A delirium risk monitoring device and system

Abstract

This invention relates to a delirium risk monitoring device, which implements a delirium risk monitoring method. The delirium risk monitoring method includes: after the subject undergoing at least one rapid assessment for delirium and confusion, a delirium factor processing module retrieves dominant and latent factors related to the subject from a medical information management system, and generates tags required by the delirium risk monitoring module based on the attributes of the dominant factors and / or the attributes of the latent factors; based on the generated tags, the delirium risk monitoring module obtains multiple case information groups matching the subject from the cloud platform through information interaction with the cloud platform; and the delirium risk monitoring module calculates the delirium risk prediction of the subject using a delirium dynamic prediction model based on the obtained multiple case information groups.

Description

[0001] The original basis for this divisional application is patent application CN202010417659.2, filed on May 15, 2020, entitled "A method and system for monitoring spectrum delusion risk based on a dynamic prediction model of delusion". Technical Field

[0002] This invention relates to the field of delirium care technology, and in particular to a delirium risk monitoring device and system. Background Technology

[0003] Delirium is an acute cognitive impairment syndrome characterized by attention deficit, confusion, altered consciousness, and impaired level of consciousness. It is an acute or subacute onset disorder, typically progressing within hours to days. Identification requires brief cognitive screening and keen clinical observation. Key diagnostic features include acute onset and fluctuating altered level of consciousness, inattention, impaired level of consciousness, and cognitive impairment (e.g., disorientation, memory impairment). Clinically, delirium occurring in patients in the Intensive Care Unit (ICU) is often referred to as ICU delirium. Literature reports that 14%–24% of hospitalized patients develop delirium during hospitalization, with the incidence rate reaching 30%–50% in elderly patients and 35%–80% in ICU patients.

[0004] Although the incidence of delirium is not to be underestimated, it has long been mistakenly considered a minor "ICU psychosis" in ICU patients, a neglect that can lead to various adverse consequences. Delirium is a complex neurological syndrome associated with multiple adverse outcomes, such as increased medical costs, prolonged hospital stays, cognitive impairment, decreased independence, increased complications, decreased cumulative survival, prolonged postoperative recovery time, and increased postoperative mortality. Studies by Trogrlic Z et al. have found that delirium complicates patient treatment and increases the difficulty of care, and may also lead to permanent irreversible brain damage. Prolonged delirium can cause potential organ dysfunction, with an incidence rate of approximately 70%–92%, a 10-fold increased risk of aspiration and iatrogenic pneumonia, a significantly increased incidence of complications such as pulmonary embolism and pressure sores, and difficulties in weaning mechanically ventilated patients, leading to accidental extubation or re-intubation after extubation, ultimately prolonging ICU stays and increasing mortality. Studies show that patients with delirium have an average hospital stay that is 8 days longer than those without delirium. Once delirium occurs, it prolongs the time patients need mechanical ventilation, the time spent in the ICU, and the length of hospital stay. According to related research, 75.7% of patients with delirium still have cognitive impairment at discharge. Delirium in the elderly often indicates a poor prognosis, including overall functional impairment, inability to live independently, and extreme need for care. Related studies show that compared with patients without delirium, patients with postoperative delirium also have an increased incidence of pulmonary complications and a higher probability of receiving further postoperative care. The occurrence of ICU delirium can increase the risk of iatrogenic pneumonia by 10 times and can also lead to accidental extubation, re-intubation, and difficulty weaning patients on mechanical ventilation.

[0005] However, delirium is easily misdiagnosed or missed because its early symptoms are atypical and its onset is insidious. At the same time, non-specialist psychiatrists often lack knowledge and understanding of the disease, and therefore it is usually not given timely attention and treatment.

[0006] Prior art, such as the system and method disclosed in patent document CN109069081A for predicting, screening, and monitoring encephalopathy / delirium, detects the presence of diffuse slowing in a patient's brain waves (a hallmark of encephalopathy attacks). This system and method can detect diffuse slowing by performing spectral density analysis on brain waves recorded at a small number of discrete locations on the patient's head, thus enabling relatively easy bedside assessment, for example, using a handheld device. That is, the system and method can record brain waves using two or more leads placed on the patient's head, execute algorithms to evaluate the ratio of recorded low-frequency waves to high-frequency waves, and compare this ratio to a defined threshold to identify encephalopathy attacks. In further embodiments, the system and method utilize machine learning and additional data, such as from medical records, to improve assessment accuracy.

[0007] According to the current version of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) developed by the American Psychiatric Association and generally considered the gold standard for diagnosing delirium, the diagnosis of delirium requires meeting the following criteria: A) Impaired consciousness accompanied by a decline in attention, persistence, or diversion; B) Cognitive changes (including memory impairment, disorientation, and language impairment), or perceptual disturbances that cannot be explained by dementia; C) Onset of symptoms within a short period (usually from several hours to several days), with fluctuations throughout the day.

[0008] The aforementioned patent document utilizes a dozen physiological or brain sensors placed on the patient to continuously monitor the patient's physiological information, including at least brain signals, and outputs an indication of the presence, absence, or likelihood of delirium. However, in practice, on the one hand, electroencephalography (EEG) is mainly for early monitoring of atypical and difficult-to-detect abnormal brain discharges, and cannot provide the patient information required for the DSM-IV-TR diagnostic criteria. That is, delirium assessment cannot be determined solely through continuous EEG monitoring, easily missing the opportunity to take timely preventive measures. On the other hand, one of the main purposes of screening for postoperative delirium is to take timely preventive measures and avoid high delirium treatment costs. The current assessment scheme using expensive EEG-assisted continuous monitoring actually increases the patient's treatment costs.

[0009] Besides using electroencephalography (EEG) as an auxiliary examination for the assessment and differentiation of delirium, clinical practice often employs delirium scales derived from the DSM-IV-TR diagnostic criteria to assess the severity of delirium in the ICU, provide prognostic assessments for delirium patients, and serve as a basis for treatment. Currently, the main scales used are as follows: Confusion Assessment Method (CAM), Memory Delirium Assessment Scale (MDAS), Delirium Rating Scale (DRS), Delirium Rating Scale-98 (DRS-R-98), Cognitive Test for Delirium (CTD), and 3-Minute Diagnostic Interview for CAM-Defined Delirium (3D-CAM). The aforementioned patent document also mentions a solution for continuing patient assessment using one of the commonly used scales when EEG cannot detect delirium. However, it failed to consider that although the delirium scale itself incorporates objective cognitive testing and assessment into delirium assessment, it places high demands on the delirium assessment ability of nursing staff. Based on the solution proposed in the aforementioned patent document, which is based on the nursing staff's own understanding and assessment of EEG and scale, it is difficult for medical staff to achieve reliable and effective delirium assessment. Summary of the Invention

[0010] Currently, the field of delirium nursing technology faces several challenges. For example, delirium scales place high demands on nurses' delirium assessment abilities, making reliable and effective delirium risk prediction difficult if based solely on the nurses' understanding of the scales. Existing technologies propose clustering and extracting similar medical data from big data and using this data for risk prediction. However, these solutions cluster and extract multiple medical data related to the disease itself, failing to consider the significant impact of individual differences on delirium risk. In other words, a single piece of medical data cannot reflect the current state and response of the patient. Furthermore, these solutions extract data from patients with confirmed illnesses, resulting in extremely low reliability of risk predictions based on post-diagnosis patient data. Therefore, existing solutions are unsuitable for delirium risk prediction, especially for patients who have not yet developed delirium.

