Intelligent accompanying and visiting support system and method for intensive care

Abstract

The invention provides an intelligent accompanying and visiting support system and method for intensive care, relates to the technical field of medical information, and aims to solve the technical problems that in the related technology, a critical patient lacks continuous emotion support, the visiting experience of family members is poor, the family members are easily impacted by emotions, and the doctor-patient communication efficiency is low. Comprising the steps of obtaining original environment information flow; the original environment information flow is decomposed into a plurality of emotional information quanta, and each emotional information quantum is a discrete data unit bearing specific information; acquiring a patient emotional state vector representing the current physiological or emotional state of the patient; for each emotion information quantum, calculating a quantum transmission probability based on the patient emotion state vector and a preset resonance cavity configuration matrix; and based on the quantum transmission probability, executing a preset information processing operation on each emotion information quantum to generate a patient-oriented accompanying information stream and a family member-oriented visiting information stream.

Description

Technical Field

[0001] This application relates to the field of medical information technology, and in particular to an intelligent companionship and visitation support system and method for intensive care. Background Technology

[0002] The Intensive Care Unit (ICU) is a crucial facility in modern medicine for treating critically ill patients. However, its high-tech, high-intensity environment also presents significant physical and psychological challenges for patients. Patients not only endure the pain of the disease itself but also face an environment filled with cold instruments, unfamiliar faces, and constant auditory and visual stimulation. This environment can easily lead to negative emotions such as loneliness, anxiety, fear, and even delirium. These emotions not only affect the patient's subjective experience but have also been shown to potentially have an adverse impact on the recovery process.

[0003] Meanwhile, the strict visitation policies in the ICU also pose challenges to patients' families. Family visits are severely limited, making it difficult to provide continuous emotional support to patients. While traditional remote video visits have addressed the "face-to-face" issue to some extent, the experience is often crude and one-way. Families may see scenes filled with medical equipment, even disturbing images of exposed wounds and drainage tubes, which not only fail to reassure patients but may also exacerbate the family's own panic and anxiety. Furthermore, the constant noise from medical equipment alarms in the ICU is transmitted to families via video call, affecting communication and emotional exchange.

[0004] While healthcare workers bear the primary responsibilities of monitoring and communication, they face immense work pressure, making it difficult to provide continuous, personalized emotional care to each patient. Communication between healthcare workers and families is also often inefficient due to time constraints and information asymmetry.

[0005] Existing technological attempts, such as playing preset music for patients or providing simple video call functions, are typically isolated and passive. They lack the ability to perceive the patient's real-time status and cannot adjust companionship strategies according to the dynamic changes in the patient's emotions. They also fail to systematically solve the problem of "information overload" and "emotional shock" coexisting in remote visits, namely, how to effectively convey necessary information to family members while filtering out sensitive content that may cause trauma. Summary of the Invention

[0006] This application provides an intelligent companionship and visitation support system and method for intensive care, which aims to improve the technical problems in related technologies, such as the lack of continuous emotional support for critically ill patients, poor visitation experience for family members who are easily emotionally impacted, and low efficiency of doctor-patient communication.

[0007] To achieve the above objectives, the embodiments of this application adopt the following technical solutions: In a first aspect, this application provides an intelligent companionship and visitation support method for intensive care, comprising: acquiring an original environmental information stream, the original environmental information stream including real-time environmental data characterizing the intensive care environment and companionship data characterizing preset interactive content; decomposing the original environmental information stream into multiple emotional information quantum, each of the emotional information quantum being a discrete data unit carrying specific information; acquiring a patient emotional state vector characterizing the patient's current physiological or emotional state; for each of the emotional information quantum, calculating its quantum transmission probability based on the patient emotional state vector and a preset resonance cavity configuration matrix; and performing a preset information processing operation on each of the emotional information quantum based on the quantum transmission probability to generate a companionship information stream for the patient and a visitation information stream for family members.

[0008] In one possible implementation of the first aspect, the step of calculating the quantum transmission probability of each of the emotional information quantum includes: calculating a perturbation potential value for each of the emotional information quantum, the perturbation potential value being used to characterize the potential degree to which the emotional information quantum has a negative impact on the patient's emotional state vector; calculating a resonance affinity for each of the emotional information quantum, the resonance affinity being used to characterize the degree of consistency between the emotional information quantum and the target companionship strategy defined by the resonance cavity configuration matrix; and obtaining the quantum transmission probability by combining the perturbation potential value, the resonance affinity, and the patient's emotional state vector.

[0009] In one possible implementation of the first aspect, the step of obtaining the quantum transmission probability by combining the perturbation potential value, the resonance affinity, and the patient's emotional state vector includes: constructing a state adjustment function that maps the patient's emotional state vector to a dynamic adjustment factor; combining the resonance affinity with the perturbation potential value to obtain a basic transmission intention value; and modulating the basic transmission intention value using the dynamic adjustment factor to obtain the quantum transmission probability.

