Medical rehabilitation nursing monitoring system for old people

Abstract

The invention relates to the technical field of service supervision and management systems, and discloses an elderly medical rehabilitation nursing monitoring system, which comprises the following steps: collecting a service request event and a paired service request release event; the mode analysis module is used for judging the stability of the system by analyzing the space-time mode of the request event; the service response analysis module is used for calculating the time difference between the request and the release event so as to evaluate the service response quality; and the cross validation module is used for calibrating the stability by using the service response quality when the system is stable so as to output a supervision signal representing the stable state quality. The technical problem that in the prior art, ordered stability under high efficiency and silent stability under request suppression caused by service delay cannot be distinguished is solved, and an objective decision basis is provided for managers in resource allocation and management intervention.

Description

Technical Field

[0001] This invention relates to an elderly medical rehabilitation nursing monitoring system, belonging to the technical field of service supervision and management systems. Background Technology

[0002] Currently, in the administrative management and supervision of elderly medical rehabilitation and nursing institutions, a service request system is typically configured. This system, with a call bell device as a typical example, is used to convert the immediate needs of the patients into discrete work task instructions. This is the basic component of the current service management process and provides clear operational guidance when processing independent requests to maintain the daily operation of the institution. However, the actual operation of a nursing unit is characterized by a continuous interaction between the supply of service resources and dynamic demands. When multiple service demands occur simultaneously in time and space, the total amount of immediate service demands within the unit will exceed the normal service supply capacity, leading to a general delay in service response. Since this imbalance in the service supply and demand relationship is not an independent event that can directly trigger a request, the existing management system can only record and count the discrete tasks that have been completed. It lacks corresponding monitoring and evaluation methods for the load status and changing trends of the entire unit's service capacity.

[0003] To address this issue, two common technical approaches are increasing staffing to supplement resources and utilizing video surveillance and other equipment for status monitoring. The former directly improves service supply capacity but increases operating costs; the latter introduces a new dimension of information collection, but its deployment complexity, the immediacy of data processing, and the protection of service recipients' privacy all limit its practical application. Furthermore, neither approach provides a method for directly quantifying the overall service load status of the nursing unit. Looking at other management systems in this field, even software platforms that focus on information fusion and process management often neglect dynamic analysis of the real-time load and response quality of the service system. For example, Chinese invention patent CN108648126A discloses an information control system and information data processing terminal integrating medical and elderly care. The core of the system lies in building a medical and elderly care information integration platform. By extracting data from the medical HIS system and elderly care institutions, a shared electronic health record database and physical examination database are established. Although it includes a supervision module and an evaluation support module, its evaluation is a periodic assessment of the elderly's behavioral abilities (such as daily living activities and mental state), and its supervision is also based on macro-management of static or quasi-static information such as records, appointments, and physical examinations. This design is essentially a data collection and record management system. It completely lacks the ability to perform real-time dynamic analysis of the core service process of service request response. It also cannot monitor the real-time spatiotemporal distribution pattern of service demand, nor can it quantify the real-time load pressure and response efficiency of the service system. Naturally, it cannot solve the key management problem that this invention aims to address: how to distinguish between high-efficiency stability and silent stability caused by service lag.

[0004] Analysis reveals the following unresolved issues with existing management technologies: 1. Lack of supervisory indicators for quantitatively assessing the overall service load of a nursing unit; 2. Inability to distinguish between sporadic service request concentrations and systemic service capacity saturation at the data level; 3. Resource allocation is based on statistics of completed discrete tasks, rather than real-time, continuous system load status. Therefore, the technical problem this invention aims to solve is how to utilize existing service request systems in service-oriented institutions to establish a technical method capable of quantitatively assessing the overall service load status of service units and generating corresponding supervisory signals, serving as a real-time basis for service resource allocation and management intervention. Summary of the Invention

[0005] This invention provides a monitoring system for elderly medical rehabilitation and nursing care. Its main purpose is to solve the problem that the existing technology lacks a technical method to quantitatively evaluate the overall service load status of the service unit and generate monitoring signals by utilizing the existing service request system.

[0006] To achieve the above objectives, the present invention provides an elderly medical rehabilitation and nursing monitoring system. This system includes a processor and a memory connected to the processor. The memory stores computer program instructions, which, when executed by the processor, implement the following modules: The data acquisition module is configured to collect service request events generated by multiple service request terminals within a nursing unit within a specified time, as well as service request cancellation events paired with the service request events. Each service request event includes a first timestamp and location information, and each service request cancellation event includes a second timestamp. The pattern analysis module is configured to construct a steady-state baseline model representing the spatiotemporal distribution pattern of normal service requests in a nursing unit based on the first timestamp and location information of historical service request events, and to compare newly generated service request events with the steady-state baseline model in real time to output a state signal representing the current steady-state level of the nursing unit. The service response analysis module is configured to determine a service quality benchmark based on the difference between the first timestamp of a historical service request event and the second timestamp of a paired service request cancellation event, and to calculate a service response index characterizing the dispersion of service response duration based on the difference between the first timestamp of a newly generated service request event and the second timestamp of a paired service request cancellation event. The cross-validation module is configured to follow preset validation rules: when the status signal output by the pattern analysis module is at a preset stability level, the service response index calculated by the service response analysis module is compared with the service quality benchmark, and when the service response index fails to reach the service quality benchmark, a steady-state quality warning signal that is different from the status signal is generated and output.

[0007] Preferably, the steady-state baseline model includes a temporal distribution model of service request events and a spatial distribution model of service request events; the pattern analysis module is configured to compare newly generated service request events with the steady-state baseline model to detect whether a pattern transition has occurred in the current service request pattern. The pattern transition includes the frequency of service request events being higher than the baseline frequency defined by the temporal distribution model, the time intervals between service request events being statistically dense, or the service request events being spatially concentrated in a local area defined by the spatial distribution model.

