in the child's medication adherence and psychological and behavioral state.

[0006] The present invention has the following beneficial effects: The core advantage of the intelligent follow-up robot system of this invention lies in the construction of a new intelligent follow-up paradigm with an "assessment-intervention" closed loop, which realizes the leap from traditional extensive management to precise and personalized services.

[0007] I. Breaking through clinical bottlenecks and achieving precision management. The system captures real-time data on children's medication behavior, emotions, and physiology from multiple dimensions through ecological instantaneous assessment, far exceeding the static information of traditional follow-up. Based on this, ecological instantaneous intervention can implement personalized and scenario-based reminders and incentives, addressing the core challenge of low adherence during the adolescent transition period from the source, significantly reducing the risk of missed doses, and safeguarding long-term health after transplantation.

[0008] II. Innovative Human-Computer Interaction, Reshaping the Follow-up Experience. With gamified narrative and affective computing at its core, the tedious task of taking medication is transformed into an engaging interactive game, providing anthropomorphic emotional support. This effectively breaks down the psychological barriers of children with illnesses, transforming "passive treatment" into "active participation," and greatly enhancing user engagement and management efficiency.

[0009] Third, optimizing resource allocation and improving service efficiency. The robot handles routine follow-up work, freeing medical staff from repetitive tasks and allowing them to focus on high-risk case intervention, thus achieving precise allocation of medical resources. The system's low-bandwidth adaptability helps break down geographical barriers, facilitating the downward flow of high-quality medical resources and yielding significant social benefits. In summary, this invention is not only a technological innovation but also an upgrade to the nursing model, providing a replicable and scalable systematic solution for the long-term management of pediatric liver transplantation and other chronic diseases. Attached Figure Description

[0010] Figure 1 This is a structural diagram of the intelligent follow-up robot system for the transition period after pediatric liver transplantation provided by the present invention. Detailed Implementation

[0011] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention 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 described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a 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.

[0012] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The embodiments of the intelligent follow-up robot system for the transition period after pediatric liver transplantation in this invention include: The system includes a robot body, a communication module, a parent terminal and a transplant center follow-up system that are communicatively connected to the robot body; wherein the robot body includes a processing unit, a sensing unit and an interactive output unit. The system further includes a follow-up management software module running on the processing unit, comprising at least: The adherence prediction module is used to build a medication adherence prediction model based on the child's baseline scale data, adherence interview coding data and historical clinical follow-up data, and output the child's adherence risk assessment results. The ecological instantaneous assessment module is used to call the sensing unit to collect the child's medication-related behavioral data, emotional state data and / or physiological state data in real time when the compliance risk assessment result meets the preset triggering conditions, and to integrate and analyze the real-time collected data with the clinical indicators in the transplant center follow-up system to generate the child's current follow-up status assessment result. The ecological instantaneous intervention module is used to automatically trigger corresponding level intervention measures according to the follow-up status assessment results and a preset hierarchical intervention strategy, including: The primary intervention involves sending personalized medication reminders to the child through the interactive output unit. If primary intervention is ineffective, secondary intervention, such as monitoring and reminders, is initiated by linking the communication module to the parent's terminal. If secondary intervention is ineffective, an early warning message is sent to the transplant center's follow-up system to trigger tertiary intervention with manual intervention; The ecological instantaneous assessment module and the ecological instantaneous intervention module work together based on the same real-time data stream, enabling the intervention content, timing, and intensity to be adaptively adjusted according to the dynamic changes in the child's medication adherence and psychological and behavioral state.

[0013] In an embodiment of the present invention, the compliance prediction module is configured to perform the following steps: Collect multi-source baseline assessment data of the child, which includes at least immunosuppressive drug adherence assessment scale data, medication behavior assessment scale data, and scale data reflecting the child's self-management ability or disease cognition status. Feature engineering is performed on scale data from different sources, including numerical standardization and feature vectorization. The processed baseline assessment data is stored in the patient's archive to serve as the input feature set for the compliance prediction model.

[0014] In an embodiment of the present invention, the compliance prediction module further includes an interview data processing step, the interview data processing step comprising: According to the preset interview coding system, the follow-up interview content of the child or his / her guardian is semantically transcribed to obtain transcribed text; the transcribed text is then used to extract features from the transcribed text using natural language processing methods to generate interview coding feature vectors that reflect medication resistance, family support and psychological state; the interview coding feature vectors and the baseline assessment data are used together as inputs to the compliance prediction model.

