Intelligent sensing dressing system for multi-mode wound state monitoring and active intervention

Abstract

The invention discloses an intelligent sensing dressing system for multi-modal wound state monitoring and active intervention, which relates to the field of medical equipment and comprises a flexible multi-modal sensing layer, a microenvironment regulation dressing layer, an integrated flexible electronic control unit, a local intelligent terminal module and a cloud data platform. The system can collect wound pressure, temperature, pH value and blood flow parameters in real time, operate an artificial intelligence risk prediction algorithm to generate risk levels and nursing suggestions, and send intervention prompts when preset thresholds are reached. The cloud platform realizes data storage, trend analysis and risk alarm pushing, supports remote monitoring and cooperation, and improves wound management efficiency and healing quality.

Description

Technical Field

[0001] This specification relates to the field of medical devices, and more specifically, this application relates to an intelligent sensing dressing system for multimodal wound condition monitoring and active intervention. Background Technology

[0002] Pressure injury (PI), also known as pressure ulcer, is a common clinical problem caused by sustained pressure or the combined effect of pressure and shear force, resulting in localized damage to the skin and / or subcutaneous soft tissues. It commonly occurs at bony prominences and may also be related to contact sites with long-term medical devices. Its pathogenesis is closely related to local tissue blood flow obstruction, ischemia-reperfusion injury, tissue temperature changes, humidity imbalance, and infection processes. Elderly individuals, long-term bedridden patients, patients with limited mobility, patients with spinal cord injuries, diabetic patients, and malnourished individuals are all high-risk groups. Epidemiological studies show that the prevalence of pressure injury in hospitalized adults can reach 8%–13%, significantly increasing patients' pain and infection risk, prolonging hospital stays, reducing quality of life, and imposing a heavy medical and nursing burden.

[0003] Existing measures for the prevention and management of pressure injuries mainly include regular positional changes, the use of pressure-relieving mats or mattresses, and enhanced skin care. However, these measures have the following shortcomings:

[0004] (1) Relying on manual inspection and subjective judgment, it is impossible to continuously obtain physiological parameters of wounds or high-risk areas, and it is easy to miss the opportunity for early injury intervention.

[0005] (2) Conventional dressings cover the wound during use, forming a "black box" state. Medical staff cannot understand the changes in the wound microenvironment in real time without removing the dressing. Frequent removal of the dressing will damage the wound microenvironment and delay healing.

[0006] (3) Some existing smart dressing products can only monitor a single parameter, such as temperature or pressure, lack multimodal information fusion, cannot fully reflect the pathological process of the wound, and cannot provide personalized intervention suggestions for nursing staff.

[0007] (4) The lack of a closed-loop system linked with the cloud data management platform results in low utilization of historical data, making it impossible to conduct trend analysis, risk prediction, and cross-scenario nursing collaboration.

[0008] In summary, there is an urgent clinical need for an intelligent system that can monitor wound status in real time with multiple parameters, conduct risk assessment based on pathological mechanisms, and proactively output personalized nursing intervention suggestions. This system can dynamically monitor the wound microenvironment while the wound is covered by dressings, and avoid the adverse effects of frequent dressing changes. This would improve the early identification and intervention capabilities for pressure injuries, improve patient prognosis, and reduce the nursing burden. Summary of the Invention

[0009] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. This summary section is not intended to limit the key and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0010] This invention proposes an intelligent sensing dressing system for multimodal wound condition monitoring and active intervention, comprising:

[0011] The flexible multimodal sensing layer includes a pressure sensing unit, a temperature sensing unit, a pH sensing unit, and a blood flow monitoring unit, which are used to collect multimodal parameters in real time. These multimodal parameters include pressure, temperature, pH value, and tissue blood flow parameters in the wound area.

[0012] The microenvironment-regulating dressing layer includes a pressure-reducing buffer structure and a humidity-regulating structure, which are used to disperse concentrated pressure in the wound area and maintain a moist healing environment;

[0013] An integrated flexible electronic control unit is electrically connected to the aforementioned flexible multimodal sensing layer for signal acquisition, processing, and wireless transmission.

[0014] The local intelligent terminal module communicates wirelessly with the aforementioned flexible multimodal sensing layer to receive the aforementioned multimodal parameters and execute an artificial intelligence risk prediction algorithm to output the wound damage risk level and nursing recommendations.