[0010] In one possible implementation of the first aspect, the step of performing a preset information processing operation on each of the emotional information quanta based on the quantum transmission probability includes: comparing the quantum transmission probability with a first preset threshold; if the quantum transmission probability is lower than the first preset threshold, performing a quantum decay operation on the emotional information quantum to filter or replace it when generating the companionship information stream; if the quantum transmission probability is not lower than the first preset threshold, performing a quantum resonance transmission operation on the emotional information quantum to retain or enhance it when generating the companionship information stream.

[0011] In one possible implementation of the first aspect, the preset information processing operation further includes: comparing the quantum transmission probability with a second preset threshold for family members; if the emotional information quantum is a sensitive information quantum and its quantum transmission probability is lower than the second preset threshold, then the image or sound corresponding to the sensitive information quantum is masked when generating the visitation information stream.

[0012] In one possible implementation of the first aspect, the method further includes: establishing an emotional homeostasis regulation mechanism; the emotional homeostasis regulation mechanism continuously monitors the patient's emotional state vector, and when it deviates from a preset homeostasis target range, adjusts at least one parameter of the resonance cavity configuration matrix through a feedback loop to guide the patient's emotional state vector back into the homeostasis target range.

[0013] In one possible implementation of the first aspect, the method further includes: establishing a predictive feedforward intervention mechanism; the predictive feedforward intervention mechanism identifies precursor information quantum associated with emotional information quantum with subsequent high perturbation potential value; when the precursor information quantum is detected, actively introducing compensatory emotional information quantum with high resonance affinity into the companion information stream to intervene in the patient's emotional state vector in advance.

[0014] In one possible implementation of the first aspect, the step of obtaining the patient's emotional state vector includes: acquiring facial image data of the patient through a non-contact sensor; extracting micro-expression feature sequences from the facial image data; and inputting the micro-expression feature sequences into a pre-trained emotional state analysis model to obtain the patient's emotional state vector.

[0015] Secondly, this application provides an intelligent companionship and visitation support system for intensive care, comprising: an information flow acquisition module for acquiring an original environmental information flow, the original environmental information flow including real-time environmental data characterizing the intensive care environment and companionship data characterizing preset interactive content; an information quantization module for decomposing the original environmental information flow into multiple emotional information quanta, each of the emotional information quanta being a discrete data unit carrying specific information; a state perception module for acquiring a patient emotional state vector characterizing the patient's current physiological or emotional state; a probability calculation module for calculating the quantum transmission probability of each emotional information quanta based on the patient emotional state vector and a preset resonance cavity configuration matrix; and an information flow generation module for performing preset information processing operations on each emotional information quanta based on the quantum transmission probability to generate a companionship information flow for the patient and a visitation information flow for family members.

[0016] In one possible implementation of the second aspect, the probability calculation module is configured to: calculate a perturbation potential value for each of the emotional information quantum, the perturbation potential value being used to characterize the potential degree of negative influence of the emotional information quantum on the patient's emotional state vector; calculate a resonance affinity for each of the emotional information quantum, the resonance affinity being used to characterize the degree of consistency between the emotional information quantum and the target companionship strategy defined by the resonance cavity configuration matrix; and combine the perturbation potential value, the resonance affinity, and the patient's emotional state vector to obtain the quantum transmission probability. Attached Figure Description

[0017] Figure 1 A flowchart illustrating an intelligent companionship and visitation support method for intensive care, provided for some embodiments of this application; Figure 2 This is a schematic diagram of the structure of an intelligent companionship and visitation support system for intensive care, provided for some embodiments of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0019] In the following description, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] Furthermore, in this application, directional terms such as "upper," "lower," "left," and "right" may be defined relative to the orientation of the components shown in the accompanying drawings. It should be understood that these directional terms can be relative concepts, used for relative description and clarification, and may change accordingly depending on the orientation of the components in the accompanying drawings.

[0021] In this application, unless otherwise expressly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral part; it can be a direct connection or an indirect connection through an intermediate medium. Furthermore, the term "electrical connection" can refer to the manner in which an electrical connection is used to achieve signal transmission.

[0022] As used herein, “about,” “approximately,” or “approximately” includes the stated value and a reference value within an acceptable range of deviation from the given value, characterized in that the acceptable range of deviation is determined by a person skilled in the art taking into account the measurement under discussion and the error associated with the measurement of the given quantity (i.e., the limitations of the measurement method).

[0023] This application provides an intelligent companionship and visitation support system and method for intensive care units (ICUs). The method constructs a dynamic and personalized information interaction environment for patients in the ICU, which is abstracted as an "emotional resonance chamber." All information entering this environment, whether from alarms of medical equipment, conversations of medical staff, or videos and music pre-stored by family members, is considered as discrete "emotional information quanta." The system, by sensing the patient's emotional and physiological state in real time and combining it with preset companionship strategies, calculates a dynamic "quantum transmission probability" for each "emotional information quantum." This probability determines whether the information quantum is allowed to be "resonantly transmitted" to the patient and family, or whether it needs to be processed by "quantum attenuation" (such as filtering, shielding, or replacement). In this way, the system can proactively create an intelligent environment conducive to patient recovery while providing effective information and emotional support to family members.

