Hospital intelligent internet-of-things system

Abstract

The invention discloses a hospital intelligent internet-of-things system, and the system comprises a data collection unit which is used for synchronously obtaining a voice signal and a video image in a ward; an analysis decision unit which is in communication connection with the data acquisition unit and is used for carrying out fusion analysis on the voice signal and the video image and generating an event instruction; the communication unit is used for sending the event instruction in a wireless mode; the first terminal is arranged at a nurse station and is used for receiving and prompting event instructions related to nursing; and the second terminal is arranged in a security duty room and is used for receiving and prompting event instructions related to security. According to the scheme, emergencies in a ward can be timely and correspondingly recognized, and early warning is carried out.

Description

Technical Field

[0001] The present invention relates to the fields of intelligent voice and intelligent vision, and in particular to a hospital intelligent Internet of Things system based on voice and vision processing technologies. Background Art

[0002] Ward management is a core component of hospitals' efforts to safeguard the lives and property of patients. However, traditional manual inspections not only consume significant human resources but also suffer from delayed responses and poor timeliness, making it difficult to establish a comprehensive, all-weather security barrier. While information technology has driven the widespread adoption of intercom systems for medical staff, existing solutions still have significant limitations: For patients unable to operate physical call devices due to paralysis or hand injuries, their requests for help are difficult to promptly address. Furthermore, existing systems lack the ability to proactively identify and rapidly warn of unexpected incidents within wards, such as violent conflict and theft, preventing hospital administrators from being immediately notified and intervening. Summary of the Invention

[0003] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a hospital smart Internet of Things system that can respond to emergencies in the ward in a timely manner and actively identify and warn.

[0004] To achieve the above object, the present invention provides the following technical solutions: A hospital smart Internet of Things system includes: a data acquisition unit for synchronously acquiring voice signals and video images in a ward; an analysis and decision-making unit, communicatively connected to the data acquisition unit, for performing a fusion analysis of the voice signals and video images and generating event instructions; a communication unit for wirelessly transmitting the event instructions; a first terminal, arranged at a nurse's station, for receiving and prompting event instructions related to nursing; and a second terminal, arranged in a security duty room, for receiving and prompting event instructions related to security.

[0005] As a further improvement of the present invention, the data acquisition unit includes: a voice acquisition subunit for converting the voice signal into text information in real time; and a visual acquisition subunit for continuously acquiring the video image.

[0006] As a further improvement of the present invention, the voice collection subunit starts an interactive mode after detecting a preset wake-up word to receive a voice request associated with a specific bed.

[0007] As a further improvement of the present invention, the analysis and decision-making unit includes: a keyword recognition module for extracting preset help keywords from the text information; a semantic analysis module for parsing the semantic content of the text information; a visual analysis module for identifying abnormal behavior in the video image; and a fusion judgment module for generating the event instructions by integrating the keywords, semantic content and abnormal behavior.

[0008] As a further improvement of the present invention, the abnormal behavior includes at least one of quarreling, violence, stealing or falling.

[0009] As a further improvement of the present invention, the communication unit adopts LoRa, NB-IoT or Sigfox low-power wide-area wireless protocol for signal transmission.

[0010] As a further improvement of the present invention, the event instruction carries a bed identifier, so that the first terminal or the second terminal can accurately locate the event source.

[0011] As a further improvement of the present invention, the visual acquisition subunit is at least one of a wide-angle camera, a depth camera or an infrared camera.

[0012] As a further improvement of the present invention, it further includes: a data storage unit for storing the voice signal, video image and corresponding analysis results to support post-audit.

[0013] As a further improvement of the present invention, the analysis and decision-making unit, while generating event instructions, prioritizes the same event and preferentially sends high-priority event instructions through the communication unit.

[0014] The beneficial effect of the present invention is that through the coordinated work of the data acquisition unit, the analysis and decision-making unit, the communication unit and the first and second terminals, the system realizes the real-time collection and fusion analysis of voice and video information in the ward, and can automatically generate event instructions and accurately distribute them to the nurse station or security duty room, thereby significantly improving the response speed and coverage of ward management, effectively solving the problems of low efficiency of traditional manual inspections and limited functions of existing intercom systems, and comprehensively ensuring the safety of patients and medical staff. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a system block diagram of the present invention. DETAILED DESCRIPTION

[0016] The present invention will be further described below with reference to the embodiments shown in the accompanying drawings.