[0011] In response to this, the solution provided by this invention for risk prediction using similar medical condition information from medical big data utilizes, on the one hand, the objective cognitive testing and assessment characteristics of the delirium scale itself. The assessment data, closely related to individual patient differences, obtained through the assessment process, is used as a latent factor for clustering and capturing big data. This achieves a high degree of matching in data capture while fully satisfying individual patient differences. On the other hand, considering the problem of low accuracy in risk prediction results due to the superposition and cancellation of multiple medical data, the solution proposed in this invention analyzes and processes big data from two different levels: overall trend and local trend. This further improves the accuracy and reliability of risk prediction, and is particularly suitable for assessing subjects who have not yet developed delirium or for delirium risk prediction with a low potential risk.

[0012] To address the shortcomings of existing technologies, this application proposes a delirium risk monitoring device based on a delirium dynamic prediction model. The delirium risk monitoring device includes at least: a delirium factor processing module, used to retrieve dominant and latent factors related to the subject after the subject has completed at least one rapid assessment for delirium confusion, and generate tags required by the delirium risk monitoring module based on the attributes of the dominant factors and / or the attributes of the latent factors; and a delirium risk monitoring module, used to obtain multiple case information groups matching the subject in the cloud platform based on the generated tags through information interaction with a cloud platform. The delirium risk monitoring module calculates the delirium risk prediction of the subject based on the obtained multiple case information groups using the delirium dynamic prediction model.

[0013] For predicting the risk of delirium onset or deterioration, domestic and international scholars mainly use factors such as the patient's own disease, treatment, and environment (which can be regarded as dominant factors) as indicators to construct a risk system for delirium risk prediction. However, the above factors are not only numerous, but the correlation between them is also unclear. That is, although many factors may have a potential impact on the induction of delirium, they include factors that are unrelated to the occurrence or induction of delirium, weakly correlated, or redundant. These factors not only do not contribute significantly to the prediction results, but also increase the amount of data calculation and reduce the prediction efficiency.

[0014] In response, the delirium risk monitoring method proposed in this application utilizes the objective cognitive testing and assessment characteristics of the delirium scale itself. It uses assessment data closely related to individual patient differences obtained through the assessment process as latent factors for clustering and capturing large datasets. On one hand, the number of latent factors is far lower than the number of dominant factors, resulting in less computational data and improved data processing efficiency. On the other hand, the delirium risk monitoring method proposed in this application, based on the partial case information group already selected based on latent factors, further utilizes dominant factors that may have a potential impact on delirium induction to continue screening the case information group, thus accurately identifying case samples for risk monitoring. This allows delirium risk monitoring to fully satisfy individual patient differences while achieving high matching degree and high processing efficiency in data capture.

[0015] More preferably, the delirium risk monitoring module calculates the delirium risk prediction of the subject to be assessed using a delirium dynamic prediction model based on multiple historical information groups and the overall and / or local change trends corresponding to each historical information group. Especially for individual patients, if current mainstream big data analysis methods are used, the data collected typically includes factors, delirium status, and the correlation between factors and delirium status at the current time point. However, delirium, as a time-series disease, is susceptible to induction at subsequent time points due to changes in relevant factors at a certain time point (or simply understood as the implementation of treatment measures). This time-series disease has a certain lag and significant randomness, and due to the influence of other poorly controllable factors (such as sudden pain caused by the patient's own disease), there is a possibility of drastic fluctuations in the short term.

[0016] To address this, the delirium risk monitoring method proposed in this application analyzes big data from two different perspectives: the overall trend of latent factors and the local trend of latent factors. This method is particularly suitable for the lagging characteristics of delirium as a time-series illness. By analyzing the trends, the impact of changes in relevant factors at a certain point in time on the induction of delirium at subsequent points in time can be determined. However, the big data processing based on overall trends is relatively coarse. For samples where the delirium trend fluctuates repeatedly due to factors with poor controllability, the overall trend cannot reflect the true delirium trend of the sample. Therefore, this application combines overall and local trends, further improving the accuracy and reliability of risk prediction. This method is particularly suitable for assessing subjects who have not yet developed delirium or for delirium risk prediction with low potential risk.

[0017] According to a preferred embodiment, after obtaining the tag, the delirium risk monitoring module first compares the similarity of the latent factor tag of the current subject to be evaluated with the latent factor tags of multiple case information in the cloud platform based on a pre-set similarity interval, thereby determining multiple case information in the cloud platform that meet the similarity interval and are used to form a case information group.

[0018] According to a preferred embodiment, when the number of case information in at least one of the multiple case information groups does not reach the sample size threshold, the delirium risk monitoring module corrects the similarity interval by selectively expanding the interval range, thereby maximizing the matching degree between the case information group and the subject to be evaluated while meeting the sample size required for delirium risk prediction of the subject to be evaluated.

[0019] According to a preferred embodiment, obtaining multiple case information groups refers to, after determining multiple case information that meet the similarity range in the cloud platform, the delirium risk monitoring module filters out multiple case information that meet the same label as the label of the object to be evaluated based on the labels of the multiple case information and the label of the object to be evaluated, and forms a case information group.

[0020] According to a preferred embodiment, the dominant factors include at least one major risk factor for delirium and at least one minor risk factor for delirium, and the latent factors include at least the delirium assessment data determined during the rapid assessment of delirium confusion in the subject to be assessed, which includes at least one or more of the trend of delirium potential risk level changes and delirium characteristic k.

[0021] According to a preferred embodiment, the delirium risk monitoring device includes a delirium assessment module, which is configured to: acquire feedback information about the subject to be assessed and / or about the assistant, acquire behavioral information of the subject to be assessed during a rapid assessment of delirium confusion, and / or generate parameters required for a delirium assessment model based on the attributes of the behavioral information and the attributes of the feedback information, and / or calculate using the delirium assessment model based on the generated parameters to obtain an assessment value for at least one delirium feature obtained by the subject to be assessed during a rapid assessment of delirium confusion.

[0022] According to a preferred embodiment, the delirium risk monitoring device further includes: an audio-visual processing module, which is used to acquire behavioral information and / or feedback information about the subject under assessment by video acquisition of the behavior of the subject under assessment who has been diagnosed with delirium or has the potential risk of delirium during a rapid assessment of delirium confusion; and / or an external input device, which is operated by the subject under assessment and is used to acquire feedback information input by the subject under assessment for the assessment content, and to detect the subject under assessment's autonomous operation during the assessment process to acquire behavioral information and / or feedback information about the subject under assessment.

