Digital bed perception management system and method based on RFID and internet of things

Abstract

The application discloses a digital sickbed perception management system and method based on RFID and the Internet of Things, and belongs to the technical field of intelligent medical Internet of Things. The system comprises a perception layer, an edge processing layer, a cloud service center and a multi-level application terminal, wherein the perception layer is composed of a fixed RFID reader and writer arranged on a sickbed and a passive RFID wristband label bound with a patient, and is used for continuously collecting patient bed occupation original data. The technical problems of the prior art, such as the inability to automatically perceive the patient-bed binding relationship in real time, the lag of bed operation data, the lack of objective evidence chain in medical insurance supervision, the multi-department data island and the rough semantic analysis of the off-bed state, are solved.

Description

Technical Field

[0001] This invention belongs to the field of smart medical Internet of Things technology, specifically relating to a real-time sensing system and method for inpatient bed status, multi-source data fusion, and cross-level business collaboration management based on radio frequency identification (RFID) and edge-cloud collaborative computing architecture. Background Technology

[0002] Managing the inpatient status is a core aspect of hospital operations and healthcare safety. Current mainstream management models include:

[0003] ① Manual patrol mode: This mode relies on nurses making regular rounds of the wards and manually recording whether patients are in bed. This mode has high labor costs, cannot achieve continuous monitoring, and has significant blind spots in supervision at night and on holidays.

[0004] ② Discrete Information System Mode: Bed status is entered and updated based on the nurse workstation computer (HIS system). Data updates rely on manual triggering, resulting in lag and failing to reflect the patient's real-time physical location status.

[0005] ③ Isolated RFID application mode: Some hospitals use RFID wristbands for patient identification, but the application scenarios are limited to access control, medication verification or surgical safety checks. They do not establish a continuous physical binding relationship with the bed itself, and cannot form an automated and periodic perception closed loop for the patient-bed occupancy status.

[0006] The aforementioned prior art has at least the following drawbacks:

[0007] (1) The patient-bed binding relationship lacks real-time physical perception capability. Existing hospital beds are only physical facilities and do not have the function of automatically identifying the patient's identity and status. Medical staff need to manually check the wristband and bed card to confirm the identity, which poses the risk of "misplacement" and "mixed-placement", and cannot automatically identify patient leaving the bed.

[0008] (2) Bed operation data is lagging and coarse-grained. Key operational indicators such as bed occupancy rate, turnover rate, and average length of stay rely on post-event statistics and manual summarization. The time resolution is in the "day" unit, and it cannot provide dynamic data at the hour or even minute level, resulting in a lack of real-time data support for bed scheduling decisions.

[0009] (3) There are data gaps and room for fraud in medical insurance supervision. The medical insurance department mainly relies on random checks and medical record tracing to verify the true hospital status of patients. It lacks an objective, continuous and tamper-proof chain of hospitalization evidence based on Internet of Things perception. Violations such as "bed-hanging hospitalization" and "fake hospitalization" are difficult to be identified and warned in real time.

[0010] (4) Data silos and heterogeneous system barriers among multiple departments: Clinical (HIS / EMR), hospital management (BI system), medical insurance supervision, and regional health platforms have the same source but are not synchronized. Patient hospitalization status data needs to be repeatedly entered into multiple systems, forming information silos and making it difficult to support collaborative management and decision analysis.

[0011] (5) Lack of multi-dimensional state semantic analysis capability: Even if the existing technology collects the patient's bed leaving signal, it cannot distinguish between "brief bed leaving (toilet / examination)," "regular activity (rehabilitation training)," and "abnormal bed leaving (fall risk / unauthorized departure from the hospital)," resulting in a high false alarm rate and poor clinical usability. Summary of the Invention

[0012] The purpose of this invention is to provide a digital hospital bed sensing management system and method based on RFID and the Internet of Things to solve the technical problems of existing technologies, such as the inability to automatically sense patients and beds in real time, lagging bed operation data, lack of objective evidence for medical insurance supervision, data silos among multiple departments, coarse semantic parsing of bed leave status, and high false alarm rate. This invention aims to achieve automated sensing of digital hospital beds, accurate judgment through edge cloud collaboration, multi-level business collaboration, intelligent medical insurance supervision, and coordinated scheduling of regional medical resources.