[0017] Reference Figure 1 As shown, A hospital smart IoT system, comprising: Data acquisition unit, used to synchronously acquire voice signals and video images in the ward; The analysis and decision-making unit is connected to the data acquisition unit for performing fusion analysis on the voice signal and the video image and generating event instructions; a communication unit, configured to send event instructions wirelessly; The first terminal is set at the nurse station and is used to receive and prompt instructions for nursing-related events; The second terminal is set in the security duty room and is used to receive and prompt security-related event instructions.

[0018] The data acquisition unit uses a microphone array and camera to capture real-time sounds and images from within the ward, converting them to analog and digital signals before sending them to the analysis and decision-making unit. This unit then uses a fusion algorithm to correlate sound and visual events, generating event instructions containing event type and location information. The communication unit then broadcasts these instructions to the first and second terminals via the LoRa wireless module. Upon receiving a nursing instruction, the first terminal immediately alerts the nurse with sound and light, while upon receiving a security instruction, the second terminal immediately alerts the security guard with sound and light. This process completes a closed loop of real-time perception, analysis, and distribution of ward status, enabling nursing or security personnel to arrive at the scene as quickly as possible, significantly reducing incident response time.

[0019] Preferably, the data acquisition unit includes a voice acquisition subunit and a visual acquisition subunit; the voice acquisition subunit is used to convert voice signals into text information in real time; and the visual acquisition subunit is used to continuously acquire video images.

[0020] The voice acquisition subunit uses an embedded speech recognition chip to perform noise reduction, endpoint detection, and feature extraction on analog voice signals, and outputs text sequences in real time using a local acoustic model. The visual acquisition subunit uses a CMOS image sensor to continuously output a YUV-formatted video stream. Both streams of data are synchronously fed into the analysis and decision-making unit via a parallel bus, ensuring strict temporal alignment between voice and visual events. This provides a high-precision data foundation for subsequent fusion analysis and avoids misjudgments caused by sampling time differences.

[0021] To facilitate daily patient interactions, the voice acquisition subunit in this solution starts the interaction mode after detecting the preset wake-up word to receive voice requests associated with a specific bed.

[0022] The wake-up word model is deployed in the DSP of the voice acquisition subunit. Upon detecting a wake-up word such as "Xiao Hu Xiao Hu," the system immediately raises the wake-up signal and opens a 5-second interaction window. During this window, only voice commands prefixed with "Bed Number X," such as "Bed Number 3, calling," are collected and recognized, thus binding subsequent voice requests to a specific bed. This mechanism significantly reduces power consumption and privacy risks associated with continuous monitoring while ensuring the uniqueness and traceability of voice requests.

[0023] Specifically, the analysis and decision-making unit includes a keyword recognition module, a semantic analysis module, a visual analysis module and a fusion judgment module; the keyword recognition module is used to extract preset help keywords from text information; the semantic analysis module is used to parse the semantic content of text information; the visual analysis module is used to identify abnormal behavior in video images; the fusion judgment module is used to integrate keywords, semantic content and abnormal behavior to generate event instructions.

[0024] The keyword recognition module uses a Trie tree to quickly match keywords such as "help" and "uncomfortable." The semantic analysis module uses a lightweight BERT model to extract intent. The visual analysis module uses OpenPose to detect key points on the human body and calculate velocity vectors to identify falls. The fusion judgment module uses a weighted voting mechanism to integrate the three types of results, generating an event command only when at least two types of results point to the same event type. This multimodal fusion strategy significantly reduces single-modal false positives and minimizes the false positive rate.

[0025] Preferably, the abnormal behavior includes at least one of arguing, violence, stealing or falling.

[0026] The visual analysis module has four built-in behavior detection sub-models: Dispute detection: based on the distance threshold between two people and the frequency of rapid gesture waving; Violence detection: based on drastic changes in human posture and rapid movement trajectory; Theft detection: based on a hand reaching into someone else's personal belongings area and staying there for more than 3 seconds; Fall detection: Based on the sudden drop in height of the center of the human body and the horizontal velocity close to zero.

[0027] When any sub-model triggers a confidence threshold, it outputs an anomaly label to the fusion judgment module. By segmenting behavior types, the system can automatically match the most appropriate handling process for different events, improving processing efficiency.

[0028] In a further configuration, the communication unit uses LoRa, NB-IoT or Sigfox low-power wide-area wireless protocols for signal transmission.

[0029] The communication unit integrates a switchable RF front-end, automatically selecting LoRa (433 MHz / 470 MHz), NB-IoT (Band 8), or Sigfox (868 MHz) depending on the deployment environment. In LoRa mode, it uses an SF7 spreading factor and a 250 bps data rate to achieve 200 m indoor wall-penetrating coverage. In NB-IoT mode, it leverages carrier base stations for wide-area coverage. In Sigfox mode, ultra-narrowband technology compresses single-event commands to 12 bytes, resulting in a mere 160 ms airtime. This solution ensures low-power operation.