[0008] Preferably, the service response analysis module is configured to perform analysis within a preset statistical period. Internally, based on The response duration sequence formed by the difference between the first and second timestamps of the group. The service response index is calculated using the following formula. : ,in, For sequence number, , It is a positive integer. For statistical period The arithmetic mean of all response times.

[0009] Preferably, the system further includes: a context information interface module configured to obtain information on one or more planned events from an external scheduling system, wherein the planned event information includes an event time window and an area of ​​influence; and a pattern analysis module further configured to: determine whether the time and location of a newly generated service request event fall within the event time window and area of ​​influence of any planned event before comparing it with a steady-state baseline model; and if so, adjust the judgment logic used to detect whether the service request pattern deviates from the steady-state baseline model based on the type of the planned event.

[0010] Preferably, the pattern analysis module adjusts the judgment logic based on the type of planned event, including: temporarily increasing the trigger threshold for judging whether the frequency of service request events is higher than the baseline frequency by a preset ratio, or exempting specific types of pattern phase transitions during judgment.

[0011] Preferably, the system further includes: a micro-pattern analysis module, configured to analyze the number of service request events activated by the same service request terminal within a preset time window before the pattern analysis module makes comparisons; and to generate and output an emergency alarm signal that is distinct from the status signal and the steady-state quality warning signal when the number of events meets a preset emergency condition.

[0012] Preferably, the preset emergency condition is defined as: within a time window of 5 seconds, the number of service request events activated by the same service request terminal reaches or exceeds 3 times.

[0013] Preferably, the system further includes: a baseline monitoring module configured to monitor statistical features of state signals generated by the pattern analysis module that characterize the deviation of the nursing unit from the steady-state baseline model within a preset monitoring period; and a baseline management module configured to generate candidate steady-state baseline models based on service request events collected by the data acquisition module after the baseline drift conditions are met when the statistical features meet preset baseline drift conditions, and to provide a user interface for confirming whether to replace the current steady-state baseline model with the candidate steady-state baseline model.

[0014] Preferably, the baseline monitoring module is configured to calculate the proportion of the total time that the nursing unit is in a state of deviation from the steady state within the monitoring period to the total time of the monitoring period, as a statistical feature; the baseline drift condition is defined as the proportion being higher than 50% in three consecutive monitoring periods.

[0015] Preferably, the system further includes a status output module, which includes a visual management dashboard. The visual management dashboard is configured to display status signals with a first visual identifier, steady-state quality early warning signals with a second visual identifier, and emergency alarm signals with a third visual identifier, wherein the first visual identifier, the second visual identifier, and the third visual identifier are distinguishable from each other.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. By collecting time and location information of service request terminals, a steady-state baseline model representing the routine service request pattern of the nursing unit is constructed. Then, the continuous comparison between real-time service request events and this baseline model enables managers to identify pattern phase transitions formed by the spatiotemporal coupling of multiple isolated request events. This mechanism shifts managers from focusing on discrete, single service requests to perceiving the systemic pressure state of the entire nursing unit's service supply and demand relationship, providing an objective basis for pre-adjusting resources before service capacity approaches saturation.

[0017] 2. While establishing a steady-state baseline model for the nursing unit, this invention also incorporates service request cancellation events into the analysis. By calculating the time information between service request events and paired cancellation events, an objective index characterizing service response quality is obtained. When the system outputs a steady-state signal, this response index is used to calibrate the service quality of the steady-state baseline. This dual verification mechanism enables the system not only to identify disturbances in service request patterns but also to distinguish between orderly stability under high efficiency and silent stability under request suppression caused by service delays, thus avoiding misjudgments of service quality in management decisions.

[0018] 3. This invention introduces planned event information from an external scheduling system. Before performing pattern phase change detection, it pre-determines whether the current service request event falls within the time window and influence area of ​​known planned events, and dynamically adjusts the detection logic based on the event type. This links the administrative scheduling with the statistical model of service requests, enabling the system to adapt to different scenarios in judging service request patterns. It filters out pattern disturbances caused by foreseeable factors such as group activities, allowing the final output status signal to more accurately reflect unplanned system operation risks. Before performing overall pattern analysis of the nursing unit, a pre-processing micro-pattern analysis step is set up to specifically analyze the number of events activated by the same service request terminal within a preset time window. When this number meets the preset emergency conditions, the system generates and outputs an emergency alarm signal that differs from the regular status signal. This design enables the system to instantly capture and grade individual extreme crisis signals while monitoring the slow evolution of systemic risks, forming a complete monitoring closed loop for risks of different natures and time scales. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the system functional architecture and cross-validation logic of the present invention; Figure 2 This is a schematic diagram illustrating the basis for service quality indicator deviation and early warning in this invention; Figure 3 This is a diagram showing the physical deployment and hardware / software interaction architecture of the system of this invention. Detailed Implementation

[0020] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention is further described below with reference to specific illustrations. However, the following embodiments are only preferred embodiments of this invention and are not intended to limit the scope of protection of this invention. All equivalent changes or modifications made without departing from the spirit disclosed in this invention should be included within the scope of protection of this invention.

[0021] The elderly medical rehabilitation and nursing monitoring system disclosed in this invention is constructed as an administrative supervision and decision support tool deployed at the institutional management level. Its core objective is to transform the service operation status of nursing units from a management approach relying on subjective inspections and delayed reporting to a real-time, quantifiable monitoring system based on objective data. Physically, the system can be constructed from general-purpose server or computer hardware, including a processor and a memory connected to the processor. The memory stores computer program instructions executed by the processor, forming multiple functional units including a data acquisition module, a pattern analysis module, a service response analysis module, and a cross-validation module. These modules work collaboratively to provide managers with dual insights into the stability and effectiveness of the service system. During system operation, the data acquisition module is configured as the system's foundation. The basic information input terminal directly connects to existing service request terminals in the nursing facility, typically the most basic call bell device. This module operates continuously in bypass monitoring mode, utilizing the existing processor in the system to capture and structure the storage of two types of core events. The first type is service request events. When any terminal is activated, the module immediately records its [call initiation timestamp] (i.e., the first timestamp) and [call initiation location (room number)] (i.e., location information). The second type is service request cancellation events. When a nursing staff member completes the service on-site and resets the call, the module captures and records a [cancellation timestamp] (i.e., the second timestamp) and pairs this cancellation event with the previously corresponding service request event, thereby constructing a complete service request-response closed-loop event flow at the data layer. This data flow is the sole objective basis for all subsequent analyses.