[0024] The following will refer to Figure 1 This paper elaborates on the method provided in a specific embodiment of the present application. Figure 1 This is a flowchart illustrating an intelligent companionship and visitation support method for intensive care disclosed in one embodiment of this application.

[0025] S100: Obtain the raw environment information stream.

[0026] The system needs to collect all information related to the patient, environment, and preset interactive content to form a multimodal, heterogeneous raw environmental information flow. The original environmental information stream includes real-time environmental data and companion data.

[0027] The real-time environmental data refers to data generated in real time within the intensive care unit. This dynamically changes data, directly reflecting the patient's immediate physical and informational environment. In this embodiment, the real-time environmental data may include video stream data captured by cameras deployed at the patient's bedside. Audio stream data acquired through a microphone array and real-time physiological parameter data obtained through connection to the monitoring device's data interface. Video stream data Visual information within the ward was captured, such as the activities of medical staff, the appearance of medical equipment, and changes in lighting. Audio stream data was also included. It captured all sounds in the environment, such as equipment alarms, conversations, and background noise. Physiological parameter data. These include heart rate, respiratory rate, and blood oxygen saturation, and the above data are important objective bases for assessing the patient's condition.

[0028] The companionship data refers to data pre-set or uploaded by family members or medical staff to provide emotional support or information transmission to patients. Unlike the passive generation of real-time environmental data, companionship data is actively provided content. In this embodiment, companionship data may include family video clips uploaded by family members. Encouraging voice messages And a music library tailored to patients' cultural background, religious beliefs, or personal preferences. and image library (Such as pictures of natural scenery and family photos). This data forms the basis for the system's proactive companionship and intervention.

[0029] For example, suppose that within a time window, the system acquires the raw environmental information stream. One segment includes: a sequence of video frames captured by a camera showing a nurse adjusting an IV pump for a patient; a microphone recording the "beep-beep" sound from a monitor and indistinct voices in the distance; and a data interface receiving heart rate data. The signal is transmitted at a rate of [number] times per minute; simultaneously, a pre-stored voice message from a family member, "Be strong, we are all waiting for you to come back," is in the playback queue. All these data together constitute the current situation. .

[0030] S200: Decompose the original environmental information flow into multiple emotional information quantum.

[0031] A continuous stream of raw environmental information was obtained. Subsequently, for more refined analysis and processing, it is necessary to transform the continuous data stream into discrete units with clearly defined semantics or characteristics. This application defines these discrete units as "Affective Information Quantums," denoted as... Every quantum of emotional information It is a structured discrete data unit that carries specific information.

[0032] The decomposition process employs different strategies depending on the data modality.

[0033] For video stream data The system can employ camera change detection or scene segmentation algorithms to divide a continuous video stream into multiple event segments. For example, the process of a nurse approaching, operating equipment, and leaving can be identified as a complete event and encapsulated into an emotional information quantum. Simultaneously, to address privacy protection needs in visitation scenarios, the system runs an object detection model to identify sensitive areas in video frames in real time, such as exposed wounds or catheters on patients, and labels the location information of these areas as one or more specific emotional information quantumes. For example, a video quantum... It can be represented as: ,in It is a unique identifier. and These are the start and end timestamps. These are key visual feature vectors extracted from video frames (e.g., extracted via a convolutional neural network CNN). This is a list of coordinates of the identified sensitive areas; if none are found, it will be empty. It is understood that the target detection model can be trained by those skilled in the art based on a large amount of relevant data to create an image recognition model; this is a common practice among those skilled in the art, and this application does not impose any restrictions on it.

[0034] For audio stream data The system can employ Sound Event Detection (SED) technology. This technology can segment a continuous audio stream into different sound events and classify each event. For example, the system can identify "high-frequency alarm from a monitor," "human voice (male / female)," "soothing music," and "device operating noise." Each identified sound event is encapsulated into an emotional information quantum. For example, an audio quantum... It can be represented as: ,in It is the acoustic feature vector of audio (such as Mel frequency cepstral coefficients MFCC). It is an event category label, such as "alarm_high_urgency".

[0035] For physiological parameter data Companionship data , Emotional information quanta are typically discrete or easily discretized. For example, a voice message, a photograph, or a complete piece of music can all be directly considered as a quantum of emotional information.

[0036] After this step, the continuous, mixed raw environmental information flow is... It is transformed into a structured quantum set of emotional information. .

[0037] S300: Obtain the patient's emotional state vector, which represents the patient's current physiological or emotional state.

[0038] To achieve truly personalized and adaptive companionship, the system must be able to perceive the patient's state in real time and accurately. This application introduces a "patient emotional state vector." The concept is a multidimensional vector designed to quantify a patient’s overall emotional and physiological state at a given moment.

[0039] In this embodiment, obtain The process primarily relies on non-contact sensing technology to minimize disruption to the patient. Specifically, cameras deployed in front of the bed are used not only to collect environmental video but also to continuously capture high-resolution facial images of the patient.