[0023] This application also proposes a delirium risk monitoring system based on a delirium dynamic prediction model. The risk monitoring system includes at least: a memory; at least one computer processor coupled to the memory; a first computer processor, configured to, after a first user completes at least one assessment, retrieve dominant and latent factors related to the first user from a database, and generate tags required by a second computer processor based on the attributes of the dominant factors and / or the attributes of the latent factors; and a second computer processor, configured to, based on the generated tags, obtain multiple historical information groups matching the first user by interacting with the database, wherein the second computer processor calculates the risk prediction of the first user using a dynamic prediction model based on the obtained multiple historical information groups and the overall and / or local change trends corresponding to each historical information group.

[0024] According to a preferred embodiment, the second computer processor is further configured to, after acquiring the tag, first compare the similarity of the tag of the latent factor of the current first user with the tag of the latent factor of multiple historical information in the database based on a pre-set similarity interval, thereby determining multiple historical information in the database that meet the similarity interval and are used to form a historical information group.

[0025] According to a preferred embodiment, the second computer processor is further configured to correct the similarity interval by selectively expanding the interval range when the number of historical information in at least one of the multiple historical information groups does not reach the sample number threshold, thereby maximizing the matching degree between the historical information group and the first user while meeting the sample number required for risk prediction of the first user.

[0026] The processing modules mentioned in this invention can be described as being "configured to" to perform one or more functions. Generally, an element configured to perform or configured to perform a function is capable of performing that function, suitable for performing that function, operable to perform that function, or otherwise performs that function. It should be understood that "at least one of X, Y, Z" and "one or more of X, Y, Z" can be understood as only X, only Y, only Z, or any combination of two or more of X, Y, Z (e.g., XYZ, XY, YZ, XZ, etc.). Similar logic can be applied to any two or more objects appearing in the statements "at least one..." and "one or more...". As used in this specification, the singular forms of "a" or "the" include the plural referent unless the content and context explicitly indicate otherwise. That is, for example, a reference to "device" includes a combination of two or more such devices. Unless otherwise specified, the "or" connector is intended to be used in its proper meaning as a Boolean logic operator, including both alternative feature selection (A or B) and conjunction feature selection (A or B). The intelligent electronic devices include, but are not limited to, various terminal devices such as computers, mobile phones, and tablets.

[0027] The apparatus proposed in this invention includes at least one processing module, a system storage device, and at least one computer-readable storage medium. The at least one computer-readable storage medium carries computer-executable instructions for causing a processor to implement various aspects of the invention. Figure 2 For example, multiple processors and interfaces are interconnected via a communication bus (solid line) such as a motherboard (system storage devices are not shown). Interfaces include at least communication interfaces and I / O interfaces. Each module is operatively coupled to a computer network via a communication interface (such as a network adapter). The computer network can be the Internet, the Internet of Things, and / or an extranet, or an intranet and / or extranet communicating with the Internet. Each module communicates with the intelligent electronic device via the computer network or via a direct (e.g., wired, wireless) connection.

[0028] At least one processing module, such as the first processing module, is used to execute the computer-executable instructions. The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Each block in the flowchart or block diagram may represent a module, program segment, or part of an instruction, which includes one or more executable instructions for implementing a specified logical function. Each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions. Various aspects of the present invention are described herein with reference to flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present invention. It should be understood that each block in the flowchart and / or block diagram, and combinations of blocks in the flowchart and / or block diagram, can be implemented by computer-readable program instructions. A processing module is a functional unit that interprets and executes instructions, also known as a central processing unit or CPU, which, as the core of the computer system's computation and control, is the final execution unit for information processing and program execution.

[0029] The aforementioned computer-readable storage medium can be a tangible device capable of holding and storing instructions used by an instruction execution device. Computer-readable storage media can be, for example, but not limited to, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination thereof. The computer-executable instructions described herein can be downloaded from the computer-readable storage medium to various computing / processing modules, or downloaded via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network, to an external computer or external storage device. The network can include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing module receives computer-readable program instructions from the network and forwards the computer-executable instructions for storage in the computer-readable storage medium of the respective computing / processing module.

[0030] The computer program instructions used to perform the operations of this invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as C or similar languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of the invention. Attached Figure Description

[0031] Figure 1 This is a flowchart of a preferred method for monitoring delirium risk provided by the present invention; and

[0032] Figure 2 This is a simplified module connection diagram of the preferred delirium risk monitoring system provided by the present invention.

[0033] List of reference numerals

[0034] 101: Mobile electronic device; 102: Delirium factor processing module; 103: Delirium assessment module; 1011: Audio-visual processing module; 1012: External input device; 104: Assessment and processing module; 105: Delirium risk monitoring module; 106: Cloud platform; 107: Medical information management system. Detailed Implementation

[0035] The following is a detailed explanation with reference to the accompanying drawings.

[0036] The present invention will now be described in detail with reference to the accompanying drawings.

[0037] To address the shortcomings of existing technologies, such as the high demands placed on nurses' delirium assessment abilities by delirium scales, relying solely on nurses' understanding of the scale makes reliable and effective delirium risk prediction difficult. Existing technologies propose solutions that cluster and extract similar medical condition information from big data medical data and use the extracted data for risk prediction. However, because these solutions cluster and extract multiple medical data related to the disease itself, they fail to consider the significant impact of individual differences in the individual being assessed on delirium risk. In other words, a single piece of medical data cannot reflect the current state and response of the individual being assessed. Furthermore, the medical data extracted by these solutions are all from patients whose diseases are already confirmed, resulting in extremely low reliability of risk predictions based on the medical data of diagnosed patients. Therefore, the solutions proposed in existing technologies are not suitable for delirium risk prediction, especially for individuals being assessed who have not yet developed delirium.

[0038] The solution for risk prediction using similar medical condition information provided by this invention utilizes the objective cognitive testing and assessment characteristics of the delirium scale itself. It uses assessment data closely related to individual patient differences obtained through the assessment process as implicit factors for clustering and capturing big data. This achieves high matching degree in data capture while fully satisfying individual patient differences. Furthermore, considering the problem of low accuracy in risk prediction results due to the superposition and cancellation of multiple medical data, the solution proposed in this invention analyzes and processes big data from two different levels: overall trend and local trend. This further improves the accuracy and reliability of risk prediction, and is particularly suitable for assessing subjects who have not yet developed delirium or for delirium risk prediction with a low potential risk.

[0039] like Figure 1 As shown, this invention proposes a delirium risk monitoring method based on a dynamic delirium prediction model. This delirium risk monitoring method is particularly suitable for assessment subjects that have not yet developed delirium or have a low potential delirium risk level. Figure 1 As shown, this delirium risk monitoring method mainly includes:

[0040] Preprocessing steps (not shown in the figure): After obtaining the patient information of the subject to be evaluated, the system interacts with the medical information management system 107 to retrieve the patient's medical history and transmit it to the preprocessing module. The preprocessing module then makes a preliminary judgment on the subject's language expression ability, physical activity ability, and facial expression ability.

[0041] S1: At least one mobile electronic device 101 acquires feedback information about the subject to be assessed and / or about the assistant, and acquires behavioral information of the subject to be assessed during a rapid assessment of delirium and confusion.