[0022] The core task of the pattern analysis module is to provide managers with a basis for judging the macro-stability of nursing units. To this end, in the initial stage of system deployment, this module first executes a self-learning procedure for baseline model construction. It uses historical service request events accumulated by the data acquisition module within an initial calibration period (e.g., 72 hours of data collection) (using only their first timestamp and location information) to construct a dedicated steady-state baseline model for each independent nursing unit (e.g., the second floor, east wing of Building 3) that represents its normal service request pattern. This steady-state baseline model specifically includes two sub-models: a temporal distribution model of service request events, which quantifies the average call frequency at different times of the day (e.g., morning, afternoon, and night) and the statistical distribution of time intervals between call events (e.g., characteristic parameters of a Poisson distribution); and a spatial distribution model of service request events, which quantifies the historical distribution probability or heat map of calls in different rooms. After the construction is completed, the pattern analysis module enters a real-time monitoring state, comparing the newly generated service request event stream with the established steady-state baseline. The model performs continuous statistical comparisons to detect in real time whether the current service request pattern deviates from historical norms and undergoes a pattern shift. Pattern shifts specifically include several typical forms of accumulated system pressure, such as: frequency resonance, where the total call frequency within the current unit of time (e.g., 15 minutes) exceeds the normal fluctuation range of the historical baseline frequency defined by the time distribution model (e.g., exceeding the baseline mean by 3 standard deviations); temporal compression, where the time intervals between call events become statistically abnormally dense, for example, more than 5 consecutive call intervals are less than 20% of the historical average interval; or spatial collapse, where calls are abnormally concentrated in a local area defined by the spatial distribution model within a short period, for example, within 10 minutes, more than 40% of calls come from a few rooms that historically accounted for less than 10% of calls. When the module detects any pattern shift, it immediately generates and outputs a status signal characterizing the current steady-state level of the care unit, for example, defining the status signal as green (steady state), yellow (disturbance), or red (instability) for managers to make macro-level judgments.

[0023] Meanwhile, the service response analysis module operates independently. Its core task is to provide the system with an objective benchmark for service effectiveness to address potential quality limitations in the main solution, namely, its inability to distinguish between high-quality, orderly service and low-quality, stagnant service. To this end, this module first determines a historical service quality benchmark for the nursing unit based on historical data (i.e., the difference between the first timestamp of the service request event and the second timestamp of the paired service request cancellation event). For example, it calculates the unit's historical average response time. The timeframe is 180 seconds, and a benchmark metric characterizing the stability of historical responses. Subsequently, during system runtime, this module, based on the newly generated service request-response closed-loop event flow, performs statistical analysis within a preset period. Within a timeframe (e.g., a 1-hour sliding window), a service response metric is calculated in real-time to characterize the dispersion of current service response time. The variance, explicitly defined as the variance of the response duration series, is calculated using the following formula: ,in, For sequence number, , This represents the total number of closed-loop events within the statistical period. For the first The response time of a closed-loop event (i.e., the difference between the second and first timestamps), and For statistical period The arithmetic mean of all response times; The higher the value, the worse the reliability of the service response, that is, the greater the dispersion.

[0024] The cross-validation module performs secondary verification on the status signals from the pattern analysis module. It receives the status signals output by the pattern analysis module and the service response indicators calculated by the service response analysis module. This module is configured to follow preset administrative verification rules: when the status signal output by the pattern analysis module is at a preset stability level (e.g., the status signal is green (steady state)), this superficially indicates that the system is operating in an orderly manner. However, at this time, the cross-validation module immediately initiates a second verification, which checks the service response indicators calculated in real-time by the service response analysis module (e.g., the current...). ), compared with previously established service quality benchmarks (e.g., The comparison will be made; if the comparison reveals that the service response indicators do not meet (i.e. are inferior to) the service quality benchmark (e.g., ... If the current state is determined to be a quiescent steady-state state with management risks, the module will generate and output an independent steady-state quality warning signal that is different from the state signal. For example, it will be presented on the management dashboard as a green light with a faint blue light flashing around the outer ring, along with management insights (such as note: the service request mode of unit XX is stable, but the service response efficiency is consistently lower than the baseline, indicating potential operational risks), thereby preventing managers from making serious misjudgments about service quality.