[0015] The cloud-based data platform interacts with the aforementioned local smart terminal module to store the multimodal parameters, generate trend analysis reports, and push risk alarm information to nursing staff or medical institutions. Specifically, when the aforementioned local smart terminal module detects that the risk level of wound damage has reached a preset threshold, it sends intervention prompt information to the nursing staff or patient terminal.

[0016] In one feasible implementation, the pressure sensing unit is a flexible thin-film structure, distributed in an array on the wound contact surface of the dressing.

[0017] In one feasible implementation, the temperature sensing unit and the pH sensing unit are embedded in a flexible substrate and connected to an integrated flexible electronic control unit via the same signal bus.

[0018] In one feasible implementation, the blood flow monitoring unit is an optical detection module or an impedance detection module, used to assess the local blood perfusion status of the wound.

[0019] In one feasible implementation, the aforementioned pressure-reducing buffer structure uses polyurethane foam material to form a support layer to disperse concentrated pressure at bony prominences.

[0020] In one feasible implementation, the aforementioned wireless transmission module employs a low-power wireless communication protocol and dynamically adjusts the sampling and transmission frequencies based on risk prediction results.

[0021] In one feasible implementation, the pH sensing unit adopts an ion-selective electrode structure and is combined with a display film layer to achieve color visualization to assist in judgment.

[0022] In one feasible implementation, the aforementioned artificial intelligence risk prediction algorithm is used to calculate the risk level of wound damage based on historical multimodal monitoring data and generate personalized care recommendations, including the frequency of turning over, dressing change time, and local decompression plan.

[0023] In one feasible implementation, the specific operational steps of the aforementioned artificial intelligence risk prediction algorithm include:

[0024] The above multimodal parameters are denoised, time-aligned, and individualized baseline normalized based on a sliding time window.

[0025] Calculate multidimensional characteristic indicators of pressure-time integral, overthreshold time, temperature change, pH drift, perfusion index and their interaction terms within the above time window.

[0026] Short-term pattern features are extracted from the above multidimensional feature indicators by a convolutional temporal network, and long-term evolution features are extracted from the above multidimensional feature indicators by a bidirectional long short-term memory network. The above multidimensional feature indicators are then input into a multilayer perceptron for processing.

[0027] A multi-head attention mechanism is used to fuse the aforementioned short-term pattern features, long-term evolutionary features, and multi-dimensional feature indicators to generate a fused feature representation;

[0028] Based on the above-mentioned fusion feature representation, the wound injury risk score is calculated and mapped to the above-mentioned wound injury risk level;

[0029] The effects of different turning frequencies, dressing change times, and local decompression strategies were simulated using counterfactual simulation methods, and the above nursing recommendations were output.

[0030] In one feasible implementation, the cloud data platform includes a data storage module, a trend analysis module, and a remote consultation interface module. The trend analysis module is used to generate a trend analysis report that includes multi-parameter change curves of the wound.