[0042] S2: The delirium factor processing module 102 generates the parameters required for the delirium assessment model based on the attributes of the behavioral information and the attributes of the feedback information.

[0043] S3: The delirium assessment module 103 calculates the value of at least one delirium feature obtained by the delirium assessment model based on the generated parameters.

[0044] S4: After the delirium assessment module 103 obtains the assessment value of at least one delirium feature obtained from the rapid assessment of delirium confusion of the subject to be assessed, the assessment processing module 104 combines the above assessment values ​​of at least one delirium feature and determines whether the combination meets the predetermined assessment conditions.

[0045] S5: The delirium factor processing module 102 retrieves the dominant and latent factors related to the subject to be evaluated from the medical information management system 107, and generates the tags required by the delirium risk monitoring module 105 based on the attributes of the dominant factors and / or the attributes of the latent factors.

[0046] S6: Based on the generated labels, the delirium risk monitoring module 105 obtains multiple case information groups in the cloud platform 106 that match the subject to be evaluated by interacting with the cloud platform 106.

[0047] S7: The delirium risk monitoring module 105 calculates the delirium risk prediction of the subject to be assessed based on the multiple sets of case information obtained using the delirium dynamic prediction model.

[0048] The following is a step-by-step explanation of S5~S7:

[0049] For step S5: the delirium factor processing module 102 retrieves the dominant and latent factors related to the subject to be evaluated from the medical information management system 107, and generates the tags required by the delirium risk monitoring module 105 based on the attributes of the dominant factors and / or the attributes of the latent factors.

[0050] "Dominant factors" refer to the primary and secondary risk factors for delirium. Primary risk factors for delirium include at least the APACHE-II score, history of chronic illness, sleep disturbances, use of sedatives or anesthetics, infection, indwelling urinary catheter, and hearing loss. Secondary risk factors for delirium include those targeted by the ABCDEF bundle of interventions: pain, mechanical ventilation, use of sedatives or analgesics, limited mobility, and lack of family accompaniment. For example, information such as the subject's age (D1), sex (D2), body mass index (D3), education level (D4), history of alcoholism or smoking (D5), underlying medical conditions (D6), type of illness admitted to the hospital (D7), and use of analgesics (D8) are also considered (Dm).

[0051] Dominant factors refer to information that can be determined without assessment, such as the personal information, medical history, or medication information of the subject to be assessed. "Attributes of dominant factors" refer to situations that differ from the actual circumstances of the subject to be assessed, such as whether the subject may or may not possess the dominant factor, or whether the dominant factor falls within a defined range. Based on this attribute, the corresponding label Dmn can be determined. Since the values ​​corresponding to these attributes are all non-numerical, the delirium factor processing module 102 in step S5 needs to generate the labels required by the delirium risk monitoring module 105 based on the attributes of the dominant factors. Taking "age D1" and "history of alcoholism or smoking D5" as examples, there are four age ranges: {10~30, 31~50, 51~70, 71~100}. Each of these ranges corresponds to a numerical identifier from 1 to 4. If the subject to be assessed is 52 years old, falling within the {51~70} range, then the label for the subject to be assessed must include at least D13. For whether or not a person has a history of alcoholism or smoking, there is a corresponding number 1 and 0, respectively. If the person being evaluated has a history of either alcoholism or smoking, then the label of the person being evaluated shall include at least D51.

[0052] "Latent factors" refer to the delirium assessment data of the subject to be assessed. Latent factors are information determined through assessment. Delirium assessment data may include the trend of delirium potential risk level changes δQ. Delirium assessment data may also include information such as combinations Σ, k1, k2, k3, and k4. Among these, "trend of delirium potential risk level changes δQ" refers to the information on the change of the delirium potential risk level over time based on each assessment of the subject. Since the values ​​corresponding to its attributes are all numerical, the attributes of these latent factors are the labels required to generate the delirium risk monitoring module 105. For example, if the delirium potential risk levels Q obtained from four assessments of the subject are 0, 1, 2, and 1 respectively, then the labels for the subject must at least include ΩQ{0, 1, 2, 1}.

[0053] Regarding step S6: Based on the generated labels, the delirium risk monitoring module 105 obtains multiple case information groups in the cloud platform 106 that match the subject to be evaluated by interacting with the cloud platform 106.

[0054] "Cloud platform 106" can be a third-party service database storing a large amount of case information. Each case information includes a dominant factor label Hmn and a recessive factor label Φ. Preferably, the delirium risk monitoring module 105, through information interaction with cloud platform 106, obtains multiple groups of case information in cloud platform 106 that match the subject to be evaluated, based on satisfying preset similarity screening conditions. The "preset similarity screening conditions" are used to screen a subset of case information from the large amount of case information in cloud platform 106 that matches a preset similarity. More specifically, step S6 also includes one or more of the following steps:

[0055] S61: Based on the similarity range of 90%~100%, the label Ω of the latent factor of the current object to be evaluated is compared with the label Φ of the latent factor of multiple case information in the cloud platform 106 to determine X case information in the cloud platform 106 that meet the similarity range.

[0056] The label Ω for the latent factor used in similarity comparison can be one or more time-related trend information of Σ, k1, k2, k3, k4, and the potential risk level of delirium Q obtained from each assessment of the subject under evaluation. Since the "evaluation value" of delirium features k1, k2, k3, and k4 only includes two results: negative and positive (negative results can be represented by "-" and positive results by "+"), the evaluation value of its latent factor is the label required to generate the delirium risk monitoring module 105. For example, the delirium feature k1 obtained from four assessments of the subject under evaluation is -, +, -, - respectively, meaning that the label of the subject under evaluation must include at least Ωk1{-, +, -, -}. In the early stages of the assessment, since the subject under evaluation may only have undergone one or two rapid assessments of delirium confusion, the label Ω will compare a large number of case information. Therefore, "comparing the similarity of the latent factor label Ω of the current subject to be evaluated with the latent factor label Φ of multiple case information in the cloud platform 106" is performed on the premise that the precondition is met. The precondition is that the current subject to be evaluated has undergone at least λ evaluations λ∈{1,2,3,4,5}, that is, the label Ω must contain at least λ values. In the initial stage of evaluation, if the current subject to be evaluated has undergone at least λ evaluations, there is insufficient reference data for the subject to be evaluated. Preferably, based on a similarity range of 90%~100%, the dominant factor label Dmn of the current subject to be evaluated is compared with the dominant factor label Hmn of multiple case information in the cloud platform 106 to determine X case information in the cloud platform 106 that meet the similarity range.

[0057] S62: Based on the labels Hmn of each of the X case information and the label Dmn of the object to be evaluated, select multiple case information that satisfy the same Hmn and Dmn from the X case information and form a case information group;

[0058] S63: Based on a preset sample number threshold, determine whether the number of case information in each of the multiple case information groups meets the sample number threshold;

[0059] S64: When the number of case information in each of the multiple case information groups meets the sample number threshold, the delirium risk monitoring module 105 retrieves the multiple case information groups.