[0013] To achieve the above objectives, the present invention provides the following technical solution:

[0014] A digital hospital bed sensing and management system based on RFID and IoT adopts a four-layer architecture: sensing layer - edge processing layer - cloud service center - multi-level application terminal;

[0015] The sensing layer includes a digital bed sensor unit deployed on each hospital bed. The digital bed sensor unit includes a fixed UHF RFID reader and a passive RFID wristband tag. The fixed RFID reader continuously reads the passive RFID wristband tag in the area of ​​the bed at a preset scanning cycle, and collects the tag ID, RSSI value, phase value and timestamp raw sensing data. This breaks through the limitation of existing RFID wristbands being used only for discrete identity verification. Through the collaboration of the fixed reader and the passive wristband, a continuous sensing network of the patient's physical presence status is constructed with the hospital bed as the anchor point, realizing 24 / 7 automated monitoring and fundamentally eliminating errors from manual verification.

[0016] The edge processing layer deploys an edge computing gateway to aggregate raw data from multiple digital bed sensor units in the ward, perform signal filtering and noise reduction, perform initial edge judgment of bed leave status, and upload the data after compression and encryption.

[0017] The cloud service center includes a cloud server, a time-series database, a relational database, a status determination engine, a multi-dimensional semantic parsing engine, and a standard interface service module, which are used for data aggregation and storage, accurate determination of patient bed status, semantic annotation of bed exit events, and standardized data connection with external systems.

[0018] The architecture adopts a layered computing structure of "preliminary edge judgment + precise cloud judgment". The edge side realizes low-latency state pre-judgment and data cleaning, while the cloud side performs high-precision semantic parsing based on historical models and business rules, balancing real-time performance and judgment accuracy, and reducing network bandwidth consumption and cloud computing load.

[0019] The multi-level application terminals include clinical terminals, hospital management terminals, medical insurance supervision terminals, and regional health terminals, which provide differentiated business application services to medical staff, hospital managers, medical insurance agencies, and health management departments, respectively.

[0020] In the aforementioned management system, the fixed UHF RFID reader conforms to the ISO18000-6C / EPCGen2 standard, operates at a frequency of 920-925MHz, and has an adjustable transmission power of 10-30dBm. The reader is installed above the headboard of the hospital bed at a height of 1.2m from the ground, and uses a circularly polarized antenna. Through power control and antenna beam orientation optimization, the effective reading distance is limited to 0.5-1.5m, providing independent coverage for each bed. In conjunction with RSSI threshold range [-60dBm, -30dBm] filtering, signal crosstalk between adjacent beds is suppressed.

[0021] In the aforementioned management system, the passive RFID wristband tag has a built-in unique EPC code, is encapsulated with medical-grade hypoallergenic flexible material, has a waterproof rating of IP67, and the chip storage information is encrypted using AES-128. The wristband is equipped with a one-time tamper-proof buckle, which destroys the tag's antenna circuit upon removal. Upon admission, the wristband ID is uniquely bound to the patient's electronic medical record number, basic information, and medical insurance participation number.

[0022] In the aforementioned management system, the edge computing gateway adopts an industrial-grade ARM architecture device with a built-in finite state machine (FSM) algorithm. It defines three states: S0 in bed, S1 suspected of being out of bed, and S2 confirmed to be out of bed. Based on the continuous scanning cycle tag detection results and the valid RSSI value, it completes the state pre-judgment and uploads the preliminary judgment result, original feature data, and confidence score to the cloud through a TLS1.3 encrypted channel.

[0023] In the aforementioned management system, the multi-dimensional semantic parsing engine of the cloud service center integrates RSSI spatiotemporal features, patient care level, medical appointment schedule, historical bed leaving behavior patterns, and day / night bed leaving time multi-dimensional factors to semantically classify bed leaving events into normal activities, examinations, rehabilitation training, abnormal bed leaving, and unauthorized departure from the hospital; at the same time, it constructs a patient-bed-time triplet time-series data model, records state switching event logs, and persists evidence.

[0024] In the aforementioned management system, the cloud service center achieves interconnection and interoperability with the hospital's HIS / EMR system, medical insurance platform, and regional health platform through RESTful API and HL7 FHIR protocol, supporting the synchronization of basic patient information, automatic writing back of bed status, and acquisition of treatment schedule data; multi-level application terminals achieve one-time perception and multi-party data sharing based on a unified data source.