[0031] In summary, the intelligent sensing dressing system for multimodal wound condition monitoring and proactive intervention provided by this invention has significant beneficial effects on addressing the problems of single monitoring parameters, delayed information acquisition, passive nursing intervention, and lack of data closure in existing technologies. This invention integrates pressure, temperature, pH, and blood flow monitoring units into a flexible multimodal sensing layer, enabling simultaneous acquisition of multiple physiological and microenvironmental parameters highly correlated with the mechanisms of pressure injury. Compared to traditional intelligent dressings that only monitor a single indicator, this system comprehensively reflects the multidimensional changes in wound pressure, local ischemia, inflammation, infection, and healing processes, providing a more reliable data foundation for risk assessment. The flexible integrated design allows the system to conform to the skin's curvature, connecting various sensing units and an integrated flexible electronic control unit via flexible wiring, reducing thickness and local hard spots, improving wearing comfort, and allowing for stable signal acquisition even during prolonged application without frequent dressing removal. The system introduces an AI-based risk prediction method based on pathological mechanisms. By calculating mechanistic features such as pressure-time integral, overthreshold time, temperature change rate, pH drift, perfusion index, and their interaction terms, and combining this with multi-branch temporal networks and attention mechanisms, it achieves accurate prediction of wound damage risk. This not only reflects instantaneous anomalies but also captures long-term cumulative effects and multi-parameter interactions, thus issuing early risk warnings and significantly improving the timeliness of early intervention. Based on risk prediction, the system uses counterfactual simulation to model the impact of different nursing protocols on risk, automatically recommending the lowest overall risk turning interval, dressing change timing, and local decompression plan. This allows nurses to directly refer to personalized suggestions, reducing reliance on experience-based judgment and improving the targetedness and effectiveness of interventions. The decompression buffer structure in the microenvironment-regulating dressing layer disperses concentrated pressure at bony prominences, reducing the risk of local tissue ischemia. The humidity-regulating structure, through the combination of a hydrophilic absorbent layer and a breathable backing, maintains a moist but not overly immersed healing environment for the wound, accelerating tissue repair. The cloud-based data platform can store historical monitoring data long-term and generate trend analysis reports, supporting retrieval during ward rounds, remote consultations, and inter-departmental transfers. Simultaneously, it pushes risk alarms and nursing suggestions to hospital information systems or nursing terminals in elderly care facilities in real time, enabling continuous nursing and data sharing. By achieving real-time monitoring and early warning while the wound is covered, this invention reduces the need for frequent dressing removal for wound observation, lowering the risk of infection and alleviating patient pain. Furthermore, intelligent analysis and suggestions reduce the workload of manual inspections and recording, improving nursing efficiency. This allows for accurate identification and timely intervention during high-risk periods without disrupting the wound microenvironment, improving patient prognosis and reducing the burden on caregivers.

[0032] Other advantages, objectives and features of this application will be apparent in part from the description which follows, and in part from what those skilled in the art will understand through study and practice of this application. Attached Figure Description

[0033] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit this specification. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0034] Figure 1 A structural schematic diagram of an intelligent sensing dressing system for multimodal wound condition monitoring and active intervention provided in an embodiment of this application;

[0035] Figure 2 This is a flowchart illustrating an artificial intelligence risk prediction algorithm provided in an embodiment of this application. Detailed Implementation

[0036] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or 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. The technical solutions of the embodiments of this application will now be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0037] Please refer to the figure. Figure 1 This application provides a structural schematic diagram of an intelligent sensing dressing system 10 for multimodal wound condition monitoring and active intervention, which specifically includes:

[0038] The flexible multimodal sensing layer 11 includes a pressure sensing unit, a temperature sensing unit, a pH sensing unit, and a blood flow monitoring unit, which are used to collect multimodal parameters in real time. The multimodal parameters include pressure, temperature, pH value, and tissue blood flow parameters in the wound area.

[0039] The microenvironment regulation dressing layer 12 includes a pressure-reducing buffer structure and a humidity regulation structure, which are used to disperse concentrated pressure in the wound area and maintain a moist healing environment;

[0040] An integrated flexible electronic control unit 13 is electrically connected to the aforementioned flexible multimodal sensing layer and is used for signal acquisition, processing, and wireless transmission.

[0041] The local intelligent terminal module 14 wirelessly communicates with the aforementioned flexible multimodal sensing layer to receive the aforementioned multimodal parameters and execute an artificial intelligence risk prediction algorithm to output the wound damage risk level and nursing recommendations.

[0042] The cloud data platform 15 interacts with the aforementioned local smart terminal module to store the aforementioned multimodal parameters, generate trend analysis reports, and push risk alarm information to nursing staff or medical institutions. Specifically, when the aforementioned local smart terminal module detects that the risk level of wound damage has reached a preset threshold, it sends intervention prompt information to the nursing staff or patient terminal.

[0043] For example, the smart sensing dressing is constructed with a layered structure from the inside out. The layer closest to the wound is a flexible multimodal sensing layer 11, which is covered by a microenvironment regulation dressing layer 12. The data acquisition and transmission components and the power supply components are integrated into an integrated flexible electronic control unit 13. It is preferably placed at the edge of the dressing or in the stepped area of ​​the backing to reduce pressure on the wound and improve wearing comfort. The local smart terminal module 14 serves as the user-side human-computer interaction and edge inference entry point, interacting with the system via low-power wireless means. The cloud data platform 15 provides long-term storage, trend analysis, and cross-departmental sharing capabilities.