[0025] To further enhance the accuracy of the monitoring system in real-world administrative environments, the system may also include a contextual information interface module to address potential contextual limitations arising from the main scenario's inability to understand planned events. This module is configured to retrieve information on one or more planned events from the organization's external scheduling system (e.g., an iCalendar subscription link or a structured text file), where the planned event information includes at least the event's [event time window] and [area of ​​impact] (e.g., Wednesday afternoon 2:00-3:00, rehabilitation hall). Furthermore, the internal logic of the pattern analysis module is configured to first determine the timing of newly generated service request events before comparing them with the steady-state baseline model. The system determines whether the time and location fall within the event time window and influence area of ​​any planned event; and if so, the pattern analysis module dynamically and temporarily adjusts the judgment logic used to detect pattern phase transitions based on the type of planned event. For example, if it is a group rehabilitation activity, the system can temporarily increase the trigger threshold for judging frequency resonance by a preset ratio (e.g., increase by 50%), or temporarily exempt (i.e., not alarm) the spatial collapse pattern occurring near the rehabilitation hall during the judgment. This measure filters out benign pattern disturbances caused by foreseeable factors by integrating administrative calendar information, enabling the system's early warning to more accurately focus on unplanned operational risks. To ensure the stability of the system in a real operating environment, the invention also includes a response mechanism for external system failures. In particular, when the context information interface module fails to obtain planned event information from the external scheduling system due to network interruptions, external interface changes, or unresponsive target servers, the module incorporates heartbeat detection or timeout judgment logic. If it fails to obtain external schedule data multiple times within a preset time period, it will proactively enter a context information missing fault-tolerant mode and alert the administrator with a specific indicator on the visual management dashboard. In this mode, the context information interface module will notify the pattern analysis module to suspend all dynamic adjustments to the judgment logic based on planned events and make judgments based on its internal steady-state baseline model. This ensures that the system's core monitoring functions will not be interrupted due to external dependency failures, but will automatically degrade to the basic monitoring module. This approach avoids inaccurate judgments and provides managers with a clear understanding of the system's state. Furthermore, to address the limitation of the pattern analysis module in perceiving systemic risks, which might overlook individual or acute crises, the system may also include a micro-pattern analysis module. This module acts as a pre-processor for the pattern analysis module, and its sole task is to analyze the number of service request events activated by the same service request terminal within a very short, preset time window (e.g., 5 seconds) before the pattern analysis module makes comparisons. When the number of events meets a preset emergency condition representing an urgent semantic, for example, it is defined as: within a 5-second time window, the number of service request events activated by the same service request terminal reaches or exceeds 3 times.At this point, the micro-pattern analysis module will immediately generate and output a highest-priority emergency alarm signal, distinct from the status signal and steady-state quality warning signal. This signal will bypass the conventional status lights and directly trigger an interruptible, location-specific audible and visual alarm, thus enabling the system to instantly capture individual extreme crisis signals while monitoring systemic risks.

[0026] Furthermore, to address the long-term operational issues of baseline drift or outdated knowledge faced by any system based on historical baselines, this system may also include a mechanism for adaptive evolution. This mechanism includes a baseline monitoring module that, instead of analyzing the original call signals, monitors the statistical characteristics of state signals generated by the pattern analysis module, characterizing the deviation of the care unit from the steady-state baseline model. For example, the baseline monitoring module is configured to calculate the proportion of the total time the care unit is in a deviation from the steady-state level (e.g., yellow or red) within a monitoring period (e.g., 72 hours) relative to the total monitoring period, and uses this as a statistical characteristic. When the statistical characteristic meets a preset baseline drift condition, for example, defined as the proportion being higher than a high threshold (e.g., 50%) for three consecutive monitoring periods, the system triggers the baseline drift hypothesis. At this time, the baseline management module is activated, which automatically starts a new learning instance in the background, generating a candidate steady-state baseline model based on service request events collected by the data acquisition module after the baseline drift condition is met (i.e., using only the latest data). Once the candidate model converges and stabilizes, the module provides a user interface on the management interface for managers (such as head nurses) to review, compare, and ultimately confirm whether to replace the current, potentially outdated, steady-state baseline model with the candidate model. This ensures the long-term timeliness and accuracy of the monitoring tool. To address potential delays or oversights by managers during the confirmation process, this invention incorporates a timeout and reminder mechanism: When the baseline management module generates a candidate model and prompts the user for confirmation, the system starts a preset confirmation timer (e.g., 72 hours). During this period, the system continues to use the current steady-state baseline model for analysis to ensure business continuity. If no user confirmation instruction is received after the timer expires, the system will prioritize the matter to be confirmed and remind the user with a more prominent pending task icon on the visual management dashboard. This aims to ensure that managers do not overlook this critical decision. The candidate model will be retained until the manager explicitly chooses to confirm replacement or ignore and abandon it. The system itself will not automatically update the baseline without authorization, thus guaranteeing the final control of management decisions.

[0027] Finally, to construct a monitoring interface that allows managers to take direct action, this system also includes a status output module, the core of which is a visual management dashboard. This dashboard is configured to integrate and display all the aforementioned monitoring signals using distinct visual identifiers of different priorities: for example, first visual identifiers (such as green / yellow / red zone status lights) display regular status signals; second visual identifiers (such as a faint blue flashing light around the outer ring of a green status light) display steady-state quality warning signals; and third visual identifiers (such as a full-screen red flashing light and a buzzer) display the highest-priority emergency alarm signals. This transforms a complex, multi-dimensional operational status into a hierarchical decision-making basis that managers can perceive immediately. To address the issue of emergency alarm signals, status signals, and steady-state quality warning signals potentially triggering simultaneously, the status output module follows a clear set of signal priorities and hierarchical display logic. The system pre-sets emergency alarm signals to have the highest priority, followed by status signals, with steady-state quality warning signals having the lowest priority. On the visual management dashboard, when multiple signals trigger simultaneously, they will... The system displays information in parallel across non-interfering visual areas and formats. For example, an emergency alarm (individual event) could occur in a unit where the overall service mode is green (steady state). At this time, the manager would simultaneously see a red flashing light for the corresponding room and a green status light representing the unit as a whole. If a steady-state quality warning signal is also triggered at this time, a blue halo would be superimposed on the outer ring of the green status light. This multi-layered information visualization strategy allows the manager to clearly interpret complex states on a single interface. For example, the unit as a whole may be stable but the response quality may be poor, and there may be an emergency in a certain room. This avoids the possibility of signal conflicts and ensures the accuracy of decision-making. The system of this invention can be implemented through software program modules running on general-purpose computing devices (such as servers, workstations, or embedded systems). General-purpose computing devices include, but are not limited to, processors for executing program instructions and memory (such as RAM, ROM, or hard disk drives) for storing computer program instructions. The functional logic of the modules can be implemented through programming languages ​​(such as Python, Java, or C++) combined with a database (for storing baseline models and event data).