[0060] S65: When the number of case information in at least one of the multiple case information groups does not reach the sample number threshold, the similarity interval is corrected by expanding the interval range. The corrected similarity interval is substituted into step S61, and S61~S63 are repeated until step S64 is met, at which point the delirium risk monitoring module 105 retrieves the multiple case information groups.

[0061] Here, Dmn refers to the label of a dominant factor of the subject being evaluated. For example, D51 indicates that the subject has a history of alcoholism or smoking. For a given m value, its corresponding n value is unique. Therefore, based on multiple Dmn values, multiple case information groups can be formed. Each case information group corresponds to one label Dmn, and the labels Dmn for each of the multiple case information groups are different. "Multiple case information groups" are actually multiple combinations classified according to different labels Dmn. All case information in a single case information group has a label Hmn that is the same as Dmn.

[0062] The "similarity comparison" method primarily involves comparing two aspects: the overall trend and the local trend. The overall trend refers to the trend of the final determined potential risk level Q of delirium after assessment, while the local trend refers to the trend of the assessed values ​​of delirium features k1, k2, k3, and k4 determined during the assessment process. If the overall trend conforms to the similarity interval, the local trends are then compared to see if they also conform to the similarity interval. "Modifying the similarity interval by expanding the range" refers to a selective correction method. Specifically, when the number of case information in at least one case information group among multiple case information groups does not reach the sample size threshold, the similarity interval requirement for the overall trend remains unchanged, while the similarity interval requirement for the local trend is reduced. This maximizes the number of case information groups with high similarity as risk prediction samples for the delirium risk monitoring module 105.

[0063] Existing technologies typically employ a specified screening range to select data groups from the database that meet the specified screening range. Then, each sample within a data group is compared one by one with the test sample to obtain the similarity between the two samples. This increases the unnecessary amount of data processing. Furthermore, some existing technologies stop screening once a sufficient number of samples have been obtained, leading to low confidence intervals for the selected data groups. Other existing technologies screen all samples in the database before selecting the required number of samples, which drastically increases CPU load and reduces data processing efficiency. To address this, the delirium risk monitoring system proposed in this application employs a hierarchical screening structure to obtain the required number of case information groups. It selects the case information groups with the highest similarity from the database by satisfying the specified screening range of maximum similarity. After the initial screening, the number of samples within each group is counted. If the required number is not reached, the database is further screened by reducing the specified screening range. The delirium risk monitoring method using a hierarchical screening structure in this application completes the screening process and the similarity comparison process simultaneously, reducing unnecessary data processing and enabling the acquisition of samples with high confidence intervals without having to completely screen all data.

[0064] The above screening process first filters out case information that meets the delirium assessment data of the current subject to be assessed. Then, the number of cases initially screened is counted. If the minimum sample size threshold is not reached, the screening range is appropriately expanded and a second screening is conducted. This ensures that the number of samples obtained is sufficient to support delirium risk prediction and also guarantees the validity of the obtained data, thereby improving the accuracy of delirium risk prediction.

[0065] Regarding step S7: The delirium risk monitoring module 105 calculates the delirium risk prediction of the subject to be assessed based on the multiple case information groups obtained and using the delirium dynamic prediction model.

[0066] To facilitate understanding, the following explanation addresses the labels Φ for multiple case information entries: Since the labels Φ for the selected case information entries all meet the similarity screening criteria with the label Ω, meaning that the trend of partial delirium potential risk level changes in label Φ matches the trend of delirium potential risk level changes in label Ω, the subsequent delirium potential risk level changes in label Φ are considered trend values. In other words, each of the selected case information entries corresponds to a trend value, which is used to provide calculation data for the delirium risk prediction model of the subject under evaluation. "Subsequent delirium potential risk level changes" refers to the trend of partial delirium potential risk level changes within a preset time range. The preset time range can be one month or two months.

[0067] For example, if the trend of delirium potential risk level changes within a label Φ over a month is 1, 0, 1, and its average trend remains flat with no higher risk levels, then its trend value is 0, indicating a low delirium risk and a relatively stable state. Conversely, if at least one of the following conditions exists—an increasing average trend or the appearance of a higher risk level—then its trend value is 1, indicating an increased delirium risk. If the average trend decreases and there are no higher risk levels, then its trend value is -1, indicating a decreased delirium risk and a relatively stable state.

[0068] Similarly, since the labels Φ of the selected case information all meet the similarity screening criteria with the label Ω, that is, the changing trends of the evaluation values ​​of delirium features k1, k2, k3, and k4 in label Φ meet the similarity screening criteria with the changing trends of some evaluation values ​​of delirium features in label Ω. The changing trends of the evaluation values ​​of delirium features in label Φ that follow this partial changing trend are considered as trend values. Based on different delirium evaluation data, trend values ​​for several different labels can be determined.

[0069] Regarding the "delirium dynamic prediction model" in step S7, this model pre-stores several calculation formulas. For example, it includes a delirium risk prediction calculation formula. Therefore, more specifically, step S7 includes at least one or more of the following steps:

[0070] S71: Obtain multiple case information groups and determine the proportion of each case information group based on the number of case information in each group;

[0071] S72: Determine the trend value of each case information based on the label Φ that meets the similarity screening condition between each case information and the label Ω;

[0072] S73: Statistically analyze the trend values ​​of each case information group to generate the trend values ​​of each case information group;

[0073] S74: Based on the proportion and trend value of each case information group, determine and output the delirium risk prediction for the current subject to be evaluated within a preset time range.

[0074] The "trend value" includes at least the trend value corresponding to the overall trend and the trend value corresponding to the local trend. The calculation is preferentially based on the trend value corresponding to the overall trend, i.e., based on the trend of the delirium potential risk level Q. Since the calculation involves comparing proportions, there may be small differences between the proportions. In this case, the true trend of the sample cannot be reflected, leading to a calculation result that is heavily biased towards either a worsening or improving delirium risk prediction. Therefore, for step S74, more preferably: if the differences between the proportions corresponding to the calculated trend values ​​of 1, 0, and -1 are small, making it impossible to determine the delirium risk prediction for the current assessment object within a preset time range, then the calculation is performed based on the trend value corresponding to the local trend, i.e., the trend of the evaluation values ​​of delirium features k1, k2, k3, and k4. This allows at least one delirium potential risk level Q to be determined, and based on this delirium potential risk level Q, the delirium risk prediction for the current assessment object in the near term is determined and output. Since the local change trend / delusion characteristic assessment value change trend only includes two results, negative and positive, and the assessment value combination Σ for delusion characteristics can obtain at least one corresponding delusion potential risk level Q, by comparing it with the current delusion potential risk level of the subject to be assessed, the delusion risk prediction of the subject to be assessed within a preset time range can be determined—whether the delusion potential risk level increases, remains the same, or decreases.

[0075] Regarding the "proportion of each case information group," further explanation is needed: First, before risk prediction, the weights of each dominant factor are pre-set. As mentioned earlier, dominant factors differ from primary and secondary risk factors; correspondingly, different dominant factors have different weights in influencing delirium risk. Support staff can modify the pre-set weights of multiple dominant factors based on actual circumstances. Since each case information group corresponds to a unique label Dmn, it also corresponds to only a unique weight value. By statistically analyzing the number of cases within a single case information group, the proportions of each case information group can be obtained. These proportions are then further optimized based on the weight values ​​of individual case information groups; the optimized result is the proportion of each case information group. For example, multiplying the two yields the proportion of each case information group.