[0044] The flexible multimodal sensing layer 11 employs a bendable flexible substrate and arrays pressure sensing units, temperature sensing units, pH sensing units, and blood flow monitoring units. Each unit is electrically connected to the integrated flexible electronic control unit 13 via flexible wiring. The pressure units are preferably a thin-film array to form a surface-distributed pressure thermogram. The temperature and pH units are embedded within the same flexible layer to reduce interlayer thickness and interfacial impedance. The blood flow monitoring units are located in the window area of ​​the pressure array to minimize interference from surface pressure on the optical or impedance paths.

[0045] The microenvironment-regulating dressing layer 12 includes a pressure-reducing buffer structure for dispersing concentrated loads at bony prominences, and a humidity-regulating structure for maintaining a moist healing environment. The pressure-reducing buffer structure provides compliance in the vertical direction and shear support in the tangential direction through porous elastomers or gradient support regions. The humidity-regulating structure, through the synergistic effect of hydrophilic absorption and a breathable backing, absorbs excess exudate when there is significant exudation and maintains local humidity balance in a dry environment.

[0046] The integrated flexible electronic control unit 13 includes signal conditioning and sampling circuitry, a microprocessor unit, and wireless communication circuitry. Its function is to filter, remove spurious signals, and synchronously sample the multimodal raw signals from the flexible multimodal sensing layer 11, encapsulating them into data frames in the form of timestamps. To reduce the additional load on the wound surface, unit 13 is connected to the sensing layer via flexible cabling or conductive fabric. Its encapsulation shell is made of biocompatible material and features a liquid-proof sealing structure.

[0047] The local intelligent terminal module 14 receives multimodal parameters through the wireless communication circuit built into unit 13. The terminal runs a risk prediction algorithm that aligns with the pathological mechanism of pressure ulcers. First, on the terminal side, noise reduction, time alignment, and individualized baseline normalization are performed on the pressure, temperature, pH, and tissue blood flow sequences according to a preset sliding window. Then, mechanism features related to injury occurrence are calculated from within the window, including the pressure-time integral for characterizing continuous pressure dose, the overthreshold time for characterizing high-pressure exposure duration, temperature changes and pH drift for reflecting signs of inflammation or ischemia, and the perfusion index for assessing local perfusion status. Several features can also be interactively combined to capture combined risks.

[0048] Based on this, the terminal employs a multi-branch temporal network to extract short-term local patterns and long-term evolutionary trends, and then merges these mechanistic features into a unified representation in an attention fusion unit. The system outputs a wound injury risk score and maps it to a risk level. To avoid frequent disruptions to the nursing workflow, the terminal applies smoothing and consistency constraints to the risk output over continuous time. Intervention prompts are triggered only when the risk level reaches a preset threshold, prompting nursing staff to adjust the patient's position, reduce local pressure, or change dressings. Simultaneously, the suggested turning frequency, dressing change timing, and local pressure reduction plan are displayed.

[0049] The cloud-based data platform 15 receives and stores multimodal parameters and risk assessment results from the local smart terminal module 14. The platform establishes longitudinal trend curves and periodic overviews based on individual patient histories and supports the export of standardized trend analysis reports. When the same patient is transferred between different nursing units, the platform pushes risk alarm information and treatment suggestions to the corresponding medical information system, achieving continuous care across different scenarios.

[0050] In use, after cleaning the wound, the nursing staff applies the dressing to the target area, ensuring the flexible multimodal sensing layer 11 fully adheres to the wound. Upon power-up, the integrated flexible electronic control unit 13 begins periodically collecting multimodal parameters and establishes a session with the local smart terminal module 14 via wireless link. During patient rest, turning, or receiving nursing care, the system continuously updates pressure thermograms, temperature and pH trends, and perfusion levels. If persistent high pressure is accompanied by a decrease in perfusion or an abnormal increase in temperature, the terminal module 14 immediately generates a high-priority alert and, based on the optimal compromise suggestion obtained from counterfactual simulation, provides turning intervals, positioning guidance, and key points for local decompression.

[0051] After the intervention is completed, the system records the changes in key indicators before and after the intervention and synchronizes them to the cloud data platform to form closed-loop evidence.