[0040] Get The specific steps are as follows: S310: Acquires facial image data of the patient through a non-contact sensor. The camera continuously captures the patient's face at a preset frame rate (e.g., 30 frames per second).

[0041] S320: Extract micro-expression feature sequences from the facial image data. Since critically ill patients may be unable to make large facial expressions, the system focuses on subtle, involuntary movements of facial muscles, i.e., micro-expressions. The system first accurately locates the facial region using a face detection algorithm, and then uses facial keypoint detection technology (such as using a depth-aligned network) to track the coordinate changes of dozens of facial keypoints (such as the corners of the eyes, corners of the mouth, and eyebrows). The displacement, velocity, and acceleration of these keypoints between consecutive frames constitute the original micro-expression feature sequence.

[0042] S330: The micro-expression feature sequence is input into a pre-trained emotion state analysis model to obtain the patient's emotion state vector. This application constructs a temporal emotion analysis model, for example, a recurrent neural network (RNN) based on a long short-term memory network (LSTM) or a gated recurrent unit (GRU). This model takes the micro-expression feature sequence as input and outputs a multidimensional... .

[0043] The emotional state vector It can be defined within a standardized emotional space, such as the Valence-Arousal two-dimensional space. Under this definition, Among them, valence Indicates the positive or negative nature of emotions (from very negative to very positive), arousal level. Indicates the intensity of an emotion (from calm to excited).

[0044] For example, if the system analyzes subtle movements such as a patient's furrowed brows and slightly downturned lips, the model might output an emotion state vector. This indicates that the patient is currently in a moderately intense negative emotional state (such as pain or anxiety). If the patient's facial muscles are relaxed, it may output... This indicates that the patient is in a calm and slightly positive state.

[0045] In some alternative implementations, in order to improve The accuracy and robustness of the data can be enhanced by integrating physiological parameter data obtained from monitoring devices. For example, sympathetic / parasympathetic activity indices obtained from heart rate variability (HRV) analysis can be added as an additional dimension to the analysis. Alternatively, it can be used as another input, along with micro-expression features, and fed into a multimodal fusion sentiment analysis model.

[0046] S400: For each of the emotional information quantum, calculate its quantum transmission probability based on the patient's emotional state vector and the preset resonance cavity configuration matrix.

[0047] This step aims to process each quantum of emotional information to be processed. Calculate a "quantum teleportation probability" This probability value This determines the system's processing strategy for this information quantum: the probability approaches... This means that the information quantum should be transmitted without obstruction; the probability approaches 1 / 2. This means that the information quantum should be strongly attenuated or filtered.

[0048] The calculation process needs to take into account three factors: the nature of the information quantum itself, the pre-set companionship strategy, and the patient's current state.

[0049] First, a "resonance cavity configuration matrix" needs to be introduced. The concept of companionship. This is a set of parameters set by family members or medical staff to define the ideal companionship environment. It concretizes the abstract desire for companionship into calculable parameters. Including but not limited to: Content Preference Vector This defines the degree of preference for different content categories (such as classical music, religious chants, natural scenery, and family videos). For example, .

[0050] Sensitivity settings Defines the filtering strength for different types of potentially negative information (such as medical alerts, nude images).

[0051] Interactive initiative This defines the frequency and intensity of the system's proactive push of companionship content.

[0052] Quantum transmission probability The calculation process can be broken down into the following sub-steps: S410: Calculate a perturbation potential value for each of the said emotional information quantum.

[0053] Disturbance potential value It is a scalar used to characterize the quantum of emotional information. Patient emotional state vector The potential for negative impact, its range is: A larger value indicates a greater likelihood and intensity of negative impact. This value is calculated based on the inherent characteristics of information quantum itself.

[0054] For audio quantum ,That It is mainly determined by its acoustic characteristics. This application constructs a perturbation potential evaluation function. In one specific embodiment, the perturbation potential assessment function This could be a standardized weighted sum model. For example, ,in It's the clock speed. It is energy. It is spectral entropy. It normalizes the input physical quantity to Functions on intervals These are preset weights, for example, they can be set to... , It is to crop the result to A function that operates within a range to ensure that the output value is strictly within that range.

[0055] For example, a monitor emits a high frequency ( High-energy alarm sounds, with low spectral entropy (pure but sharp sound), will be assigned a high perturbation potential value after being calculated by this function, for example... A background piece of music with a stable volume and a soothing melody, whose various parameters are all within the comfort zone, may be... .

[0056] For video quantum ,That Primarily determined by its visual content. This application constructs another perturbation potential assessment function. Among them, the function Specifically, it can be defined as: if the list of sensitive areas If not empty, then ;otherwise .function The intensity of motion in a video image can be quantified by calculating the average amplitude of the optical flow field between consecutive video frames, and this amplitude can be normalized to... Interval.

[0057] For example, if a sensitive area marked as a "wound" is detected in a frame, then Regardless of the movement of the image, its All will be set to at least .

[0058] S420: Calculate a resonance affinity for each of the said emotional information quantum.

[0059] Resonance Affinity It is a scalar used to characterize the quantum of emotional information. Resonance cavity configuration matrix The degree of consistency of the defined target companionship strategy, its value range is also [missing information]. The larger the value, the more it matches the preset companionship preference.