[0015] In an embodiment of the present invention, the adherence prediction module constructs a machine learning-based medication adherence prediction model by integrating baseline assessment data, interview coding data, and historical clinical follow-up data of the child, and outputs prediction results characterizing the risk of medication adherence in the child.

[0016] In an embodiment of the present invention, the compliance prediction module uses the coefficient of variation of immunosuppressant trough concentration as an objective compliance evaluation index, and feeds the coefficient of variation of trough concentration as label data back to the compliance prediction model to update the model parameters online.

[0017] In an embodiment of the present invention, the instantaneous ecological assessment module is configured as follows: At each assessment time point, an immediate compliance risk score is generated for the child, and the ecological instantaneous assessment process is triggered when at least one of the following conditions is met: the immediate compliance risk score exceeds a preset risk threshold; or the change in the risk score at the current assessment time point relative to the risk score at the previous time point exceeds a preset change threshold.

[0018] In an embodiment of the present invention, the ecological transient assessment module collects the child's ecological transient data through multimodal sensing, and the ecological transient data includes at least: Emotional state data obtained through image or voice data recognition; Physiological parameter data acquired through wearable devices; Medication-related behavioral data obtained through robot interaction records.

[0019] In an embodiment of the present invention, the ecological instantaneous assessment module performs time-series modeling processing on the collected multimodal ecological instantaneous data to obtain the emotional state vector of the child, and uses the emotional state vector as an important input parameter for follow-up status assessment.

[0020] In an embodiment of the present invention, the ecological transient assessment module integrates the multimodal ecological transient data, the coefficient of variation of immunosuppressant trough concentration, and liver function test indicators with the patient's historical clinical records to output discrete follow-up status assessment results, which include at least good, fair, poor, and critical levels.

[0021] In an embodiment of the present invention, the content of the primary intervention is generated by a personalized generation module. The personalized generation module dynamically generates medication reminder methods, incentive feedback forms, and interaction difficulty levels based on the child's personal file, current follow-up status, and interaction preferences.

[0022] Specifically, the solution of the present invention is as follows: Adherence management is the most critical follow-up management target throughout the transition period. Poor medication adherence is one of the main factors leading to long-term mortality after pediatric liver transplantation. Failure by parents or guardians to administer medication on time and in the prescribed dosage often leads to acute rejection and chronic infection, becoming a major cause of re-transplantation and graft failure. For children outside the transplant center's location, current follow-up management resources are insufficient. Common methods to improve medication adherence in pediatric liver transplant recipients include health education, behavioral and visual reminders, digital reminders, family involvement, and intensive follow-up. However, these methods face challenges such as difficulty in converting into long-term behaviors, declining adherence over time, equipment / adherence fatigue, parental stress, and parent / child asynchrony. A diversified, intelligent, and systematic approach is urgently needed. To effectively address this issue and improve medication adherence in children, providing support for children and their parents during the transition period, this solution integrates deep learning algorithms, natural language processing, and affective computing technologies to develop an intelligent follow-up management robot specifically designed for pediatric and adolescent liver transplant patients, combining EMA and EMI technologies. The system is designed with gamification as its core concept, breaking down patients' psychological barriers through numerous innovative modules to improve medication adherence and management effectiveness. A successful transition is crucial for the continued health of pediatric liver transplant patients. An effective transition program includes many key components, achieving individualized management through EMA and EMI collaboration to optimize function within a given transplant center. This invention utilizes an intelligent robot to achieve systematic transition management and "emotional-like interaction," stabilizing and improving follow-up adherence in children, and enhancing long-term clinical outcomes and quality of life for liver transplant recipients.

[0023] This robot is designed for post-liver transplant medication management in children. It constructs an integrated architecture of "hardware-software-multi-terminal collaboration" and deeply integrates EMA and EMI to achieve closed-loop management from perception to intervention.

[0024] A: Hardware layer: The robot body integrates multimodal sensors, processing chips, built-in programs and corpora, and is equipped with an integrated medicine box, moving wheels and power switch. It supports linkage with peripherals to complete command reception and real-time feedback.