[0052] The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention provided by this invention offers significant advantages over existing technologies, which suffer from problems such as single monitoring parameters, delayed information acquisition, passive nursing intervention, and lack of data closure. This invention integrates pressure, temperature, pH, and blood flow monitoring units within a flexible multimodal sensing layer. It can simultaneously collect multiple physiological and microenvironmental parameters highly correlated with the mechanisms of pressure injury. Compared to traditional intelligent dressings that only monitor a single indicator, this system comprehensively reflects multidimensional changes in wound pressure, local ischemia, inflammation, infection, and healing processes, providing a more reliable data foundation for risk assessment. The flexible, integrated design allows the system to conform to the skin's curvature. Flexible wiring connects various sensing units to an integrated flexible electronic control unit, reducing thickness and local hard spots, improving wearing comfort, and enabling stable signal acquisition even during prolonged application without frequent dressing removal. The system introduces an AI-based risk prediction method based on pathological mechanisms. By calculating mechanistic features such as pressure-time integral, overthreshold time, temperature change rate, pH drift, perfusion index, and their interaction terms, and combining this with multi-branch temporal networks and attention mechanisms, it achieves accurate prediction of wound damage risk. This not only reflects instantaneous anomalies but also captures long-term cumulative effects and multi-parameter interactions, thus issuing early risk warnings and significantly improving the timeliness of early intervention. Based on risk prediction, the system uses counterfactual simulation to model the impact of different nursing protocols on risk, automatically recommending the lowest overall risk turning interval, dressing change timing, and local decompression plan. This allows nurses to directly refer to personalized suggestions, reducing reliance on experience-based judgment and improving the targetedness and effectiveness of interventions. The decompression buffer structure in the microenvironment-regulating dressing layer disperses concentrated pressure at bony prominences, reducing the risk of local tissue ischemia. The humidity-regulating structure, through the combination of a hydrophilic absorbent layer and a breathable backing, maintains a moist but not overly immersed healing environment for the wound, accelerating tissue repair. The cloud-based data platform can store historical monitoring data long-term and generate trend analysis reports, supporting retrieval during ward rounds, remote consultations, and inter-departmental transfers. Simultaneously, it pushes risk alarms and nursing suggestions to hospital information systems or nursing terminals in elderly care facilities in real time, enabling continuous nursing and data sharing. By achieving real-time monitoring and early warning while the wound is covered, this invention reduces the need for frequent dressing removal for wound observation, lowering the risk of infection and alleviating patient pain. Furthermore, intelligent analysis and suggestions reduce the workload of manual inspections and recording, improving nursing efficiency. This allows for accurate identification and timely intervention during high-risk periods without disrupting the wound microenvironment, improving patient prognosis and reducing the burden on caregivers.

[0053] In one feasible implementation, the pressure sensing unit is a flexible thin-film structure, distributed in an array on the wound contact surface of the dressing.

[0054] In one feasible implementation, the temperature sensing unit and the pH sensing unit are embedded in a flexible substrate and connected to an integrated flexible electronic control unit via the same signal bus.

[0055] In one feasible implementation, the blood flow monitoring unit is an optical detection module or an impedance detection module, used to assess the local blood perfusion status of the wound.

[0056] In one feasible implementation, the aforementioned pressure-reducing buffer structure uses polyurethane foam material to form a support layer to disperse concentrated pressure at bony prominences.

[0057] In one feasible implementation, the pressure sensing unit employs a flexible thin-film structure and is arrayed in the area where the dressing directly contacts the wound. A planar measurement interface, flexible to the contours of the skin, is formed by the flexible substrate and conductive mesh. This arrayed layout decomposes localized stress into the normal pressure response of multiple pixels, generating a pressure thermogram for identifying bony prominences and areas of concentrated shear at the edges. To improve stability under long-term application, a low-modulus biocompatible coating can be laminated onto the array layer to reduce microslippage. Zero-point calibration and temperature drift compensation strategies are combined at the acquisition side to suppress baseline drift. The array signal is multiplexed and pre-conditioned before being input to the control unit for synchronous or quasi-synchronous sampling, reducing misjudgments caused by timing mismatches.