[0076] The following is a detailed explanation of S1 to S4 step by step:

[0077] For step S1: At least one mobile electronic device 101 acquires feedback information about the subject to be assessed and / or about the assistant, and acquires behavioral information of the subject to be assessed during the rapid assessment of delirium and confusion.

[0078] The "feedback information about the subject to be assessed and / or the assistant" in step S1 above refers to the selections made by the subject to be assessed or the assistant on several items of the delirium confusion rapid assessment scale. This feedback information may include four pieces of information: yes, no, incorrect, and correct.

[0079] Regarding the "feedback information about the support staff" in step S1, this feedback information is the support staff's response to the relevant items. The method of collecting this response can be determined by the response input by the support staff into the mobile electronic device 101. For example, regarding one of the scale items, "During the assessment, was the patient drowsy, lethargic, or comatose?", the support staff, based on their own judgment of the assessment subject's condition, inputs feedback information of yes or no into the mobile electronic device 101.

[0080] Regarding the "feedback information about the subject to be assessed" in step S1, this feedback information is the subject's answer to the relevant items. This answer can be collected by inputting the response from an assistant into the mobile electronic device 101. Preferably, the "feedback information about the subject to be assessed" can also be an answer input by the subject themselves. For example, if an assistant asks the subject, "Have you felt confused today?", the subject can use a handheld remote control to input "yes" or "no" into the mobile electronic device 101. The above preferred embodiments are primarily designed for patients who cannot speak, such as those on mechanical ventilation or with central venous catheterization. In such cases, delirium assessment cannot be achieved through the patient's verbal description. The intelligent assessment system provided by this invention provides the subject with a handheld remote control, allowing them to answer the items displayed on the screen by sliding or pressing the buttons on the remote control.

[0081] Preferably, regarding the "feedback information about the subject to be evaluated" in step S1, this feedback information is the subject's answer to the relevant item, and the method of collecting this answer can be determined by the mobile electronic device 101. For example, an assistant asks the subject to be evaluated one of the scale items, "What year is it this year?" The subject may answer correctly or incorrectly. The mobile electronic device 101 analyzes and processes the subject's answer based on the video collected by its audio-visual processing module 1011 to determine the answer under that item. By using the above-described video acquisition method for assisted evaluation, the present invention can obtain accurate answers through video analysis and processing, and cross-verify these answers with the answers input by the assistant to the mobile electronic device 101, eliminating the problem of incorrect answers input by the assistant due to errors.

[0082] For step S2: The delirium factor processing module 102 generates the parameters required for the delirium assessment model based on the attributes of the behavioral information and the attributes of the feedback information.

[0083] The "attributes of feedback information" in step S2, such as the feedback information mentioned above, can include four pieces of information: yes, no, incorrect, and correct. The values ​​corresponding to the attributes of the feedback information are all non-numerical. Therefore, the delirium factor processing module 102 needs to generate the parameters required for the delirium assessment model based on the attributes of the feedback information. The "parameters required for the delirium assessment model" refer to at least one or more combinations of four parameters: the patient's response, the assistant's response, the patient's behavior during the assessment process, and the patient's medical history. For example, for the parameters corresponding to the attributes of feedback information about the subject of assessment and / or about the assistant, "yes or incorrect" can be set to 1, and "no or correct" can be set to 0. For example, when the patient gives "yes" feedback information for item 08, the parameter value corresponding to the patient's response parameter must include at least A081. A indicates the patient's response parameter, 08 indicates for item 08, and 1 indicates "yes" feedback information. For example, when an assistant provides "no" feedback for item 13, the parameter value corresponding to the assistant's response must include at least B130. B indicates the assistant's response parameter, 13 indicates item 13, and 0 indicates "no" feedback. The required parameters are determined based on the delirium assessment model. To determine which parameters are needed for the delirium assessment model, the corresponding parameters are generated based on the attributes of the feedback information and the attributes of the behavioral information, and the corresponding parameter values ​​are determined.

[0084] The phrase "behavioral information of the subject during a rapid assessment of delirium and confusion" refers to the collection and analysis of the subject's external behavior during the assessment process from a third-party perspective using a mobile electronic device 101. For example, if an assistant asks the subject one of the scale items, "What year is it?", the subject might only answer correctly after the assistant repeats the item at least twice. In this case, the patient may appear unable to keep up with the conversation or show inappropriate distraction due to environmental stimuli. The mobile electronic device 101 analyzes the subject's response process based on the video it collects to determine the behavioral information related to that item. This video-based assessment differs from the assistant's sensory perspective; it analyzes the patient's actual reactions from an objective standpoint, avoiding reliance solely on the subjective and potentially biased interpretation of the assistant, thus ensuring the accuracy and reliability of the intelligent assessment system proposed in this invention.

[0085] More preferably, the "behavioral information" may include several pieces of information such as reaction time (a), blinking frequency (b), relevance of feedback information (c), and speech rate (d), as well as a third-party judgment (C) related to the patient's answer. For items in the scale that require assistance from an assistant, this is done after the subject of assessment has completed the corresponding item. The assistant mainly relies on memory and senses to answer. The intelligent assessment system provided by this invention, however, utilizes the video processing technology of the mobile electronic device 101 to provide a third-party judgment (C) that can cross-verify with the assistant's answer. For example, when the patient gives a correct answer to item 08 "What is this place?", and the mobile electronic device 101 analyzes and determines that the subject of assessment answered correctly, then the parameter value corresponding to the patient's behavior parameter during the assessment process will at least include C081.

[0086] More preferably, for the relevant items in the scale that require assistance from an assistant, each item in items 12-20 answered by the assistant is pre-set to correspond to at least one behavioral information. For example, item B11, "During the assessment, did the patient experience drowsiness, lethargy, or coma?", is pre-set to be associated with blinking behavior b in the behavioral information. As another example, item B18 is pre-set to be associated with blinking behaviors a and b in the behavioral information. More preferably, for the relevant items in the scale that require assistance from an assistant, each item in items 1-10 answered by the subject to be assessed is pre-set to correspond to a third-party judgment C in the behavioral information. For example, item 1 is pre-set to be associated with a third-party judgment C1 regarding item 1 in the behavioral information.

[0087] Here, we take "behavioral information of reaction time 'a'" as an example: The assistant asks questions to the subject of assessment one by one according to a pre-set item order. After the questioning is completed, based on the mobile electronic device 101 in step 1, the reaction time corresponding to each item in the pre-set item order is determined. Based on its analysis of the changing trends of the reaction times corresponding to each item in items 1 to 10 answered by the subject of assessment, the mobile electronic device 101 obtains the attribute of the behavioral information of reaction time 'a'—whether 'a' fluctuates or not.