[0025] B: Software Layer: Centered on medication adherence management, it organically integrates EMA and EMI mechanisms. Through sensors and sensing modules, it assesses children's medication behavior, emotional state, and physiological indicators in real time, triggering personalized and contextualized intervention strategies, including intelligent reminders, incentive feedback, and multi-level early warnings. Simultaneously, it seamlessly connects to the transplant center's follow-up system, enabling remote collaboration and data exchange.

[0026] Medication adherence is the core of management. 1. Establish a predictive model for medication adherence 1.1 First, a baseline survey was conducted on the children using structured scales such as the Immunosuppressive Drug Adherence Assessment Scale and the Morisky Drug Adherence Scale. The survey results were stored and used as important variables for modeling.

[0027] 1.2 Secondly, compliance interviews were conducted, and the interview results were encoded and transcribed as one of the modeling variables.

[0028] 1.3 Use different scales to enrich the predictive model, such as recognized scales for self-management ability, disease identification, and self-efficacy, and store the obtained assessment data.

[0029] 1.4 The coefficient of variation of trough concentration of immunosuppressants was used as an objective indicator for evaluating medication adherence. The coefficient of variation of trough concentration = (standard deviation of trough concentration / mean of trough concentration) * 100%.

[0030] 2. Implementation of EMA 2.1 Predict medication adherence of children using predictive models.

[0031] 2.2 The child's emotional state and physiological indicators are assessed in real time through sensors and sensing modules.

[0032] 2.3 The assessment data will be comprehensively analyzed with the patient's personal data (such as liver function test data at each follow-up visit since the transplant surgery) in the transplant center's follow-up database to give the final assessment results.

[0033] 3. Implement EMI 3.1 Use game mechanics to interact with the child, issue medication reminders, and implement primary intervention.

[0034] 3.2 If the first-level intervention is ineffective, the parents will be notified via a wearable smart device, and the parents will be responsible for supervising medication administration, thus implementing the second-level intervention.

[0035] 3.3 If the secondary intervention is ineffective, the transplant center's follow-up management personnel will be notified via the terminal, and professional personnel will intervene to implement the tertiary intervention.

[0036] In response to the psychological and behavioral characteristics of children of different age groups—young children rely on direct experiences, school-aged children have limited cognition, and adolescents have strong autonomy and are prone to resistance—this system, based on the EMA / EMI collaborative mechanism, achieves the breaking down of psychological barriers and the improvement of compliance through the following modules: 1. Deep integration of gamification mechanics with EMA / EMI With gamification as the core design concept, EMA is embedded in interactive plots, and EMI is integrated into game rewards, forming a "assessment-intervention" closed loop: (1) Virtual Partners and Narrative Intervention (a) Design Principles: An AI virtual companion with emotional feedback is constructed. A generative adversarial network (GAN) generates personalized storylines, embedding tasks such as medication reminders and symptom reporting into the narrative flow. While monitoring participation in EMA (Emergency Medical Efficacy Assessment), EMI (Emergency Medical Efficacy Assessment) is achieved through plot progression, improving treatment adherence. The GAN consists of two independent neural networks: a generator and a discriminator. After the intelligent follow-up robot inputs the child's personal data and real-time EMA data, the generator creates a customized storyline for the child. The discriminator then judges whether the generated storyline is "realistic, reasonable, and interesting." Through this adversarial process, the generator eventually learns to automatically create: ① Emotional connection: When EMA detects that the child is depressed, it generates encouraging and supportive scenarios; ② Cognitive matching: Generate tasks and dialogues of varying complexity for children of different ages; ③ Behavioral guidance: Seamlessly and interestingly integrate medical tasks such as medication and symptom reporting into the main storyline.

[0037] (b) EMA Integration: During narrative interaction, EMA uses multimodal data, such as speech response time, task completion rate, and facial expressions, to assess the patient's engagement and emotional changes in real time. This data is used to dynamically adjust the narrative pace and content difficulty to ensure that the intervention matches the patient's state.

[0038] (c) EMI Triggering: Based on EMA data, EMI is embedded in the game reward mechanism. For example, after completing a medication task, the patient receives a virtual reward, which drives the story forward and forms a positive reinforcement loop, which is in line with the "positive reinforcement" principle in behaviorist theory and can effectively promote medication adherence in children.