[0058] In another feasible implementation, the temperature sensing unit and the pH sensing unit are embedded in the same flexible substrate layer and connected to an integrated flexible electronic control unit via the same signal bus. This integrated layout reduces interlayer thickness and the number of wirings, resulting in a thinner, more conforming dressing and reduced additional load on the wound. The temperature unit captures temperature changes caused by local inflammation or ischemia-reperfusion. The pH unit reflects the microenvironment's pH trend by forming a stable contact interface with wound exudate. The shared bus for both units uses multi-device addressing and time-division polling to achieve data integration. Shielding and isolation wiring are incorporated within the substrate to reduce cross-channel crosstalk. To improve the repeatability of the pH signal, a hydrophilic buffer layer can be formed on the surface of the sensing area to balance transient exudate fluctuations and shorten the response settling time.

[0059] In another feasible implementation, the blood flow monitoring unit can employ either an optical detection module or an impedance detection module to assess the perfusion status of the wound and its adjacent tissues. The optical module extracts pulse-related components by detecting volumetric pulsation absorption changes between the emitting and receiving elements, and is supplemented with ambient light suppression and mechanical shading structures to improve the signal-to-noise ratio. The impedance module indicates microcirculation flux by injecting a safe, small-amplitude alternating current and measuring periodic changes in tissue impedance. A four-electrode arrangement is preferred to reduce contact electrode polarization errors. Both approaches can be time-aligned with pressure array data, using pressure pixels as priors to weighted suppress motion artifacts, thereby distinguishing between true perfusion decline and transient pressure-induced blockage.

[0060] In another feasible implementation, the pressure-reducing buffer structure uses medical-grade polyurethane foam to form the support layer. Through a gradient design of porosity, thickness, and hardness, it provides compliance in the vertical direction to disperse concentrated pressure at bony prominences. Simultaneously, it provides sufficient shear support in the tangential direction to reduce relative displacement at the skin-dressing interface. The foam structure and sensing layer are geometrically windowed or thinned to avoid obstructing the blood flow optical path and to ensure sufficient contact between the pH sensing area and exudate. A breathable backing can be combined with the outer surface to maintain the moisture exchange required for moist healing and reduce the risk of maceration. All the above units are connected to an integrated flexible electronic control unit via flexible wiring. This control unit performs signal conditioning, sampling, and wireless transmission, enabling pressure, temperature, pH, and perfusion information to be continuously and minimally disturbedly converged to the terminal for assessment and intervention decisions without removing the dressing.

[0061] In one feasible implementation, the aforementioned wireless transmission module employs a low-power wireless communication protocol and dynamically adjusts the sampling and transmission frequencies based on risk prediction results.

[0062] For example, the wireless transmission module employs a low-power wireless communication protocol to reduce energy consumption during long-term continuous monitoring. This module can dynamically adjust sampling and transmission frequencies based on AI risk prediction results. When the prediction results indicate a low risk level and stable monitoring parameters, the module operates at a low sampling rate and low transmission frequency, only periodically sending summary data. When the risk level increases or parameters experience abnormal fluctuations, the module automatically switches to a high sampling rate and high transmission frequency mode, uploading key multimodal data in real time to support timely intervention and risk management. This approach ensures timely risk alerts while significantly extending the device's battery life and improving the effectiveness and stability of data transmission.

[0063] In one feasible implementation, the pH sensing unit adopts an ion-selective electrode structure and is combined with a display film layer to achieve color visualization to assist in judgment.

[0064] For example, the pH sensing unit employs an ion-selective electrode structure, integrating a specific ion-sensitive membrane onto a flexible substrate to detect the electrical signal of the wound environment's pH level. Simultaneously, this sensing unit is combined with a display membrane layer; when the pH value shifts, the display membrane layer produces a corresponding color change, thus providing intuitive visual assistance in judgment based on electronic data acquisition. This design allows medical personnel to quickly understand the wound's pH trend by observing color changes, even without connecting to a terminal device, achieving dual protection of electronic monitoring and intuitive judgment, improving ease of use and flexibility in clinical application.

[0065] In one feasible implementation, the aforementioned artificial intelligence risk prediction algorithm is used to calculate the risk level of wound damage based on historical multimodal monitoring data and generate personalized care recommendations, including the frequency of turning over, dressing change time, and local decompression plan.

[0066] In one feasible implementation, such as Figure 2 As shown, the specific operation steps of the above-mentioned artificial intelligence risk prediction algorithm include:

[0067] S210. Based on the sliding time window, the above multimodal parameters are denoised, time-aligned, and individualized baseline normalized.