[0025] This invention also provides a digital hospital bed sensing and management method based on RFID and the Internet of Things, applied to the above-mentioned system, comprising the following steps:

[0026] S1: Patient identity binding and bed allocation: When a patient is admitted to the hospital, a passive RFID wristband is issued. The wristband ID is uniquely mapped in the cloud with the patient's electronic medical record number, basic information, and medical insurance participation number. The corresponding bed is then allocated and the digital bed sensor unit of that bed is activated.

[0027] S2: Raw sensing data collection: Fixed RFID readers continuously scan the bed area according to a preset cycle. If a valid tag is detected, the tag ID, RSSI, phase value and timestamp are recorded. If no tag is detected for several consecutive cycles, a suspected bed leaving event is generated.

[0028] S3: Preliminary status judgment and data upload at the edge: The edge gateway filters interference from neighboring beds through RSSI dynamic threshold, and uses a finite state machine to complete the preliminary judgment of being in bed, suspected leaving bed, body position change, and confirmed leaving bed. After compression and encryption, the data is uploaded to the cloud.

[0029] S4: Cloud-based state calibration and semantic parsing: Cloud-based spatiotemporal consistency verification, historical model confidence weighting, and business rule engine threshold determination are performed. Multi-dimensional semantic annotation is completed by combining nursing level, treatment schedule, and bed leave time to generate structured state switching event logs.

[0030] A semantic parsing method for bed exit based on multi-source data fusion: It not only relies on RFID presence detection, but also integrates multi-dimensional information such as RSSI spatiotemporal characteristics, patient care level, treatment schedule, historical behavior patterns and bed exit time to construct a semantic classification model for bed exit events, which significantly reduces the false alarm rate and improves clinical usability.

[0031] S5: Business Process Triggering and Terminal Updates: When abnormal bed leaving / unauthorized departure from the hospital is detected, a graded alarm is pushed to the clinical terminal and nursing documents are synchronized; when bed transfer events are detected, HIS bed allocation information is automatically updated; data dashboards and operational statistics indicators at all levels of terminals are refreshed in real time.

[0032] S6: Cross-level data services and decision support: Medical staff, hospital management, medical insurance supervision, and regional health departments can access differentiated data services through authorized terminals to achieve real-time bed monitoring, operational analysis, remote medical insurance auditing, and regional bed emergency dispatch.

[0033] Through a unified IoT sensing data source, differentiated data services are provided to multiple parties, including clinicians, managers, medical insurance, and health authorities, via a standardized API (HL7 FHIR), breaking down data silos and building a collaborative medical data ecosystem of "one-time sensing, multi-party sharing".

[0034] Leveraging the continuity, objectivity, and non-repudiation of IoT sensing, it provides medical insurance departments with a genuine chain of inpatient evidence based on physical sensing rather than manual reporting, supporting the automated identification and early warning of violations such as "bed-occupancy fraud".

[0035] In the above management method, the business rule engine preset thresholds in step S4 include: leaving the bed for more than 30 minutes between 22:00 and 06:00 at night without an appointment is marked as abnormal leaving the bed; leaving the bed for more than 120 minutes during the day without a medical appointment is marked as abnormal leaving the bed; leaving the bed for more than 4 hours without system registration is marked as unauthorized leaving the hospital; leaving the bed for 15-120 minutes and matching the HIS examination appointment record is marked as examination outing.

[0036] In the above management method, system data security adopts layered protection: AES-128 encryption is used from the perception layer to the edge layer, TLS1.3 transmission encryption is used from the edge to the cloud, and OAuth2.0+JWT token authentication is used for cloud APIs; sensitive patient identity information and RFID tag IDs are stored separately under pseudonyms, in conjunction with RBAC role-based access control and field-level database encryption.

[0037] In the above management method, the system is integrated with HIS / EMR based on the HL7 FHIR R4 interface. It synchronizes basic patient information through Patient resources, writes back bed status changes through Encounter resources, and obtains treatment appointment schedules through Appointment resources, supporting intelligent semantic determination of bed exit events.

[0038] Compared with the prior art, the present invention has at least the following beneficial effects:

[0039] 1. Real-time fully automated perception of patient bed status: Through continuous interaction between RFID readers and wristbands, traditional physical hospital beds are upgraded to "digital hospital beds" with perception capabilities. The time resolution of status perception can reach the minute level, and the accuracy and timeliness of data far exceed that of manual mode.