[0028] Example 1: In a specific monitoring system application instance, a particular nursing unit A has a service request terminal whose total activation count over a continuous 72-hour period is consistently lower than 50% of the average of other nursing units within the same institution, and there is no obvious spatiotemporal concentration phenomenon. In traditional management reports, unit A is marked as a unit with light service load and stable operation due to its low request rate. After the monitoring system of the present invention is deployed in this scenario, its pattern analysis module constructs a corresponding steady-state baseline model based on the historical data of unit A (which already reflects this low-frequency request pattern) during the initial self-learning phase. Therefore, after switching to real-time monitoring, the pattern analysis module compares the currently generated service request event stream with the steady-state baseline model. No pattern phase change is detected, so it continuously outputs a state signal characterized as green (steady-state).

[0029] Meanwhile, the data acquisition module continuously collects a small number of complete service request events and paired service request cancellation events occurring within unit A; the service response analysis module, based on the difference between the first and second timestamps of these events, performs analysis within a preset statistical period. (For example Within this event, a service response index characterizing the dispersion of service response time is calculated. Its value is At the same time, the calculated arithmetic mean within the period The time interval was 580 seconds, indicating that although the number of requests was low, the response time for each request was extremely unstable and the average was too high. The cross-validation module was configured to follow its preset validation rules. When it received a green (steady-state) status signal from the pattern analysis module, it immediately initiated a second validation, applying the service response metrics calculated by the service response analysis module. Compared with a system-preset or calibrated service quality benchmark (e.g. ) for comparison; given If the service response indicators fail to meet the service quality benchmark, the cross-validation module determines the current state to be a high-risk quiescent steady state and immediately generates and outputs a steady-state quality warning signal that is distinct from the status signal. On the system's visual management dashboard, the manager of Unit A does not see a single green indicator indicating that everything is normal. Instead, the status output module of the dashboard is configured to display the composite status of the unit by superimposing a first visual indicator (green status light) and a second visual indicator (a faint blue flashing light around the outer ring of the green status light). This signal combination provides the manager with an objective, data-based decision-making basis, enabling them to focus their management attention on this seemingly stable unit with a serious lag in service effectiveness. This allows them to identify and intervene in operational risks caused by long-term insufficient allocation of service resources or failure of management processes.

[0030] Example 2: To objectively verify the effectiveness of the monitoring system of the present invention, especially its cross-validation module, in distinguishing steady-state conditions under different service quality levels, this example constructs a data simulation test platform. This platform utilizes a general-purpose processor to execute a discrete event simulation program. This program is configured to generate a data stream containing a service request event with [first timestamp, location information] and a service request cancellation event with [second timestamp] based on preset statistical parameters. To simulate different operational management states, the core control parameter of this simulation program is set as: average call interval time. (Used to control call frequency) and average service response time Meanwhile, by adjusting the statistical distribution of the response duration sequence, it can generate different service response indicators. The data stream represents the response time dispersion. The experiment included a control group (configured only with the pattern analysis module) and the prototype group (configured with the pattern analysis module, service response analysis module, and cross-validation module). The first phase of the experiment aimed to simulate a silent, steady-state nursing unit where, due to long-term insufficient service capacity, service recipients had suppressed non-urgent call requests. Therefore, the input parameters for the simulation platform were set as follows: Seconds (low call frequency) The system generates a highly discrete response sequence in seconds (high average response). The two system samples are configured to first use the 24-hour data generated by this parameter to construct their respective steady-state baseline models (at this time, the baseline model itself represents a low-quality normal state). Subsequently, the system enters a continuous 4-hour real-time monitoring period, and the simulation platform continues to generate event streams using the same parameters. During this monitoring period, the operating status of the two system samples is continuously recorded, and typical snapshot data of their key indicators are shown in Table 1.

[0031] Table 1: Silent Steady-State Simulation Test Data Table.

[0032] As shown in Table 1, the control group, because it only monitors spatiotemporal patterns, has low-frequency call flows that perfectly match its learned low-quality steady-state baseline model, thus continuously outputting a green (steady-state) signal. However, its output signal fails to reflect the true operational risk. The pattern analysis module of this invention also outputs a green (steady-state) signal, but its service response analysis module calculates a different real-time service response index. (For example, at T+1h) ) and its preset service quality benchmark ( The results were compared. The cross-validation module determined that this state indicated the service quality had not met the baseline, thus continuously triggering the steady-state quality warning signal. The second phase of the experiment aimed to simulate a highly efficient, steady-state nursing unit where service responses were timely and the routine needs of service recipients were met before calls were made, resulting in a low call frequency. Therefore, the input parameters of the simulation platform were adjusted as follows: Seconds (low call frequency) The system samples were processed in seconds (low average response) and a low-dispersion response sequence was generated. The two system samples (keeping the baseline model of the first stage unchanged) entered a 4-hour real-time monitoring period again. During this monitoring period, the operating status of the two system samples was continuously recorded, and typical snapshot data of their key indicators are shown in Table 2.

[0033] Table 2: High-efficiency steady-state simulation test data table.

[0034] As shown in Table 2, the control group still outputs a green (steady-state) signal; the pattern analysis module of the present invention's sample group also outputs a green (steady-state) signal, but at this time, the real-time service response index calculated by its service response analysis module is... (For example, at T+1h) It has exceeded the service quality benchmark. The cross-validation module determined that the service quality had reached the baseline, and therefore no steady-state quality warning signal was triggered. The test data showed that the monitoring system (control group) that only relied on the spatiotemporal pattern of service requests could not distinguish between two operating states with the same low service request frequency but completely different service quality at the management level. However, the sample of this invention, by introducing a service response analysis module and a cross-validation module, constructed a dual verification mechanism for service stability (state signal) and service effectiveness (steady-state quality warning signal), which enabled it to objectively and quantitatively identify the silent steady state caused by service delay.