[0088] Preferably, the "pre-set item order" for questioning the subject of evaluation refers to the order of items 4, 5, 6, 7, 8, 9, 10, 1, 2, and 3. Preferably, the "pre-set item order" requiring answers from assistants refers to the order of items 16, 17, 18, 19, 20, 13, 14, 15, 11, 12, 21, and 22. The "pre-set item order" for questioning the subject of evaluation takes precedence over the "pre-set item order" requiring answers from assistants.

[0089] Since the values ​​corresponding to the attributes of behavioral information are all non-numerical, the delirium factor processing module 102 in step S2 needs to generate the parameters required for the delirium assessment model based on the attributes of the feedback information. Similarly, the "parameters required for the delirium assessment model" in step S2 refer to at least one or a combination of four parameters: the patient's response, the assistant's response, the patient's behavior during the assessment process, and the patient's medical history. For the parameters corresponding to the attributes of behavioral information, "fluctuation" can be set to 1, and "no fluctuation" can be set to 0. For example, if the mobile electronic device 101 analyzes the changing trends of reaction time corresponding to each item in a pre-set order and obtains the attribute of the behavioral information—a—that shows fluctuation in reaction time a, then the parameter value corresponding to the patient behavior parameter during the assessment process must include at least a1. Correspondingly, for several pieces of information such as blinking status b, the correlation of feedback information c, and speech rate d, the parameter value corresponding to the patient behavior parameter during the assessment process must include at least one or more of b1, b0, c1, c0, d1, and d0.

[0090] For step S3: The delirium assessment module 103 calculates the value of at least one delirium feature obtained by the delirium assessment model based on the generated parameters.

[0091] Step S3 is more specifically: obtaining the parameter values ​​of the above-mentioned parameters; using the delirium assessment model to match and update the parameter values ​​of the above-mentioned parameters; using the delirium assessment model to calculate based on the parameter values ​​obtained after matching and updating; obtaining the assessment value of at least one delirium feature obtained by the rapid assessment of delirium confusion of the subject to be assessed.

[0092] "Delirium characteristics" include delirium characteristic k, k∈{1,2,3,4}, namely delirium characteristic 1, delirium characteristic 2, delirium characteristic 3, and delirium characteristic 4. The "assessment value" of a delirium characteristic includes two results: negative and positive. A negative result can be represented by "-", and a positive result by "+". Specifically, based on the rapid assessment method for delirium with altered consciousness, delirium characteristic 1 refers to acute onset or fluctuating changes; delirium characteristic 2 refers to inattention; delirium characteristic 3 refers to confused thinking; and delirium characteristic 4 refers to altered level of consciousness. The rapid assessment method for delirium with altered consciousness requires that delirium characteristics 1 and 2 be met, and at least one or two of delirium characteristics 3 or 4 be met.

[0093] The evaluation value of delirium feature 1 is determined based on items 8-10 and items 18-20. By pre-setting, delirium feature 1 corresponds to Aiji∈{08,09,10} and Biji∈{18,19,20}.

[0094] Among them, delirium feature 2 is determined based on items 4-7 and items 16-17. By pre-setting, delirium feature 2 corresponds to Aiji∈{04,05,06,07} and Biji∈{16,17}.

[0095] Among them, delirium feature 3 is determined based on items 1-3 and items 13-15. By pre-setting, delirium feature 3 corresponds to Aiji∈{01,02,03} and Biji∈{13,14,15}.

[0096] Among them, delirium feature 4 is determined based on entries 11-12. By pre-setting, delirium feature 4 corresponds to Biji∈{11,12}.

[0097] For delirium feature 1, more preferably, after Aij and / or Bij corresponding to items 1 to 20 are retrieved from database 106 for calculation, if delirium feature 1 is negative, delirium feature 2 is positive, delirium feature 3 is positive and / or delirium feature 4 is positive, delirium feature 1 is re-determined based on Biji∈{21,22}, and the evaluation value of delirium feature 1 is updated according to the determination result.

[0098] Regarding the "delirium assessment model" in step S3, this model pre-stores several calculation formulas. For example, it includes a delirium assessment calculation formula based on the rapid assessment method for delirium confusion.

[0099] Therefore, more specifically regarding the "delirium assessment model" in step S3, step S3 includes at least one or more of the following steps:

[0100] S31: Obtain several parameters and their corresponding parameter values ​​determined by the delirium factor processing module 102, including at least Aij, Bij, ζj, and Cij;

[0101] Aij refers to the parameter of the patient's answer, where i indicates the item i and j indicates the feedback information as "yes, incorrect, no, or correct".

[0102] Bij refers to the parameter of the answer given by the assistant, i refers to the item i, and j indicates the feedback information as "yes, incorrect, no or correct".

[0103] ζj refers to the parameter of patient behavior during the assessment process, ζ refers to at least one behavioral piece of information, and j indicates the feedback information of "fluctuation or no fluctuation".

[0104] Cij refers to the parameter of patient behavior during the assessment process, C refers to the third-party judgment contained in at least one behavioral information, i refers to the feedback information for item i, and j indicates "yes, no, no or correct".

[0105] i∈{01,02....09,10....21,22}, j∈{1,0}, ζ∈{a,b,c,d....};

[0106] S32: Based on the association between at least one Aij and at least one Cij, match Aij with Cij. If Aij and its corresponding Cij match successfully, output Aij. If Aij and its corresponding Cij fail to match, update the j value in Aij with the j value in Cij, and then output the updated Aij.

[0107] S33: Based on the association between at least one Bij and at least one ζj, match Bij with at least one ζj. If a match is successful between Bij and any one of the corresponding at least one ζj, output Bij. If a match fails between Bij and all of the corresponding ζj, update the j value in Bij with the j value in ζj, and then output the updated Bij.

[0108] The delirium assessment module 103 pre-stores the association relationship between at least one Aij and at least one Cij, and the association relationship between at least one Bij and at least one ζj;

[0109] The association between at least one Aij and at least one Cij is determined by the fact that the value between the first non-numeric digit and the last non-numeric digit can be regarded as the corresponding i value; the association between at least one Bij and at least one ζj is determined by a pre-defined method.

[0110] The matching method between Bij and at least one ζj refers to comparing the j value in Bij with the j value in at least one ζj. If the two values ​​are the same, the match is successful; otherwise, the match fails.

[0111] The matching method between Aij and Cij refers to comparing the j value in Aij with at least one j value in Cij. If the two values ​​are the same, the match is successful; otherwise, the match fails.

[0112] S34: Based on the pre-defined delirium feature k, k∈{1,2,3,4}, retrieve Aij and Bij corresponding to the entries ii∈{01,02....19,20} for at least one delirium feature k one by one, and determine whether the evaluation value of the delirium feature k is negative or positive according to the pre-defined evaluation value judgment conditions of the delirium feature k;

[0113] "The pre-set evaluation criteria for delirium feature k" refers to retrieving all Aij and / or Bij corresponding to delirium feature k. If the parameter value j of any Aij or any Bij is 1, then the evaluation value of delirium feature k is positive; if the parameter value j of all Aij and / or Bij is 0, then the evaluation value of delirium feature k is negative.

[0114] As above, the parameters corresponding to the attributes of the object to be evaluated and / or the feedback information of the assistants are set to 1 if they are "present or incorrect" and 0 if they are "absent or correct".