[0040] 2. Improve medical safety and nursing efficiency: Automatic identity verification eliminates the risk of "misplacement"; real-time early warning of abnormal bed leaving (such as postoperative patients leaving the bed at night, and the risk of falls in elderly patients) shortens emergency response time; and reduces the workload of nurses' administrative rounds, allowing them to focus more on professional nursing care.

[0041] 3. Empower hospitals to refine their operations and management: Provide multi-level real-time bed heat maps and dynamic utilization rate indicators for the entire hospital / department / ward, support bed scheduling optimization based on real data, shorten the average length of stay, and improve bed turnover efficiency.

[0042] 4. Construct a technological barrier for intelligent medical insurance supervision: Provide objective, continuous, and traceable IoT sensing data for medical insurance auditing, effectively identify violations such as "bed-occupancy fraud" and "false hospitalization," and safeguard the security of the medical insurance fund.

[0043] 5. Break down data silos across departments: Through a unified perception layer and standardized data interfaces, achieve synchronization of clinical, operational, medical insurance, and health data, supporting the integration of regional medical resources and intelligent decision-making.

[0044] 6. Low modification cost and high scalability: Minimal modification to existing hospital beds and network environment; RFID readers can be deployed quickly; the system adopts a modular architecture, which can be smoothly expanded to new hospital areas and new wards, and is compatible with existing HIS / EMR systems. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a schematic diagram of the overall system architecture (a four-layer structure of perception layer, edge layer, cloud layer, and application terminal, and data flow).

[0047] Figure 2 A schematic diagram of the deployment of sensor units for digital hospital beds (installation location of fixed RFID readers, antenna coverage, and design for suppressing interference from adjacent beds).

[0048] Figure 3The flowchart for edge-cloud collaborative status determination (including signal filtering, FSM initial judgment, cloud-based fine judgment, and semantic annotation logic);

[0049] Figure 4 A functional architecture diagram of a multi-level application terminal;

[0050] Figure 5 This is a schematic diagram of a patient-bed-time correlation data model. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0052] The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0053] Example 1

[0054] This embodiment improves the hospital bed management system, specifically providing a digital hospital bed sensing management system based on RFID and the Internet of Things.

[0055] 1. System Overall Architecture

[0056] The architecture adopts a four-layer structure: "Perception Layer - Edge Processing Layer - Cloud Service Center - Multi-level Application Terminal".

[0057] 1.1 Sensing Layer: Digital Hospital Bed Sensor Unit

[0058] The sensor unit deployed on each standard hospital bed includes:

[0059] Fixed RFID reader: A UHF RFID reader conforming to ISO 18000-6C / EPC Gen2 standard (such as Impinj R420 or equivalent domestic equipment) is selected, operating at a frequency of 920-925MHz, with an adjustable transmission power range of 10-30dBm. Employing ultra-high frequency (UHF) RFID technology, the reader is installed 1.2 meters above the ground above the headboard of the hospital bed, using a circularly polarized antenna with a beamwidth of 60° horizontally and 60° vertically. Power fine-tuning (usually set to 15-20dBm) ensures the effective reading area covers the bed surface and a 0.5-meter radius around it, ensuring stable reading of the wristband when the patient is lying flat, on their side, or sitting. Simultaneously, it avoids reading signals from wristbands in adjacent beds (usually at intervals ≥1.2 meters). Power control and antenna pattern design suppress crosstalk between adjacent beds.

[0060] Passive RFID wristband tag: Utilizing a passive UHF tag conforming to the EPC Gen2 standard, with a chip capacity of 96-128 bits in the EPC area, a built-in globally unique EPC code, and encapsulation in medical-grade silicone or thermoplastic polyurethane (TPU) material, it has an IP67 waterproof rating. The surface can be printed with a QR code containing basic patient information, which is encrypted and stored using AES-128. The wristband has a built-in one-time tamper-evident clip; removing it destroys the tag's antenna circuit, preventing transfer. Upon patient admission, the wristband tag ID is uniquely linked to the patient's electronic medical record number (EMRID) in the system.

[0061] The fixed RFID reader is continuously activated at a preset scanning cycle (e.g., 10 seconds / time) and reads the wristband tags in its coverage area to obtain the tag ID, received signal strength indicator (RSSI), phase value, and timestamp as the raw sensing data stream.