[0035] Example 3: This example combines Figures 1 to 3 A description of a medical rehabilitation and nursing monitoring system for the elderly, such as... Figure 1As shown, service request terminals, such as call bell devices, generate request and cancellation events, which are collected by the data acquisition module. The collected data stream enters the service response analysis module to calculate the response time and output service indicators. On the other hand, before entering the pattern analysis module, a micro-pattern analysis module performs pre-analysis to capture individual emergency signals and directly output emergency alarm signals. Subsequently, the data stream enters the pattern analysis module to analyze spatiotemporal patterns and output status signals. The output signals from the pattern analysis module and the service response analysis module are jointly fed into the cross-validation module. This module performs dual validation to identify silent steady states and output steady-state quality warning signals. At the same time, a contextual information interface module subscribes to planned events from an external scheduling system, such as iCalendar, to dynamically adjust the judgment logic of the pattern analysis module to improve judgment accuracy. Finally, the multi-dimensional monitoring information composed of emergency alarm signals, status signals, and steady-state quality warning signals is uniformly presented by the status output module on a visual management dashboard.

[0036] like Figure 2 As shown, the left vertical axis represents the response time in seconds, and the right vertical axis represents... Metrics, the service response time sequence (dotted line) and service response metrics in the graph. (Solid line) Throughout the first to second-to-last hour of the monitoring period, the results were consistently higher than the benchmark for service quality. (Dashed line) This data format objectively and reproducibly presents a system state where service effectiveness has seriously deviated, thus providing direct quantitative evidence for triggering steady-state quality early warning signals. For example... Figure 3 As shown, this solution centers on the organization's internal network, transmitting request and cancellation events generated by multiple service request terminals, such as multiple call bell devices, to a general-purpose application server hardware running the core application of the monitoring system, including data acquisition modules, pattern analysis modules, cross-validation modules, status output modules, and other core modules. During operation, this application server acquires planned event information from an external scheduling system and performs data read and write operations with a database server that stores the system database, which contains steady-state baseline models, service request event stream data, and service quality benchmarks. The final generated dashboard data is then transmitted to the administrator workstation, which is configured with a visual management dashboard for administrators to use, thus forming a complete closed-loop monitoring system.

[0037] Example 4: This example aims to explain in detail the core model construction procedure, algorithm path, and key parameter calibration method in the monitoring system, ensuring that those skilled in the art can fully reproduce the solution without creative effort. In a specific system initialization deployment scenario, the pattern analysis module needs to first execute a steady-state baseline model construction procedure. The input of this procedure comes from the data acquisition module, and a preset calibration period (e.g., ...). The processor collects all historical service request event data streams within a day, and has the ability to perform statistical operations and matrix storage; the first step of the procedure is to perform spatiotemporal discretization, in which the processor divides the physical space of the nursing unit into sections based on location information (e.g., room number or bed number). A discrete, managerial spatial unit ( ), and at the same time, divide a 24-hour day into Equal time units (e.g., 1 hour intervals), Corresponding to 00:00-01:00, ..., (Corresponding to 23:00-24:00); the second step of the procedure is to construct the frequency matrix, which the processor traverses. All historical data for the day is compiled into each spatiotemporal cell. (i.e., spatial unit) In time unit within) in, Total number of service request events occurring within the day The third step of the procedure is to generate a baseline model, which the processor is based on. Calculate each spatiotemporal cell Historical average hourly call frequency This is made up of all Composition The matrix, which is stored as a steady-state baseline model, objectively quantifies operational routines such as room 105 should have an average of 0.7 calls per hour between 2 pm and 3 pm.

[0038] After the system enters real-time monitoring mode, the mode analysis module uses the established steady-state baseline model to perform algorithmic mode phase transition detection. This detection is performed within a preset sliding monitoring time window. (For example Runs within minutes; the processor first counts the data from the past... Within the window, in each spatial unit The actual number of service requests that occurred At the same time, the processor obtains the current time. Time unit (For example belong (i.e., 14:00-15:00), and extract the corresponding time units from the steady-state baseline model. baseline frequency vector Due to the baseline It is an hourly frequency, and the processor first converts it to a window frequency. Expected number of calls within Subsequently, the processor executes algorithmic judgments; for example, to detect frequency resonance, the processor calculates the total number of real-time calls in the current window. And calculate the total expected number of calls. ,like Greater than (in It is a configurable sensitivity coefficient, for example... This calculation is based on the statistical properties of the Poisson distribution to determine the occurrence of low-probability events, thus determining frequency resonance; to detect spatial collapse, the processor calculates each spatial unit. deviation If any of them exist Make Exceeding a preset local trigger threshold (e.g.) If the condition is met, then spatial collapse is determined to have occurred.

[0039] To eliminate service quality benchmarks in the service response analysis module To address the inherent arbitrariness in the settings, this embodiment provides a standardized calibration procedure: After system deployment, the nursing unit manager designates one or more operational cycles (e.g., two consecutive work weeks) as service quality calibration cycles, which are considered representative periods for service response levels; within this cycle, the service response analysis module records the response duration of all service request-response closed-loop events occurring within that cycle, forming a response duration sequence; after the calibration cycle ends, the processor calculates the response duration according to the formula... Calculate the variance of the sequence. The value (or the 85th percentile of the sequence taken to exclude extreme cases) is set and stored as the service quality benchmark specific to that nursing unit. For example, following this procedure, Unit A's identified as Unit B, with its varying management levels, identified as For the preset time window (5 seconds) and preset emergency condition (3 times) in the micro-pattern analysis module, and the preset monitoring period (72 hours), baseline drift condition (3 consecutive periods) and statistical characteristic threshold (above 50%) in the baseline monitoring module, these parameters are provided as the preferred default values ​​for the system in this embodiment. The setting is based on technical considerations of the trade-off between efficiency and reliability in management practice: for example, 3 times within 5 seconds is a balance between filtering 1-2 activations caused by accidental touches and instantaneously capturing continuous activation intentions caused by high-risk situations (such as falls or severe pain); while being above 50% for 3 consecutive 72-hour periods is a high-confidence statistical threshold set to trigger the baseline drift hypothesis, which is used to retain the system's ability to adapt to structural and long-term changes in the operating mode while ensuring the stability of the baseline model.