[0039] 2. Adaptive Interaction and Cognitive Matching (1) Design Principles: To break down psychological barriers among children of different ages and achieve precise intervention, this system adopts a reinforcement learning framework as its intelligent decision-making engine. This framework constructs follow-up interactions as a continuous learning sequential decision-making process, and its core implementation mechanism is as follows: First, the system constructs the RL state space from the multidimensional data collected in real time by EMA—including the child's age, cognitive level scale score, real-time emotion recognition results, and interaction behavior history. This state space constitutes the system's dynamic and quantitative perception of the child's current situation. Based on the perception of the current state, the system needs to select and execute the optimal action from the action space (i.e., the pre-set ecological instantaneous intervention strategy library).

[0040] To enable the system to learn from the results of interactions, we designed a sophisticated reward function. This function transforms abstract clinical goals into computable scalar signals: the system receives a positive reward when the EMA detects that the child takes medication on time, is in a positive mood, or is deeply engaged in the interaction; and a negative reward is generated when resistance, missed medication, or deteriorating mood is detected.

[0041] (2) Age stratification strategy: (a) Young children (3-6 years old): Rely on sensory experiences. Medication administration is completed through auditory and visual rewards, such as colored lights and cheerful sound effects emitted by robots, and tactile feedback. EMA monitors medication behavior through sensors, triggering simple EMI. According to Piaget's theory of cognitive development, children in this age group are in the preoperational stage and require visual stimulation to reinforce their behavior; (b) School-aged children (7-12 years old): Cognitive development, enjoys challenges. Design a robot growth system where behaviors such as taking medication accumulate "energy" to unlock new functions or appearances. EMA tracks the frequency of behavior, EMI provides progress feedback and virtual rewards, and the achievement system enhances a sense of competence; (c) Adolescents (13-18 years old): Strong sense of autonomy, socially oriented. Embedded social modules allow sharing progress with peers or participating in team challenges. EMA monitors social interactions and emotional data, while EMI provides personalized challenges or peer support.

[0042] 3. Real-time monitoring and tiered intervention of psychological state By deeply integrating multimodal perception and real-time intervention capabilities, and combining computer vision (based on a facial motion coding system) with speech emotion analysis (analyzing pitch, speech rate, and semantic content), the system captures children's facial expressions, speech features, and interactive behavior data in real time, forming a dynamic, instantaneous ecological assessment stream. This high-dimensional temporal data is input into a Long Short-Term Memory (LSTM) network model. LSTM, with its unique gating mechanism, effectively learns the temporal dependence of facial expressions and speech, thereby accurately identifying emotional fluctuation trends and risk levels. Based on the LSTM's output of emotional risk quantification, the system initiates a tiered, instantaneous ecological intervention mechanism: when mild anxiety is identified, built-in soothing tasks are automatically triggered, such as guided breathing exercises or soothing music, to help children self-regulate their emotions; once the model determines severe resistance or a high-risk emotional state, the intervention level is immediately escalated, the system automatically interrupts the current automated process, and seamlessly transfers to manual follow-up, ensuring timely intervention by professional medical staff. This constructs a closed-loop management system from "intelligent perception" to "precise tiered intervention." 4. The liver donor follow-up module extends to EMA / EMI applications. Extending the EMA / EMI mechanism to postoperative management of living donor liver transplant recipients: (a) Intelligent follow-up reminder Using temporal convolutional networks to analyze liver donor recovery indicators and the treatment stages of children, we can predict key follow-up milestones and implement dynamic priority reminders.

[0043] (b) Tiered psychological intervention By combining micro-expression analysis and speech-semantic fusion detection, the psychological state of liver donors can be assessed in real time. Risk stratification is constructed through GNN, and differentiated EMI is implemented: low-risk patients are pushed with mindfulness training, while medium- and high-risk patients are triggered to receive human intervention and generate family communication suggestions.

[0044] (c) Human-computer emotional dialogue We designed an emotionally supportive dialogue agent, employing generative dialogue and empathic language models to alleviate the psychological stress of liver donors and achieve continuous emotional support.

[0045] 5. Wearable devices collaboratively enhance EMA capabilities The robot continuously monitors 11 parameters, including heart rate variability, blood oxygen, and blood pressure, using devices such as medical-grade smart bracelets. Motion compensation algorithms ensure data accuracy, expanding the robot's physiological EMI (extra-invasive prenatal testing) capabilities. Data is transmitted to the robot via 5G, where a risk identification system determines whether to initiate EMI or require manual intervention.

[0046] The detailed operating steps are as follows: 1. Command Reception and Connection: The robot's main program connects to the parent's mobile app to receive commands such as medication time and type of medication.