[0068] S220. Calculate the multidimensional characteristic indicators of pressure-time integral, overthreshold time, temperature change, pH drift, perfusion index and their interaction terms from the above time window.

[0069] S230. Short-term pattern features are extracted from the above multidimensional feature indicators through a convolutional temporal network, long-term evolution features are extracted from the above multidimensional feature indicators through a bidirectional long short-term memory network, and the above multidimensional feature indicators are input into a multilayer perceptron for processing.

[0070] S240. Employ a multi-head attention mechanism to fuse the aforementioned short-term pattern features, long-term evolutionary features, and multi-dimensional feature indicators to generate a fused feature representation;

[0071] S250. Calculate the wound injury risk score based on the above-mentioned fusion feature representation and map it to the above-mentioned wound injury risk level;

[0072] S260. Use counterfactual simulation to simulate the effects of different turning frequencies, dressing change times, and local decompression schemes, and output the above nursing recommendations.

[0073] For example, the artificial intelligence risk prediction algorithm of the present invention uses multimodal monitoring data collected collaboratively by the dressing end and the terminal as input, including pressure, temperature, pH value and tissue blood flow parameters of the wound area. The algorithm performs the following steps within a sliding time window.

[0074] First, in the data preprocessing stage, let the pressure, temperature, pH, and perfusion sequence be respectively...

[0075] P(t), T(t), pH(t), B(t), within a time window of length L W k ={t k -L+1,…,t k Within this scope, individualized baseline means for each modality are calculated based on individual static information (such as weight, location of the bulge, comorbidities, etc.). with standard deviation And perform normalization:

[0076]

[0077] This step can eliminate the effects of individual differences and sensor drift.

[0078] In the mechanism feature construction stage, various feature indicators related to the mechanism of pressure ulceration are calculated within a time frame, including:

[0079] Pressure-Time Integral (PTI):

[0080]

[0081] Where, τ P The safe pressure threshold is Δt, and the sampling interval is Δt.

[0082] Time to overthreshold (TAT):

[0083]

[0084] Where τ is the statistical threshold and 1 / {·} is the indicator function.

[0085] Temperature change and rate of change:

[0086]

[0087] pH change rate and fluctuation:

[0088]

[0089] Perfusion Index (PI):

[0090]

[0091] Here, AC and DC represent the pulsating component and the direct current component, respectively. The above features and their interaction terms combine to form the mechanism feature vector:

[0092] Φ k =[PTI k ,TAT k ,ΔTk ,s T ,s pH ,PI k PTI k ·s T PTI k ·(1-PI k )]

[0093] In the temporal feature extraction stage, short-term pattern features Extracted by a convolutional temporal network (TCN) to identify rapid fluctuation patterns within a window; long-term evolutionary features. Extracted by a bidirectional long short-term memory network (Bi-LSTM) to capture trend and rhythm changes across windows; the mechanism feature vector is obtained by embedding through a multilayer perceptron (MLP). Characterization used to supplement the display of pathological features.

[0094] Subsequently, the algorithm utilizes a multi-head attention mechanism to fuse three types of features and calculates weight coefficients. And generate fusion representation :

[0095]

[0096] in, This is a personalized query vector.

[0097] The risk score is calculated by integrating the input representations into the logistic regression unit. :

[0098]

[0099] It is then mapped to three risk levels—low, medium, and high—based on a preset threshold θ.

[0100] During the nursing suggestion generation phase, the algorithm uses counterfactual simulation to simulate different turning frequencies Δt, dressing change times, and local decompression schemes, and calculates the objective function:

[0101]

[0102] Where α represents the decompression intensity, c work With c comfort These represent the costs of nursing workload and patient comfort, respectively. The intervention plan with the optimal overall effect is obtained by minimizing J(Δt,α), and this is output as a personalized nursing recommendation.

[0103] In one feasible implementation, the cloud data platform includes a data storage module, a trend analysis module, and a remote consultation interface module. The trend analysis module is used to generate a trend analysis report that includes multi-parameter change curves of the wound.