[0062] 1.2 Network and Edge Processing Layer: Edge Computing Gateway

[0063] Employing industrial-grade ARM architecture edge computing devices (such as NVIDIA Jetson Nano or Rockchip RK3588 platform), equipped with at least 4GB of RAM and 32GB of storage, and running an embedded Linux system, the gateway connects up to 32 bed readers via RS485 bus or Ethernet. It incorporates a lightweight Python / Node-RED data processing framework for local data aggregation and initial analysis. Deployed in wards or nursing units, the edge computing gateway gathers raw data from multiple digital bed sensor units within its jurisdiction via wired (RS485 / Ethernet) or wireless (Wi-Fi / ZigBee) methods, performing the following edge-side processing:

[0064] Signal filtering and denoising: Based on RSSI threshold filtering and multipath effect suppression algorithm, environmental reflection interference and leakage signals from adjacent beds are eliminated;

[0065] Preliminary determination of in-bed / out-of-bed status: Based on the tag presence detection results of multiple consecutive scanning cycles, a preliminary status determination (such as in-bed / out-of-bed determination) is performed.

[0066] Data compression and encryption: The preliminary judgment results and original feature data are edge-compressed and uploaded to the cloud service center through a TLS 1.3 encrypted channel.

[0067] 1.3 Cloud Data and Service Center

[0068] The cloud service center includes cloud servers, time-series databases, relational databases, a status determination engine, a multi-dimensional semantic parsing engine, and standard interface service modules. It is used for data aggregation and storage, accurate determination of patient bed status, semantic annotation of bed exit events, and standardized external data integration. Its core functions include:

[0069] Data aggregation and storage: Receive data reported by edge gateways from multiple hospital areas and wards, and construct a structured "patient-bed-time" triplet time-series database;

[0070] Core Status Determination Engine: Based on the signal strength spatiotemporal feature fusion algorithm and business rule engine, it accurately determines the patient's physical status, identifies admission, bed transfer, bed leaving, and bed returning events, and records the status transition timestamp;

[0071] Multi-dimensional semantic analysis: Combining patient care level, treatment schedule, historical behavioral patterns and time of leaving the bed (daytime / nighttime), semantic annotation is performed on bed leaving events (such as "normal examination leave", "rehabilitation activity", "abnormal bed leaving warning").

[0072] Standard interface service: Connects with hospital HIS / EMR, medical insurance platform, and regional health platform through RESTful API and HL7 FHIR protocol to achieve data sharing and business collaboration.

[0073] 1.4 Multi-level application terminals

[0074] Based on the data interface of the cloud service center, build a software application layer for different users:

[0075] Clinical terminal medical staff Real-time bed status view, patient dynamic alerts (abnormal bed leaving, failure to return within time limit), rapid bedside identity verification, and shift handover status snapshot. Hospital Management Terminal Hospital administrators Hospital-wide bed heat map, real-time bed occupancy / vacancy rate, dynamic calculation of average length of stay, departmental turnover efficiency ranking, and resource allocation simulation. Medical insurance monitoring terminal Medical insurance management agencies Real-time inquiry of insured persons' hospitalization status, automatic comparison of hospitalization duration, inpatient rate statistics, warning of suspicious behavior (frequent short-term hospitalization, long-term bed rest at night), and storage of IoT sensing data. Regional health terminal Health and wellness management departments A summary view of hospital bed resources within the jurisdiction, regional medical load early warning, horizontal comparison of medical quality indicators, and emergency dispatch of beds for public health emergencies.

[0076] The state determination algorithm in this embodiment:

[0077] 1. Edge-side initial judgment algorithm (finite state machine FSM): Define three states: S0 (in bed), S1 (suspected to be out of bed), and S2 (out of bed).

[0078] State transition condition: If a valid RSSI (-60dBm≤RSSI≤-30dBm) is detected for 3 consecutive scan cycles (30 seconds), then the state transitions from S1 / S2 to S0.

[0079] If no tag is detected for three consecutive scanning cycles, the process is moved from S0 to S1 (suspected bed departure) and reported to the cloud.

[0080] If no tag is detected for three consecutive cycles in state S1, proceed to state S2 (confirmed to be out of bed).

[0081] 2. Cloud-based precision judgment algorithm: After receiving the state transition event reported by the edge, the cloud executes the following precision judgment logic:

[0082] Time-based weighting: The abnormal weighting coefficient for bed-leaving events at night (22:00-06:00) is set to 1.5, and for daytime events it is 1.0;

[0083] Duration thresholds: Away from bed for less than 15 minutes is marked as "brief away from bed (normal activity)"; 15-120 minutes and with an examination appointment record in the HIS system is marked as "examination out"; Away from bed for more than 120 minutes or overnight and with no examination record is marked as "abnormal away from bed".