[0040] Example 5: In a specific system deployment scenario, to ensure that the context information interface module can adjust the judgment logic of the pattern analysis module for the operational activities of a specific organization, this example discloses a preliminary engineering calibration procedure. This procedure aims to establish a correlation between planned events in the external scheduling system and the adjustment of the judgment logic of the pattern analysis module. In the execution of this procedure, a nursing unit manager first selects a planned event with a high recurrence rate and known to cause deviations in service request patterns from the external scheduling system, such as a group rehabilitation activity on Wednesday afternoon from 14:00 to 15:00. Subsequently, within the next time window of the event, the manager temporarily places the pattern analysis module in context calibration mode through a management interface.

[0041] In this mode, the pattern analysis module continues to compare newly generated service request events with the steady-state baseline model in real time, but does not output yellow or red status signals. Instead, it performs the following steps: First, the processor continuously records the quantitative characteristics of the detected pattern phase transitions within the planned event time window of one hour. The processor calculates that frequency resonance is continuously triggered during this period, and the average total call frequency per unit time is 1.8 times the historical baseline frequency. Second, the processor simultaneously records the location information of the spatial collapse, finding that during this period, more than 70% of the service request events are concentrated at the entrance of the rehabilitation hall. Third, after the planned event time window ends, the processor summarizes the above-mentioned quantitative characteristics and generates an adjustment rule set associated with the event (collective rehabilitation activity). This rule set is defined as follows: when this event is activated, the trigger threshold used to judge frequency resonance will be temporarily increased to 2.0 times the baseline value; and when judging spatial collapse, events occurring at the entrance of the rehabilitation hall will be exempted. This adjustment rule set is stored as the basis for the pattern analysis module to perform adjustments to the judgment logic used to detect pattern phase transitions in the future. Thus, the system's adjustment to foreseeable operational disturbances is based on the unit's operational data, rather than the operator's preset.

[0042] Example 6: In a newly deployed nursing unit, to ensure that the emergency alarm signals of the monitoring system can accurately adapt to the unit's unique operating rhythm, an offline calibration procedure for calibrating the preset emergency conditions of the micro-pattern analysis module is executed. This procedure first uses a data acquisition module to collect raw event sequences activated by the same service request terminal within 24 hours. The head nurse manually marks the events in the sequence, classifying them as either true emergency (fall) or routine (accidental touch). Subsequently, the processor executes an optimization algorithm to determine a preset time window and a set number of activations. The constraints that this parameter combination should satisfy are: all events marked as true emergencies are correctly detected, while the detection rate of events marked as routine (accidental touch) is lower than a preset management tolerance, i.e., 5%. Through this procedure, the preset emergency condition for the unit is determined to be 3 times within 4.5 seconds, and this set of parameters is stored to replace the system default value.

[0043] To calibrate the preset baseline drift conditions for the baseline monitoring module, the execution of this procedure depends on the input of management objectives. First, the manager sets a minimum observation period for operational adjustments, i.e., the preset monitoring period, which is set to 72 hours. This value corresponds to the shortest time it is expected to see the effect after a management intervention (such as a shift adjustment). Then, the manager sets a management tolerance limit, i.e., a statistical characteristic threshold, which is set to 50%, representing the maximum tolerable deviation from steady state within an observation period. Finally, the manager sets a consecutive period count for confirming drift, i.e., the period count in the baseline drift conditions, which is set to 3 periods. This means that a structural drift is confirmed only when the system's operational status exceeds the management tolerance limit for 3 consecutive periods. These parameters (72 hours, 50%, 3 periods), input by the manager and reflecting the specific management strategy of the unit, are stored by the system and used as the basis for the baseline monitoring module's execution judgment.

[0044] To further verify the key technical effectiveness of the cross-validation module described in this invention in distinguishing different service quality steady states, this comparative example is set up.

[0045] Comparative Example 1: The conventional monitoring system used in this comparative example is configured to correspond to the technical solution of this invention after removing the service response analysis module and cross-validation module. This solution corresponds to the prior art in the background art, which can only statistically analyze discrete tasks or request patterns. This conventional system only includes a data acquisition module and a pattern analysis module, which is configured to compare the spatiotemporal distribution pattern of service request events with the steady-state baseline model to output a state signal characterizing the steady-state level of the nursing unit. This comparative example uses the same nursing unit A as in Example 1 as the test object, and inputs the real operational data stream collected in Example 1 into the conventional monitoring system. The core characteristics of this data stream, according to the description in Example 1, are as follows: Service request pattern: The frequency of service request events is consistently low (less than 50% of the institutional average); Service response quality: The service response time is extremely unstable. According to the calculation in Example 1, its average response time is... The service response time is 580 seconds. The value is 3100; the data acquisition module of the conventional monitoring system collects service request events (including the first timestamp and location information), but because it is not configured with a service response analysis module, it does not collect or process service request cancellation events. During execution, its pattern analysis module first constructs a steady-state baseline model representing the low-frequency request pattern unique to unit A based on the historical service request events in the data stream.