[0060] The calculation of this value involves combining the content attributes of the information quantum with... The process of matching parameters in the database.

[0061] For an emotional information quantum that serves as companion data (such as music) Its resonance affinity The calculation can be directly queried. Content preference vectors in .

[0062]

[0063] For example, if the preferences set by family members are When the system is about to play a Beethoven piano piece (whose metadata is tagged "classical"), this emotional information quantum... It was calculated as .

[0064] For a quantum of environmental information generated in real time, such as a dialogue between medical staff and a patient... ,That The computation process can be more detailed. The system first obtains the text content of the dialogue through an ASR (Actions-Speech) engine. Then, this text is fed into a pre-trained sentiment classification model (e.g., a BERT-based model fine-tuned for Chinese medical dialogue scenarios). This model outputs a... Emotional polarity score within the range The fraction is obtained through a linear transformation. Mapped to The interval is used as its resonance affinity. For example, if the dialogue content is "You recovered very well today," the model might output... ,but If the instruction is neutral, such as "Please turn over," the model might output... ,but .

[0065] S430: Combine the perturbation potential value, the resonance affinity, and the patient's emotional state vector to obtain the quantum transmission probability.

[0066] The final step is to apply the above calculations... and , and the real-time patient emotional state vector Combined, this generates the final quantum teleportation probability. This combination reflects the system's dynamic adaptive capability.

[0067] In this embodiment, the joint process is implemented through a specific nonlinear mapping function. First, a state adjustment function is constructed. It will vectorize the patient's emotional state. Mapped to a dynamic adjustment factor The design principle of this function is that the more negative and agitated the patient's state, the stronger the moderating factor and the stronger the "willingness" for systemic intervention.

[0068] For example, if The state adjustment function can be defined as follows: ,in It is a hyperparameter, for example When the patient is calm or positive ( or ), It does not play a regulatory role. When the patient is in a state of anxiety, such as ,but This factor, which is greater than 1, will enhance the system's "determination" to filter information.

[0069] Next, the resonance affinity will be... With perturbation potential value By combining these values, a basic transmission intention value is obtained. A simple way to combine them is... This value may be negative.

[0070] Finally, use dynamic adjustment factors. The underlying transmission intention value is modulated and mapped to a sigmoid function. The interval is used to obtain the final quantum transmission probability. .

[0071]

[0072]

[0073] in, It is the slope parameter that controls the steepness of the curve. For example, it can be taken as... . It is an offset threshold used to calibrate the overall "openness" of the system; for example, it can be taken as... In the denominator It played a key regulatory role: when the patient was in a bad mood ( The absolute value of the entire exponent decreases, causing the output of the sigmoid function to more drastically tend towards the zero line. or This means that the system's decisions will be more "black and white," and it will tend to filter out any potentially risky information more forcefully.

[0074] Example of microcalculus: Suppose there exists an emotional information quantum. It was a monitor alarm sound.

[0075] S410: Its perturbation potential value is obtained through acoustic analysis. .

[0076] S420: This alarm sound does not belong to any preferred content; its resonance affinity is... (Give a lower default value).

[0077] S430: Assume the patient's emotional state vector at this moment is... (anxiety).

[0078] Calculate the dynamic adjustment factor: .

[0079] Calculate the basic transmission intention value: .

[0080] Calculate the final transmission probability:

[0081] This extremely low transmission probability ( This will cause the system to perform a powerful quantum decay operation on the alarm sound.

[0082] S500: Based on the quantum transmission probability, perform a preset information processing operation on each of the emotional information quantum to generate a companionship information stream for patients and a visitation information stream for family members.

[0083] Calculate each emotional information quantum transmission probability This step then translates this into specific processing actions, ultimately synthesizing two distinct, carefully planned information flows: a "companionship information flow" presented to the patient. And a "visitation information stream" presented to remote family members. .

[0084] Generate patient-oriented companion information flow

[0085] For the patient, the core objective is to create a comfortable, peaceful, and positively suggestive sensory environment. The processing logic is as follows: First, the calculated quantum transmission probability With a first preset threshold set for the patient Comparison. For example, It can be set to .

[0086] like Then, regarding the quantum of emotional information Perform a "quantum decay" operation. Quantum decay is not a simple deletion, but rather employs different strategies depending on the type of information quantum: Filtering: External environmental sounds with high disturbance potential, such as the aforementioned monitor alarm sounds. (That The system uses active noise cancellation technology to generate a sound wave with the opposite phase to the alarm sound in the patient's headphones, filtering it out of the patient's auditory perception.

[0087] Replacement: While filtering out a negative quantum, to avoid the anxiety caused by an information vacuum, the system can select a quantum with high resonance affinity from the companion database for replacement. For example, at the moment an alarm sound is filtered out, the system can smoothly mix in a short, soothing musical clip or natural sound (such as the sound of flowing water) that the patient likes to fill this time window.