[0084] Nursing care level correlation: A high-priority alarm is triggered when a Level 1 nursing patient leaves the bed; a general alert is triggered only when a Level 3 nursing patient is out of bed for more than 2 hours during the day.

[0085] Historical pattern matching: retrieve the patient's out-of-bed behavior time series data over the past 7 days. If the current out-of-bed time deviates from the historical pattern by more than 2 standard deviations, increase the abnormal confidence level.

[0086] Data security and privacy protection in this embodiment include:

[0087] Transmission security: AES-128 encryption is used from the perception layer to the edge layer; TLS 1.3 protocol is used from the edge layer to the cloud; and OAuth 2.0 authorization and JWT token authentication are used for cloud API interfaces.

[0088] Storage security: Patient identity sensitive information (name, ID number) and RFID tag ID are stored separately in the database and associated through a pseudonym mapping table; the cloud database uses field-level encryption and role-based access control (RBAC).

[0089] Privacy compliance: The system deployment complies with the requirements of the Personal Information Protection Law and the Measures for the Administration of Network Security of Medical and Health Institutions. The wristband tag data is automatically archived after the patient leaves the hospital, and the retention period follows the medical record management standards.

[0090] System integration in this embodiment: The system interfaces with the hospital's HIS / EMR system via a standard HL7 FHIR R4 interface.

[0091] Patient basic information synchronization: During admission registration, HIS pushes the patient's basic information to this system through the FHIR Patient resource interface, and this system returns wristband binding confirmation;

[0092] Bed status write-back: After this system determines a bed transfer or discharge event, it writes back the bed status change to the HIS through the FHIR Encounter resource interface;

[0093] Treatment schedule acquisition: This system obtains patient examination appointment information through the FHIR Appointment resource interface, which is used for the "examination leave" determination in the bed leave semantic parsing.

[0094] Example 2

[0095] This embodiment, based on the system described in Embodiment 1, provides a digital hospital bed sensing and management method, including the following steps:

[0096] S1: Patient Identity Binding and Bed Allocation: When a patient is admitted, the system issues a passive RFID wristband with a built-in unique EPC code. By scanning the wristband's QR code or through RFID batch registration, a unique mapping relationship is established between the wristband ID and the patient's basic information, electronic medical record number (EMR ID), and medical insurance registration number in the cloud database. Simultaneously, the system assigns the patient to a specific bed and activates the corresponding digital bed sensor unit.

[0097] S2: Raw Sensing Data Acquisition: The fixed RFID reader deployed at the bed continuously transmits radio frequency activation signals at a preset scanning cycle T (e.g., T=10 seconds) to scan wristband tags within its coverage area. If a valid tag is detected, the tag ID, RSSI value, phase value, and detection timestamp are recorded; if no valid tag is detected for N consecutive scanning cycles (e.g., N=3, i.e., 30 seconds), a "suspected bed leaving" raw event is generated.

[0098] S3: Initial Status Assessment and Data Upload at the Edge: The edge computing gateway receives the raw data reported by each reader / writer within its jurisdiction and executes:

[0099] Eliminate crosstalk between adjacent beds based on RSSI dynamic threshold filtering (e.g., setting the effective RSSI range [-60dBm, -30dBm]);

[0100] Based on the tag presence sequence within a continuous scanning cycle, a finite state machine (FSM) is used for initial judgment: if a valid tag is detected within N consecutive cycles, the initial judgment is "in bed"; if no tag is detected within M consecutive cycles (M≥N), the initial judgment is "out of bed"; if a tag is detected but RSSI fluctuation exceeds a set threshold, it is marked as "position change".

[0101] The initial assessment status, original feature data, and confidence score are encrypted and uploaded to the cloud.

[0102] S4: Cloud Status Calibration and Semantic Resolution: After receiving data reported from the edge, the cloud service center performs multi-level judgments:

[0103] Spatiotemporal consistency verification: The initial margin judgment results are weighted with confidence level by combining the historical data model of this bed (such as the distribution of the patient's previous bed-out time and the ward's daily routine).