[0047] 2. Medication reminder trigger: When the set time arrives, the robot will issue a medication reminder (such as a voice prompt, light prompt, etc.).

[0048] 3. Medication Status Detection: Sensors detect whether the child has taken their medication. If medication has been taken: The robot sends encouraging messages (such as "You're great, take your medication on time and you'll recover quickly"), and then completes the instruction. If medication has not been taken: The robot continues to send reminders, and the parent's app receives feedback that medication has not been taken.

[0049] 4. Secondary Sensing and Multi-Terminal Interaction: The robot senses the absence of medication again. If the child has taken the medication, it executes the completion command. If the child still hasn't taken the medication, the robot reminds the parents again and simultaneously notifies the transplant center's follow-up management personnel remotely through the follow-up system to ensure timely medical intervention. This intelligent robot can interact with the child through three interaction modes to ensure successful medication administration.

[0050] ① Reminders and Interactions: Medication reminders are initiated through multimodal methods such as voice and light, delivering information in a way that is easy for children to accept.

[0051] ②Incentive interaction: After the child takes the medication on time, personalized encouraging words are sent through a pre-set corpus to reinforce the child's positive behavior of taking the medication.

[0052] ③ Feedback and interaction: The system uses sensors to monitor medication behavior in real time and transforms the results into interactive feedback (encouragement or continuous reminders), forming a positive cycle of "behavior-feedback" to improve the child's attention to and cooperation with medication.

[0053] This invention establishes a medication adherence prediction model by integrating multi-source baseline assessment data, interview coding data, and historical clinical follow-up data of children. It also introduces the coefficient of variation of immunosuppressant trough concentration as an objective feedback indicator to continuously update the model, enabling the system to identify the decline in adherence before obvious abnormalities in medication behavior, thereby overcoming the shortcomings of existing technologies that only intervene passively after problems occur.

[0054] By using an ecological instantaneous assessment mechanism, data on children's medication behavior, emotional state, and physiological parameters are collected in real time in natural living scenarios. This significantly reduces the information distortion caused by recall bias and insufficient assessment frequency in traditional periodic follow-ups, enabling follow-up results to truly reflect the dynamic health status of children.

[0055] This invention integrates multimodal ecological instantaneous data with immunosuppressant trough concentrations, liver function test indicators, and previous clinical records to form a graded follow-up status assessment result, effectively avoiding misjudgments caused by single indicator assessment and improving the accuracy of identifying high-risk status.

[0056] By setting up a tiered intervention strategy consisting of real-time robot intervention, parental collaborative supervision, and human intervention from the transplant center, this invention can automatically adjust the intervention level according to changes in the child's compliance status, avoiding over-intervention or under-intervention, and ensuring that the intervention measures match the risk level.

[0057] Based on the child's age characteristics, behavioral history, and interaction preferences, this invention dynamically generates appropriate medication reminders and incentive feedback methods, ensuring that the intervention content remains novel and relevant during long-term follow-up, effectively reducing the problem of declining adherence over time.

[0058] By incorporating compliance prediction results, instantaneous ecological assessment data, and intervention feedback results into the same follow-up process, this invention constructs a sustainable self-adjusting closed-loop management mechanism, enabling the follow-up system to continuously optimize management strategies as the child's behavior and condition change.

[0059] This invention utilizes an intelligent follow-up robot to perform high-frequency, repetitive follow-up assessments and primary intervention tasks. It only triggers human intervention when a persistent abnormality or high-risk condition is detected, thereby reducing ineffective follow-ups and improving the efficiency of medical resource utilization. It is especially suitable for remote follow-up management in non-transplant center areas.

[0060] By continuously improving medication adherence during the transition period and promptly identifying high-risk conditions, this invention helps reduce the probability of adverse outcomes such as acute rejection and graft dysfunction, thereby improving the long-term quality of life for pediatric liver transplant patients from a technical perspective.