[0088] Suppression: For some information that cannot be completely filtered out or is not suitable for replacement (such as necessary instructions from medical staff), the system can modify its acoustic parameters, such as reducing its volume and filtering out its high-frequency components to make it less jarring, thereby reducing its wake-up rate.

[0089] like Then, regarding the quantum of emotional information A "quantum resonance transmission" operation is performed. This means that the information is suitable for the patient to perceive.

[0090] Retention: Information quanta are allowed to pass through, either as is or after basic optimization (such as volume normalization), and played on the patient's end. For example, an encouraging voice from a family member, whose calculated... If the value is high, the system will display it clearly to the patient.

[0091] Enhancement: For quantum particles with extremely high resonance affinity, the system can moderately enhance their presentation. For example, when playing music that perfectly matches the patient's preferences, a corresponding tranquil landscape image stored in the companionship database can be played, achieving a positive reinforcement of audiovisual synergy.

[0092] Ultimately, all the emotional information quanta processed by quantum decay and quantum resonance transmission operations are recombined according to their original time sequence, forming the deeply edited companionship information stream finally delivered to the patient. The information stream is presented to the patient via headphones (audio) and a bedside display screen (video).

[0093] Generate visitation information flow for family members

[0094] For family members, the goal is to ensure effective information transmission while filtering out images and sounds that may cause excessive panic and anxiety, thus achieving "protective visitation." The processing logic is similar to that for patients, but different thresholds and processing strategies are used.

[0095] The system will process each emotional information quantum transmission probability With a second preset threshold for family members The comparison is then made. This threshold is typically set more "lenient" than that for the patient because it needs to retain more information. For example, It can be set to .

[0096] In addition, the system pays special attention to those emotional information quanta marked as "sensitive" in step S200, such as... Includes Tagged quantum. For this type of sensitive information quantum, even its... It may not be low enough to trigger attenuation for the patient due to other factors (such as healthcare workers providing active care), but as long as it is marked as sensitive, the system will perform a shielding process specifically for family members.

[0097] Specific shielding operations include: Image blurring / blurring: For sensitive areas identified in the video feed (such as wounds, drainage tubes, and exposed body parts), the system will apply real-time, dynamic mosaic or Gaussian blurring to these areas when generating the visitation information stream. Family members can see the patient's overall condition and the actions of medical staff, but they cannot see details that might cause discomfort.

[0098] Sound masking: For specific sounds in the audio quantum, such as a patient's groans of pain or an overly sharp equipment alarm, even if these sounds need to be heard by medical staff, the system will selectively mute them or replace them with a softer alert tone in the audio stream transmitted to family members.

[0099] In this way, the system generates a visitation information flow that allows family members to understand the patient's condition and interact with the patient (or their virtual avatar), while also providing emotional protection for family members. This information stream is presented through family members' applications (mobile phones, computers).

[0100] To achieve the above method, this application also provides an intelligent companionship and visitation support system for intensive care. (Refer to...) Figure 2 , Figure 2 This is a schematic diagram of a system architecture disclosed in one embodiment of this application. The system may include: The information flow acquisition module 901 is responsible for executing S100. This module integrates or connects to various sensors and data interfaces, such as high-definition cameras, microphone arrays, and gateways for connecting to hospital information systems (HIS) or monitoring equipment. It is responsible for continuously and in parallel acquiring video, audio, and physiological data, and storing the accompanying data uploaded by family members into the database, forming a unified raw environmental information flow.

[0101] Information quantization module 902: responsible for executing S200. This module incorporates various signal processing and artificial intelligence models, including scene segmentation and object detection models for video streams, and sound event detection models for audio streams. It receives continuous data from the information stream acquisition module and efficiently transforms it into a structured set of emotional information quanta.

[0102] The State Awareness Module 903 is responsible for executing S300. At its core are one or more emotion state analysis models. It specifically processes patient facial image data and physiological data from the information flow acquisition module, calculating and outputting a multi-dimensional patient emotion state vector in real time through micro-expression analysis and physiological signal processing. .

[0103] The probability calculation module 904, responsible for executing S400, is the system's "decision-making brain." This module receives the quantum set output by the information quantization module and the output from the state perception module. And read the resonant cavity configuration matrix from storage. Its internal structure contains a function for calculating the perturbation potential energy. and resonance affinity The function, and the final combination to calculate the quantum transmission probability. The core nonlinear mapping model.

[0104] The information flow generation module 905 is responsible for executing S500. This module receives a quantum set with transmission probability tags and, based on a preset threshold ( , It operates on each quantum using processing rules (quantum decay, quantum resonant transmission). It includes an audio processing unit (for active noise reduction, replacement, and muting) and a video processing unit (for real-time decryption, blurring, and compositing). Ultimately, the module outputs two independent, processed information streams: and The message is then pushed to the patient's playback device and the family's application, respectively.

[0105] These modules can be independent hardware units or software functional modules running on one or more processors. They communicate and work together through predefined data interfaces to jointly realize the intelligent companionship and visitation support functions described in this application.