[0104] Business rule engine judgment: Set minimum continuous bed-stay time threshold (e.g., ≥4 hours at night is considered valid hospital stay) and minimum bed-off warning threshold (e.g., bed-off for more than 30 minutes and no out-of-bed examination registered in the HIS system);

[0105] Multi-dimensional semantic annotation: Combining patient care level (Level 1 / Level 2 / Level 3), current treatment stage (preoperative / postoperative / rehabilitation), time of bed leaving (daytime / nighttime / early morning), and ward geofence information, bed leaving events are classified as "normal activities", "examination leave", "rehabilitation training", "abnormal bed leaving" or "unauthorized departure from the hospital".

[0106] State transition event log: When a transition occurs in the judgment result, a structured event log is generated, which includes: event type, patient ID, bed ID, start timestamp, end timestamp, confidence level, semantic label, and is written to the time series database.

[0107] S5: Business Process Triggering and Terminal Data Update: Based on the state transition event generated by S4, the system automatically triggers the associated business process.

[0108] If the patient is determined to have "abnormally left the bed" or "left the hospital without authorization", a graded alarm (audio-visual prompts + mobile APP notification) will be pushed to the clinical terminal of the nursing station and recorded in the nursing documentation at the same time.

[0109] If it is determined to be a "bed transfer event" (the original bed is moved and the same tag is detected in the new bed), the bed allocation information in the HIS system will be automatically updated and a bed transfer confirmation notification will be sent to the relevant departments.

[0110] Real-time updates of data dashboards and statistical indicators for each application terminal.

[0111] S6: Cross-level data services and decision support: Users at each level can access the corresponding application terminals through authorization.

[0112] Medical staff can access real-time bed views and alarm handling interfaces through clinical terminals;

[0113] Hospital administrators can access operational analysis reports and predictive insights through management terminals;

[0114] Medical insurance regulators can retrieve IoT-sensed evidence data reinforced with blockchain or digital signatures through monitoring terminals to perform remote audits.

[0115] Regional health management personnel can coordinate and dispatch resources from multiple hospitals through health terminals.

[0116] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A digital hospital bed sensing and management system based on RFID and the Internet of Things, characterized in that, It adopts a four-layer architecture: perception layer - edge processing layer - cloud service center - multi-level application terminal; The sensing layer includes a digital bed sensor unit deployed on each bed. The digital bed sensor unit includes a fixed UHF RFID reader and a passive RFID wristband tag. The fixed RFID reader continuously reads the passive RFID wristband tag in the area of ​​the bed at a preset scanning cycle, and collects the tag ID, RSSI value, phase value and timestamp raw sensing data. The edge processing layer deploys an edge computing gateway to aggregate raw data from multiple digital bed sensor units in the ward, perform signal filtering and noise reduction, perform initial edge judgment of bed leave status, and upload the data after compression and encryption. The cloud service center includes a cloud server, a time-series database, a relational database, a status determination engine, a multi-dimensional semantic parsing engine, and a standard interface service module, which are used for data aggregation and storage, accurate determination of patient bed status, semantic annotation of bed exit events, and standardized data connection with external systems. The multi-level application terminals include clinical terminals, hospital management terminals, medical insurance supervision terminals, and regional health terminals, which provide differentiated business application services to medical staff, hospital managers, medical insurance agencies, and health management departments, respectively.

2. The system according to claim 1, characterized in that, The fixed UHF RFID reader conforms to the ISO18000-6C / EPCGen2 standard, operates at a frequency of 920-925MHz, and has an adjustable transmission power of 10-30dBm. The reader is installed above the headboard of the hospital bed at a height of 1.2m from the ground, and uses a circularly polarized antenna. Through power control and antenna beam orientation optimization, the effective reading distance is limited to 0.5-1.5m, providing independent coverage for each bed. In conjunction with RSSI threshold range [-60dBm, -30dBm] filtering, crosstalk between adjacent beds is suppressed.

3. The system according to claim 1, characterized in that, The passive RFID wristband tag has a built-in unique EPC code, is encapsulated in medical-grade hypoallergenic flexible material, has an IP67 waterproof rating, and the chip's stored information is encrypted using AES-128. The wristband is equipped with a one-time tamper-proof buckle; removing it will destroy the tag's antenna circuit. Upon admission, the wristband ID is uniquely bound to the patient's electronic medical record number, basic information, and medical insurance participation number.