[0061] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smart follow-up robot system for the transition period after pediatric liver transplantation, characterized in that, The system includes a robot body, a communication module, and a parent terminal and transplant center follow-up system that are connected to the robot body in communication. The robot body includes a processing unit, a sensing unit, and an interaction output unit. The system further includes a follow-up management software module running on the processing unit, comprising at least: The adherence prediction module is used to build a medication adherence prediction model based on the child's baseline scale data, adherence interview coding data and historical clinical follow-up data, and output the child's adherence risk assessment results. The ecological instantaneous assessment module is used to call the sensing unit to collect the child's medication-related behavioral data, emotional state data and / or physiological state data in real time when the compliance risk assessment result meets the preset triggering conditions, and to integrate and analyze the real-time collected data with the clinical indicators in the transplant center follow-up system to generate the child's current follow-up status assessment result. The ecological instantaneous intervention module is used to automatically trigger corresponding level intervention measures according to the follow-up status assessment results and a preset hierarchical intervention strategy, including: The primary intervention involves sending personalized medication reminders to the child through the interactive output unit. If primary intervention is ineffective, secondary intervention, such as monitoring and reminders, is initiated by linking the communication module to the parent's terminal. If secondary intervention is ineffective, an early warning message is sent to the transplant center's follow-up system to trigger tertiary intervention with manual intervention; The ecological instantaneous assessment module and the ecological instantaneous intervention module work together based on the same real-time data stream, enabling the intervention content, timing, and intensity to be adaptively adjusted according to the dynamic changes in the child's medication adherence and psychological and behavioral state.

2. The intelligent follow-up robot system according to claim 1, characterized in that, The compliance prediction module is configured to perform the following steps: Collect multi-source baseline assessment data of the child, which includes at least immunosuppressive drug adherence assessment scale data, medication behavior assessment scale data, and scale data reflecting the child's self-management ability or disease cognition status. Feature engineering is performed on scale data from different sources, including numerical standardization and feature vectorization. The processed baseline assessment data is stored in the patient's archive to serve as the input feature set for the compliance prediction model.

3. The intelligent follow-up robot system according to claim 2, characterized in that, The compliance prediction module further includes an interview data processing step, which includes: According to the pre-set interview coding system, the follow-up interview content of the child or his / her guardian was semantically transcribed to obtain the transcribed text; Natural language processing methods were used to extract features from the transcribed text to generate interview-coded feature vectors that reflect medication addiction, family support, and psychological state. The interview-encoded feature vectors and the baseline assessment data are used together as inputs to the compliance prediction model.

4. The intelligent follow-up robot system according to claim 3, characterized in that, The adherence prediction module integrates baseline assessment data, interview coding data, and historical clinical follow-up data of the children to construct a machine learning-based medication adherence prediction model and outputs prediction results that characterize the risk of medication adherence in children.

5. The intelligent follow-up robot system according to claim 4, characterized in that, The compliance prediction module uses the coefficient of variation of immunosuppressant trough concentration as an objective compliance evaluation indicator, and feeds the coefficient of variation of trough concentration as label data back to the compliance prediction model to update the model parameters online.

6. The intelligent follow-up robot system according to claim 1, characterized in that, The ecological instantaneous assessment module is configured as follows: An immediate compliance risk score is generated for the child at each assessment time point, and the ecological instantaneous assessment process is triggered when at least one of the following conditions is met: The immediate compliance risk score exceeds a preset risk threshold; The risk score at the current assessment point has changed more than the risk score at the previous point in time than the preset change threshold.

7. The intelligent follow-up robot system according to claim 6, characterized in that, The instantaneous ecological assessment module collects the child's instantaneous ecological data through multimodal sensing, and the instantaneous ecological data includes at least: Emotional state data obtained through image or voice data recognition; Physiological parameter data acquired through wearable devices; Medication-related behavioral data obtained through robot interaction records.

8. The intelligent follow-up robot system according to claim 7, characterized in that, The ecological instantaneous assessment module performs time-series modeling processing on the collected multimodal ecological instantaneous data to obtain the emotional state vector of the child, and uses the emotional state vector as an important input parameter for follow-up status assessment.

9. The intelligent follow-up robot system according to claim 8, characterized in that, The ecological instantaneous assessment module integrates and analyzes the multimodal ecological instantaneous data, the coefficient of variation of immunosuppressant concentration, and liver function test indicators with the child's historical clinical records to output discrete follow-up status assessment results, which include at least good, fair, poor, and critical levels.

10. The intelligent follow-up robot system according to claim 9, characterized in that, The content of the primary intervention is generated by a personalized generation module, which dynamically generates medication reminder methods, incentive feedback formats, and interaction difficulty levels based on the child's personal file, current follow-up status, and interaction preferences.

Images