[0106] To further enhance the system's adaptability and forward-looking nature, this application may also include an emotional homeostasis regulation mechanism. This mechanism aims to achieve a closed-loop negative feedback control, enabling the patient's emotional state to be actively maintained within an ideal, recovery-friendly "homeostatic" range.

[0107] The implementation of this mechanism relies on an extension of the core method. First, a matrix is ​​configured within the resonant cavity. In addition to parameters such as content preferences, a "steady-state target interval" also needs to be defined. This interval is a region within the emotional state space. For example, in the Valence-Arousal two-dimensional space, It can be defined as and This represents a state of "peace and joy".

[0108] The operational process of the emotional homeostasis regulation mechanism is as follows: The state-aware module 903 outputs the patient's emotional state vector at a higher frequency (e.g., 5 times per second). The system is continuously checked. Is it located in Inside. If If the deviation falls outside this range, the system will calculate a deviation vector. ,in yes The center point.

[0109] The system is based on the deviation vector The direction and size of the resonating cavity configuration matrix are automatically adjusted. One or more parameters in the [reference]. The goal of the adjustment is to produce a [parameter] that can […]. "Pull back" The "force" of action.

[0110] If a patient is detected to have excessive arousal (e.g., The system will automatically reduce Interactive initiative parameters in This reduces the frequency of proactively pushed content. Simultaneously, it may temporarily increase sensitivity to sound energy, i.e., during calculations. At that time, the energy term is given a higher weight.

[0111] If a patient's titer is detected to be too low (e.g., The system will automatically increase the content preference vector. Playback weights for the highest-scoring content categories, and possibly a slight relaxation of the quantum resonance transmission threshold in the S500. This allows more information that may bring positive emotions to be disseminated.

[0112] Adjusted This information will be immediately used by the probability calculation module 904 to calculate the transmission probability of all subsequent emotional information quanta. This will change the system's method of filtering and presenting information, thereby affecting the patient's emotional state. New It is monitored again, forming a continuously operating closed-loop control system with a time scale of minutes, thereby realizing the "automatic cruise" of emotions.

[0113] Example of microcalculus: Initial state: Defined as , . middle .

[0114] Deviation: The state awareness module detected the deviation. It deviates significantly from the target (too low valence, too high arousal).

[0115] Regulation: The steady-state regulation mechanism is activated.

[0116] For high wakefulness ( The system will from Reduce to .

[0117] Targeting low efficiency ( The system found that the patient's most preferred content was "natural scenery". Therefore, the system selects such images and videos from the companion database and raises their playback priority to the highest level.

[0118] In this way, the system reduces unintentional disruptions while presenting highly reassuring and soothing content. After a period of time, the new... The measured as It has returned to the steady-state target range. Internally, the regulatory mechanism has temporarily entered standby mode.

[0119] This application may also include a predictive feedforward intervention mechanism. The mechanism aims to transform passive filtering into proactive intervention, by learning the relationships between events to "buffer" against potential negative emotional impacts in advance.

[0120] The core of this mechanism is to establish a "precursor-event" correlation model. This model learns by mining temporal patterns from quantum sequences of emotional information over long periods (e.g., hours or days).

[0121] Associative learning: The system records and analyzes the time series of sentiment information quanta. It attempts to find a pattern: a certain (or a group of) "precursor information quanta". The appearance of [something] in the next short time window Within a short period (e.g., 30-60 seconds), there is a high probability that it will be accompanied by a value with extremely high perturbation potential. "Target event quantum" .

[0122] Example association: The system might learn from a video quantum of a "nurse preparing a syringe". This will most likely be followed by a micro-expression quantum phenomenon: "the patient's facial muscles tense due to the injection" (which is an intrinsic event with high perturbation potential). .

[0123] Feedforward intervention: Detection: During normal system operation, once the information quantization module 902 identifies an information quantum that is labeled as a "precursor" by the associated model... This will immediately trigger the feedforward intervention mechanism.

[0124] Compensation: Once the mechanism is triggered, it does not wait for the target event. When this occurs, the system immediately and forcibly schedules one or more companions with extremely high resonance affinity from the companion database. and extremely low perturbation potential The "compensatory emotional information quantum" .

[0125] Injection: These compensatory quanta are "injected" into the soon-to-be-generated companion information stream. For example, the moment the system detects a "nurse preparing syringe" image, it immediately starts playing a pre-stored, extremely soothing piece of music that the patient likes best through their headphones, or displays a tranquil landscape image on the screen.

[0126] The purpose of this feedforward intervention is to vectorize the patient's emotional state before the negative stimulus (the pain of injection) actually arrives by introducing strong positive or neutral reassurance information. Guiding towards a more resilient state (e.g., increasing valence) Or reduce wakefulness This "buffers" or "passivates" the impending negative impact, significantly improving the intelligence and care level of the companionship system.

[0127] Those skilled in the art will understand that the above embodiments are illustrative of this application and not intended to limit it. Any modifications, equivalent substitutions, or improvements made to the above embodiments within the principles of this application should be included within the protection scope of this application.