[0030] In order to facilitate timely response by nursing staff and security personnel, the event instruction carries a bed identifier so that the first terminal or the second terminal can accurately locate the source of the event.

[0031] The event command frame format is {frame header (1 byte), event type (1 byte), bed ID (2 bytes), timestamp (4 bytes), CRC16 (2 bytes)}. The bed ID is bound to the bed coordinate mapping table in the ward floor plan. Upon receiving the command, the first terminal parses the bed ID, highlights the corresponding bed on the electronic floor plan, and announces "Bed 3 calling." Upon receiving the security command, the second terminal automatically displays the real-time image of the bed in conjunction with the video surveillance client. This mechanism enables personnel to accurately locate the bed without manual searching.

[0032] Preferably, the visual acquisition subunit is at least one of a wide-angle, depth or infrared camera.

[0033] Preferably, an additional data storage unit may be provided in the system for storing voice signals, video images and corresponding analysis results to support post-audit.

[0034] The data storage unit uses a 32GB eMMC chip, which uses a cyclic overwrite strategy to store the last 30 days of data. Each data segment is named with an "event ID + timestamp" and includes JSON-formatted metadata recording the event type, triggering mode, and disposition. In the event of a medical dispute, administrators can export data for a specified time period via USB or OTA, supporting synchronized playback of H.264 video and PCM audio. This feature provides hospitals with a traceable chain of evidence, reducing legal risks.

[0035] In a further configuration, the analysis and decision-making unit prioritizes the same event while generating the event instruction, and preferentially sends the high-priority event instruction through the communication unit.

[0036] Priority grading is based on event type and patient risk level: Messages with keywords like "fall," "violence," and "life-saving" are assigned high priority; general calls are assigned medium priority; and device status information is assigned low priority. High-priority event commands are inserted into the communication queue head and have the ACK mechanism enabled. If no ACK is received from the terminal within 3 seconds, the command is automatically retransmitted twice. Medium and low-priority events use the standard queue and the no-ACK mechanism. This differentiated transmission strategy ensures that high-risk events receive the most timely response.

[0037] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A hospital smart IoT system, characterized by: include: Data acquisition unit, used to synchronously acquire voice signals and video images in the ward; An analysis and decision-making unit, in communication with the data acquisition unit, configured to perform a fusion analysis on the voice signal and the video image and generate an event instruction; a communication unit, configured to send the event instruction wirelessly; The first terminal is set at the nurse station and is used to receive and prompt instructions for nursing-related events; The second terminal is set in the security duty room and is used to receive and prompt security-related event instructions.

2. The hospital smart IoT system according to claim 1, characterized in that: The data acquisition unit includes: A voice collection subunit, configured to convert the voice signal into text information in real time; The visual acquisition subunit is used to continuously acquire the video images.

3. The hospital smart IoT system according to claim 2, characterized in that: The voice collection subunit starts the interactive mode after detecting the preset wake-up word to receive the voice request associated with the specific bed.

4. The hospital smart IoT system according to claim 1 or 2, characterized in that: The analysis and decision-making unit includes: A keyword recognition module, used to extract preset help keywords from the text information; A semantic analysis module, used to analyze the semantic content of the text information; a visual analysis module, configured to identify abnormal behavior in the video image; The fusion judgment module is used to generate the event instruction by integrating the keywords, semantic content and abnormal behavior.

5. The hospital smart IoT system according to claim 4, characterized in that: The abnormal behavior includes at least one of quarreling, violence, stealing or falling.

6. The hospital smart IoT system according to claim 1, characterized in that: The communication unit uses LoRa, NB-IoT or Sigfox low-power wide-area wireless protocols for signal transmission.

7. The hospital smart IoT system according to claim 1, characterized in that: The event instruction carries a bed identifier, so that the first terminal or the second terminal can accurately locate the event source.

8. The hospital smart IoT system according to claim 1, characterized in that: The visual acquisition subunit is at least one of a wide-angle camera, a depth camera or an infrared camera.

9. The hospital smart IoT system according to claim 1, characterized in that: Also includes: The data storage unit is used to store the voice signal, video image and corresponding analysis results to support post-audit.

10. The hospital smart IoT system according to claim 1, characterized in that: The analysis and decision-making unit, while generating event instructions, prioritizes the same event and preferentially sends high-priority event instructions through the communication unit.

Images