4. The system according to claim 1, characterized in that, The edge computing gateway adopts an industrial-grade ARM architecture device with a built-in finite state machine (FSM) algorithm. It defines three states: S0 in bed, S1 suspected of being out of bed, and S2 confirmed to be out of bed. It completes the state pre-judgment based on the continuous scanning cycle tag detection results and the valid RSSI value, and uploads the preliminary judgment result, original feature data, and confidence score to the cloud through a TLS1.3 encrypted channel.

5. The system according to claim 1, characterized in that, The cloud service center's multi-dimensional semantic parsing engine integrates RSSI spatiotemporal features, patient care level, medical appointment schedule, historical bed leaving behavior patterns, and day / night bed leaving time factors to semantically classify bed leaving events into normal activities, examinations, rehabilitation training, abnormal bed leaving, and unauthorized departure from the hospital. At the same time, it constructs a patient-bed-time triplet time-series data model, records state switching event logs, and persists them as evidence.

6. The system according to claim 1, characterized in that, The cloud service center is interconnected with the hospital's HIS / EMR system, medical insurance platform, and regional health platform through RESTful API and HL7FHIR protocol, supporting the synchronization of basic patient information, automatic writing back of bed status, and acquisition of treatment schedule data; multi-level application terminals achieve one-time perception and multi-party data sharing based on a unified data source.

7. A digital hospital bed sensing and management method based on RFID and the Internet of Things, characterized in that, The system applied to any one of claims 1-6 includes the following steps: S1: Patient identity binding and bed allocation: When a patient is admitted to the hospital, a passive RFID wristband is issued. The wristband ID is uniquely mapped in the cloud with the patient's electronic medical record number, basic information, and medical insurance participation number. The corresponding bed is then allocated and the digital bed sensor unit of that bed is activated. S2: Raw sensing data collection: Fixed RFID readers continuously scan the bed area according to a preset cycle. If a valid tag is detected, the tag ID, RSSI, phase value and timestamp are recorded. If no tag is detected for several consecutive cycles, a suspected bed leaving event is generated. S3: Preliminary status judgment and data upload at the edge: The edge gateway filters interference from neighboring beds through RSSI dynamic threshold, and uses a finite state machine to complete the preliminary judgment of being in bed, suspected leaving bed, body position change, and confirmed leaving bed. After compression and encryption, the data is uploaded to the cloud. S4: Cloud-based state calibration and semantic parsing: Cloud-based spatiotemporal consistency verification, historical model confidence weighting, and business rule engine threshold determination are performed. Multi-dimensional semantic annotation is completed by combining nursing level, treatment schedule, and bed leave time to generate structured state switching event logs. S5: Business Process Triggering and Terminal Updates: When abnormal bed leaving / unauthorized departure from the hospital is detected, a graded alarm is pushed to the clinical terminal and nursing documents are synchronized; when bed transfer events are detected, HIS bed allocation information is automatically updated; data dashboards and operational statistics indicators at all levels of terminals are refreshed in real time. S6: Cross-level data services and decision support: Medical staff, hospital management, medical insurance supervision, and regional health departments can access differentiated data services through authorized terminals to achieve real-time bed monitoring, operational analysis, remote medical insurance auditing, and regional bed emergency dispatch.

8. The method according to claim 7, characterized in that, In step S4, the business rules engine preset thresholds include: leaving the bed for more than 30 minutes between 22:00 and 06:00 at night without an appointment is marked as abnormal leaving the bed; leaving the bed for more than 120 minutes during the day without a medical appointment is marked as abnormal leaving the bed; leaving the bed for more than 4 hours without system registration is marked as unauthorized leaving the hospital; leaving the bed for 15-120 minutes and matching the HIS examination appointment record is marked as examination leaving the hospital.

9. The method according to claim 7, characterized in that, The system data security adopts layered protection: AES-128 encryption is used from the perception layer to the edge layer, TLS1.3 transmission encryption is used from the edge to the cloud, and OAuth2.0+JWT token authentication is used for cloud APIs; sensitive patient identity information and RFID tag IDs are stored separately under pseudonyms, in conjunction with RBAC role-based access control and field-level database encryption.

10. The method according to claim 7, characterized in that, The system integrates with HIS / EMR based on the HL7 FHIR R4 interface. It synchronizes basic patient information through Patient resources, writes back bed status changes through Encounter resources, and obtains treatment appointment schedules through Appointment resources, supporting intelligent semantic determination of bed exit events.

Images