[0128] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0129] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0130] The units described as separate components may or may not be physically separate. A component shown as a unit can be one physical unit or multiple physical units; that is, it can be located in one place or distributed in multiple different places. Depending on actual needs, some or all of the units can be selected to achieve the purpose of this embodiment.

[0131] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit described above can be implemented in hardware.

[0132] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An intelligent companion and visitation support method for intensive care, characterized by, The method comprises the following steps: acquiring an original environment information stream, which comprises real-time environment data representing an intensive care environment and companion data representing preset interaction content; decomposing the original environment information stream into a plurality of emotional information quanta, each of which is a discrete data unit carrying specific information; acquiring a patient emotional state vector representing a current physiological or emotional state of a patient; for each of the emotional information quanta, calculating a quantum transmission probability thereof based on the patient emotional state vector and a preset resonance cavity configuration matrix; based on the quantum transmission probability, performing a preset information processing operation on each of the emotional information quanta to generate a patient-oriented companion information stream and a family member-oriented visit information stream.

2. The method of claim 1, wherein, The step of calculating the quantum transmission probability of each of the emotional information quanta comprises: calculating a perturbation potential value for each of the emotional information quanta, the perturbation potential value representing a potential degree of negative influence of the emotional information quantum on the patient emotional state vector; calculating a resonance affinity for each of the emotional information quanta, the resonance affinity representing a degree of consistency of the emotional information quantum with a target companion strategy defined by the resonance cavity configuration matrix; obtaining the quantum transmission probability by combining the perturbation potential value, the resonance affinity, and the patient emotional state vector.

3. The method of claim 2, wherein, The step of obtaining the quantum transmission probability by combining the perturbation potential value, the resonance affinity, and the patient emotional state vector comprises: constructing a state adjustment function that maps the patient emotional state vector to a dynamic adjustment factor; combining the resonance affinity and the perturbation potential value to obtain a basic transmission willingness value; modulating the basic transmission willingness value using the dynamic adjustment factor to obtain the quantum transmission probability.

4. The method of claim 1, wherein, The step of performing a preset information processing operation on each of the emotional information quanta based on the quantum transmission probability comprises: comparing the quantum transmission probability with a first preset threshold value; if the quantum transmission probability is lower than the first preset threshold value, performing a quantum decay operation on the emotional information quantum to filter out or replace it when generating the companion information stream; if the quantum transmission probability is not lower than the first preset threshold value, performing a quantum resonance transmission operation on the emotional information quantum to retain or enhance it when generating the companion information stream.

5. The method according to claim 1 or 4, characterized in that, The preset information processing operation further comprises: comparing the quantum transmission probability with a second preset threshold value for family members; if the emotional information quantum is a sensitive information quantum and its quantum transmission probability is lower than the second preset threshold value, performing a shielding process on a picture or sound corresponding to the sensitive information quantum when generating the visit information stream.

6. The method of claim 1, wherein, The method further comprises: establishing an emotional homeostasis adjustment mechanism; The emotional homeostasis regulation mechanism continuously monitors the patient emotional state vector and, when it deviates from the preset homeostatic target interval, adjusts at least one parameter of the resonance cavity configuration matrix through a feedback loop to guide the patient emotional state vector back into the homeostatic target interval.

7. The method of claim 1, wherein, The method further comprises: establishing a predictive feedforward intervention mechanism; The predictive feedforward intervention mechanism identifies precursor information quanta associated with subsequent high-disturbance potential values of emotional information quanta; Upon detecting the precursor information quanta, actively introduce compensatory emotional information quanta with high resonance affinity into the accompanying information stream to intervene in the patient emotional state vector in advance.

8. The method of claim 1, wherein, The step of obtaining a patient emotional state vector comprises: acquiring facial image data of the patient through a non-contact sensor; extracting micro-expression feature sequences from the facial image data; inputting the micro-expression feature sequences into a pre-trained emotional state analysis model to obtain the patient emotional state vector.

9. An intelligent companion and visitation support system for intensive care, characterized by, Comprise: An information stream acquisition module is configured to acquire an original environment information stream, which includes real-time environment data representing an intensive care environment and accompanying data representing preset interaction content. An information quantization module is configured to decompose the original environment information stream into a plurality of emotional information quanta, each of which is a discrete data unit carrying specific information. A state perception module is configured to obtain a patient emotional state vector representing the current physiological or emotional state of the patient. A probability calculation module is configured to calculate the quantum transmission probability of each emotional information quantum based on the patient emotional state vector and a preset resonance cavity configuration matrix. An information stream generation module is configured to perform a preset information processing operation on each emotional information quantum based on the quantum transmission probability to generate an accompanying information stream for the patient and a visit information stream for the family members.

10. The system of claim 9, wherein, The probability calculation module is configured to: calculate a disturbance potential value for each emotional information quantum, which represents the potential degree of negative impact of the emotional information quantum on the patient emotional state vector; calculate a resonance affinity for each emotional information quantum, which represents the consistency degree of the emotional information quantum with the target accompanying strategy defined by the resonance cavity configuration matrix; obtain the quantum transmission probability by combining the disturbance potential value, the resonance affinity, and the patient emotional state vector.

Images