[0046] Subsequently, during the real-time monitoring phase, the pattern analysis module continuously compares newly generated service request event streams with the established steady-state baseline model. Since the real-time request pattern matches the historical baseline model, the system does not detect any statistical deviation or pattern shift. Because this conventional monitoring system does not include a service response analysis module for analyzing service request cancellation events, its technical solution itself cannot calculate service response efficiency (SRE) or average response time. The sole basis for judging the operating status of the nursing unit is the status signal output by the pattern analysis module. The test results are recorded as follows: Throughout the entire monitoring cycle of this conventional monitoring system, the status signal output by its status output module is always in a steady state (e.g., presented as a green indicator). This output result is consistent with the conclusion of the traditional management report mentioned in Example 1, both of which mark the nursing unit A as having a light service load and stable operation. The test results of this comparative example confirm that the conventional technical solution that relies solely on the spatiotemporal pattern of service request events for analysis cannot reflect the true operational quality and service risk when facing the silent steady-state scenario of request suppression caused by service delays in Example 1, thus leading to a fundamental misjudgment of service quality.

[0047] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0048] Finally, it should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A monitoring system for elderly medical rehabilitation and nursing care, characterized in that, The system includes a processor and a memory connected to the processor. The memory stores computer program instructions, which, when executed by the processor, implement the following modules: The data acquisition module is configured to collect service request events generated by multiple service request terminals within a nursing unit within a specified time, as well as service request cancellation events paired with the service request events. Each service request event includes a first timestamp and location information, and each service request cancellation event includes a second timestamp. The pattern analysis module is configured to construct a steady-state baseline model representing the spatiotemporal distribution pattern of normal service requests in a nursing unit based on the first timestamp and location information of historical service request events, and to compare newly generated service request events with the steady-state baseline model in real time to output a state signal representing the current steady-state level of the nursing unit. The service response analysis module is configured to determine a service quality benchmark based on the difference between the first timestamp of a historical service request event and the second timestamp of a paired service request cancellation event, and to calculate a service response index characterizing the dispersion of service response duration based on the difference between the first timestamp of a newly generated service request event and the second timestamp of a paired service request cancellation event. The cross-validation module is configured to follow preset validation rules: when the status signal output by the pattern analysis module is at a preset stability level, the service response index calculated by the service response analysis module is compared with the service quality benchmark, and when the service response index fails to reach the service quality benchmark, a steady-state quality warning signal that is different from the status signal is generated and output.

2. The elderly medical rehabilitation nursing monitoring system according to claim 1, characterized in that, The steady-state baseline model includes a temporal distribution model of service request events and a spatial distribution model of service request events. The pattern analysis module is configured to compare newly generated service request events with the steady-state baseline model to detect whether a pattern transition has occurred in the current service request pattern. A pattern transition includes service request events having a higher frequency than the baseline frequency defined by the temporal distribution model, statistically dense time intervals between service request events, or service request events being spatially concentrated in a local area defined by the spatial distribution model.

3. The elderly medical rehabilitation nursing monitoring system according to claim 1, characterized in that, The service response analysis module is configured to perform analysis within a preset statistical period. Internally, based on The response duration sequence formed by the difference between the first and second timestamps of the group. The service response index is calculated using the following formula. : ,in, For sequence number, , It is a positive integer. For statistical period The arithmetic mean of all response times.

4. The elderly medical rehabilitation nursing monitoring system according to claim 1, characterized in that, The system also includes: a context information interface module, configured to obtain information on one or more planned events from an external scheduling system, wherein the planned event information includes event time windows and influence areas; and a pattern analysis module, configured to: determine whether the time and location of a newly generated service request event fall within the event time window and influence area of ​​any planned event before comparing it with the steady-state baseline model; and if so, adjust the judgment logic used to detect whether the service request pattern deviates from the steady-state baseline model based on the type of the planned event.

5. The elderly medical rehabilitation nursing monitoring system according to claim 4, characterized in that, The pattern analysis module adjusts the judgment logic based on the type of planned events, including: temporarily increasing the trigger threshold for judging whether the frequency of service request events is higher than the baseline frequency by a preset ratio, or exempting specific types of pattern phase transitions during judgment.

6. The elderly medical rehabilitation nursing monitoring system according to claim 1, characterized in that, The system also includes a micro-pattern analysis module, which is configured to analyze the number of service request events activated by the same service request terminal within a preset time window before the pattern analysis module makes comparisons; when the number of events meets a preset emergency condition, it generates and outputs an emergency alarm signal that is distinct from the status signal and the steady-state quality warning signal.

7. The elderly medical rehabilitation nursing monitoring system according to claim 6, characterized in that, The default emergency condition is defined as: within a time window of 5 seconds, the number of service request events activated by the same service request terminal reaches or exceeds 3 times.

8. The elderly medical rehabilitation nursing monitoring system according to claim 1, characterized in that, The system also includes: a baseline monitoring module, configured to monitor statistical features of state signals generated by the pattern analysis module that characterize the deviation of the nursing unit from the steady-state baseline model within a preset monitoring period; and a baseline management module, configured to generate candidate steady-state baseline models based on service request events collected by the data acquisition module after the baseline drift conditions are met when the statistical features meet preset baseline drift conditions, and provide a user interface for confirmation of whether to replace the current steady-state baseline model with the candidate steady-state baseline model.

9. The elderly medical rehabilitation nursing monitoring system according to claim 8, characterized in that, The baseline monitoring module is configured to calculate, as a statistical feature, the proportion of the total time a nursing unit is in a state of deviation from a steady state within a monitoring period to the total time of the monitoring period. Baseline drift condition is defined as a proportion that is above 50% for three consecutive monitoring periods.

10. The elderly medical rehabilitation nursing monitoring system according to claim 1, characterized in that, The system also includes a status output module, which includes a visual management dashboard. This dashboard is configured to display status signals with a first visual identifier, steady-state quality warning signals with a second visual identifier, and emergency alarm signals with a third visual identifier, wherein the first, second, and third visual identifiers are distinguishable from each other.

Images