[0104] For example, the cloud-based data platform includes a data storage module, a trend analysis module, and a remote consultation interface module. The data storage module categorizes and saves multimodal wound monitoring data by patient, location, and time, recording the data collection source and equipment information, and supporting access control and audit trails. The trend analysis module normalizes and denoises time-series data such as pressure, temperature, pH, and blood flow, generating multi-parameter change curves. Combined with risk scores and intervention event markers, it produces a trend analysis report, providing a basis for adjusting nursing plans. When high-risk characteristics are detected, the module highlights the corresponding time period in the report and provides a prompt. The remote consultation interface module provides authorized experts with online access and annotation functions, supporting the synchronization of opinions to local nursing terminals or hospital information systems, achieving multi-party collaboration and closed-loop decision-making. This platform integrates the storage, analysis, and clinical application of wound monitoring data, improving the timeliness of risk identification and intervention.

[0105] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 this application.

Claims

1. An intelligent sensing dressing system for multimodal wound condition monitoring and active intervention, characterized in that, include: The flexible multimodal sensing layer includes a pressure sensing unit, a temperature sensing unit, a pH sensing unit, and a blood flow monitoring unit, which are used to collect multimodal parameters in real time, including pressure, temperature, pH value, and tissue blood flow parameters in the wound area. The microenvironment-regulating dressing layer includes a pressure-reducing buffer structure and a humidity-regulating structure, which are used to disperse concentrated pressure in the wound area and maintain a moist healing environment; An integrated flexible electronic control unit is electrically connected to the flexible multimodal sensing layer for signal acquisition, processing, and wireless transmission. The local intelligent terminal module wirelessly communicates with the flexible multimodal sensing layer to receive the multimodal parameters and execute an artificial intelligence risk prediction algorithm to output the wound damage risk level and care recommendations. The cloud-based data platform interacts with the local smart terminal module to store the multimodal parameters, generate trend analysis reports, and push risk alarm information to nursing staff or medical institutions. When the local smart terminal module detects that the risk level of wound damage has reached a preset threshold, it sends intervention prompt information to the nursing staff or patient terminal.

2. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 1, characterized in that, The pressure sensing unit is a flexible thin-film structure, distributed in an array on the wound contact surface of the dressing.

3. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 1, characterized in that, The temperature sensing unit and pH sensing unit are embedded in a flexible substrate and connected to an integrated flexible electronic control unit via the same signal bus.

4. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 1, characterized in that, The blood flow monitoring unit is an optical detection module or an impedance detection module, used to assess the local blood perfusion status of the wound.

5. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 1, characterized in that, The pressure-reducing and buffering structure uses polyurethane foam material to form a support layer to disperse concentrated pressure at bony prominences.

6. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 1, characterized in that, The wireless transmission module adopts a low-power wireless communication protocol and dynamically adjusts the sampling and transmission frequency based on the risk prediction results.

7. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 1, characterized in that, The pH sensing unit adopts an ion-selective electrode structure and is combined with a display film layer to achieve color visualization to assist in judgment.

8. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 1, characterized in that, The artificial intelligence risk prediction algorithm is used to calculate the risk level of wound damage based on historical multimodal monitoring data and generate personalized care recommendations, including the frequency of turning over, dressing change time, and local decompression plan.

9. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 8, characterized in that, The specific operation steps of the artificial intelligence risk prediction algorithm include: The multimodal parameters are denoised, time-aligned, and individualized baseline normalized based on a sliding time window. Calculate multidimensional characteristic indicators from the time window, including pressure-time integral, overthreshold time, temperature change, pH drift, perfusion index and their interaction terms. Short-term pattern features are extracted from the multidimensional feature indicators by a convolutional temporal network, and long-term evolution features are extracted from the multidimensional feature indicators by a bidirectional long short-term memory network. The multidimensional feature indicators are then input into a multilayer perceptron for processing. A multi-head attention mechanism is used to fuse the short-term pattern features, the long-term evolution features, and the multi-dimensional feature indicators to generate a fused feature representation; The wound injury risk score is calculated based on the fusion feature representation and mapped to the wound injury risk level; The effects of different turning frequencies, dressing change times, and local decompression schemes were simulated using counterfactual simulation methods, and the nursing recommendations were output.

10. The intelligent sensing dressing system for multimodal wound condition monitoring and active intervention according to claim 1, characterized in that, The cloud-based data platform includes a data storage module, a trend analysis module, and a remote consultation interface module. The trend analysis module is used to generate a trend analysis report that includes multi-parameter change curves of the wound.

Images