[0115] S35: Based on the evaluation value of the currently determined delirium feature k, k∈{1,2,3,4}, determine whether it meets the preset selective evaluation conditions. If it is determined that it does not meet the preset selective evaluation conditions, output the evaluation value of the above-mentioned determined delirium feature k, k∈{1,2,3,4}; otherwise, prompt the assistant to perform selective evaluation, and update the evaluation value of the currently determined delirium feature k, k∈{1,2,3,4} based on the evaluation result of selective evaluation, and output the updated evaluation value.

[0116] "Selective assessment conditions" refer to the following: after Aij and / or Bij corresponding to items 1 to 20 are retrieved from database 106 for calculation, if the currently determined delirium feature 1 is negative, delirium feature 2 is positive, delirium feature 3 is positive and / or delirium feature 4 is positive, the assistant is prompted to conduct selective assessments including items 21 to 22.

[0117] The process of “selective evaluation” refers to re-determining delirium feature 1 based on Biji∈{21,22} corresponding to delirium feature 1, and updating the evaluation value of delirium feature 1 according to the determination result.

[0118] Through the aforementioned pre-set operations, the analysis data from the mobile electronic device 101 (i.e., a third party) can be correlated with the responses input by the person being evaluated or the assistant. Therefore, the analysis data from the third party can be used as supplementary evidence to re-verify the manually input responses, especially for responses input by assistants that are difficult to avoid due to their strong subjectivity and sensory bias. This not only enables the assessment of delirium characteristics of the person being evaluated, but also generates the assessment results based on the actual reactions and states exhibited by the person being evaluated during the assessment process. It can effectively correct and prompt responses input by assistants that exhibit strong subjectivity and sensory bias, thereby improving the accuracy and reliability of delirium characteristic assessment.

[0119] To clarify the setup method of "Mobile Electronic Device 101" in this intelligent assessment system, the equipment used in the system is described below: The intelligent assessment system includes at least a handheld intelligent mobile terminal, a display, and an input device. The handheld intelligent mobile terminal is operated by an assistant, the display can be set up on the hospital bed for the person being assessed to view, and the input device is operated by the person being assessed. The handheld intelligent mobile terminal can be a smartphone, smartwatch, smart bracelet, tablet, laptop, etc. The display is the external device connected to the handheld intelligent mobile terminal, and the assistant operates it to control the display interface. The input device can be an external input device similar to a projector controller or mouse, connected to the display, allowing the person being assessed to input information onto the display by holding the input device. The input device has only two physical control buttons. One button is the mouse wheel, primarily used for scrolling through vertically arranged options. Scrolling the mouse wheel allows users to navigate up or down. For example, if an item requires the user to input a number, the display shows numbers 1-9 vertically; scrolling the mouse wheel selects the desired number. The other button is a trigger button; pressing this button inputs the currently selected option into the display. The input device is simple to use and operate, making it particularly beneficial for users, especially those who are unable to speak, such as those on mechanical ventilation or central venous catheters. This allows them to indirectly respond to questions from assistants through manual input.

[0120] Based on the above, the "mobile electronic device 101" is mainly distinguished by two methods of acquiring the external appearance of the object to be evaluated: contact-based and contactless acquisition. The contactless acquisition method includes the audio-visual processing module 1011, which is a camera mounted on the aforementioned display. The contact-based acquisition method includes the external input device 1012. The external input device 1012 refers to the input device in the intelligent evaluation system.

[0121] The following provides another preferred embodiment for steps S1-S3 above. This embodiment may be a further improvement and / or supplement to the above embodiments, and repeated content will not be described again. Where there is no conflict or contradiction, the whole and / or part of other embodiments may be used as a supplement to this embodiment:

[0122] S1: Preprocessing Step. Based on the patient's medical history, the preprocessing module generates first, second, and third pre-judgment data that are associated with at least three physiological states of the patient. The preprocessing module determines at least one feedback acquisition mode after comprehensively processing the above pre-judgment data.

[0123] The first set of pre-judgment data relates to the language expression ability of the subject being assessed. This data can be based on the patient's medical history, such as for patients wearing a breathing mask or diagnosed with post-operative speech nerve compression who are unable to speak, or for patients who can speak independently, the fluency, clarity, and logical coherence of their speech. The second set of pre-judgment data relates to the subject's physical activity ability. This data can be obtained from a behavior sensor placed on the patient's hand, which monitors hand movements and generates data on the subject's voluntary hand control and sluggishness. The third set of pre-judgment data relates to the subject's facial expression ability. This data can also be based on the patient's medical history, such as for patients on a ventilator or with oral intubation. The comprehensive condition processing here refers to a comprehensive analysis of the subject's status based on these three sets of pre-judgment data, selecting feedback acquisition modes that can effectively collect feedback from the subject. Feedback acquisition modes could include, for example, remote control operation or video capture and analysis.

[0124] S2: Under the feedback acquisition mode determined after prediction, when the assistant conducts a rapid assessment of delirium and confusion in the subject to be assessed, at least one acquisition module acquires feedback data about one or more of the subject's voice, video images, hand behavior, and point options on the assessment interface. Then, one or more of the reaction time analysis unit, eye movement analysis unit, correlation analysis unit of feedback information, speech rate analysis unit, and hand behavior analysis unit in the acquisition module process the feedback data according to the assessment items and prediction data to obtain at least one primary analysis information corresponding to each. A data processing module in the acquisition module performs secondary processing on the multiple primary analysis information to obtain third-party judgment information on the patient's behavior during the assessment process.

[0125] For the assessment items in the scale that require assistance from personnel, each item in items 12-20, answered by the assistance personnel, is pre-set to correspond to at least one analysis unit. Each acquisition module includes at least one or more of the following: reaction time analysis unit, eye movement analysis unit, feedback information correlation analysis unit, speech rate analysis unit, and hand behavior analysis unit. For example, item B11, "During the assessment, did the patient experience drowsiness, lethargy, or coma?", is pre-set to be associated with the aforementioned reaction time analysis unit. For each analysis unit, each unit corresponds to at least one or more of the first and second pre-judgment data. Based on these pre-judgment data, some disruptive information in the feedback, influenced by the subject's own habits, can be eliminated.

[0126] Preferably, the reaction time analysis unit, eye movement analysis unit, feedback information correlation analysis unit, speech rate analysis unit, etc., correspond to several pieces of information such as reaction time a, blinking status b, feedback information correlation c, and speech rate d.

[0127] Preferably, the reaction time analysis unit processes the feedback data based on the evaluation items and the predicted data to obtain the reaction time 'a'. The assistant questions the subject of evaluation one by one according to a pre-set item order. Based on a pre-set condition of the time between the assistant's questioning ending and the subject's response, the reaction time analysis unit can determine the reaction time corresponding to each item in the pre-set item order. Based on its analysis of the changing trends of the reaction times corresponding to each of the items 1 to 10 answered by the subject of evaluation, the reaction time analysis unit obtains the attribute of the behavioral information of reaction time 'a'—whether 'a' fluctuates or not.

[0128] Preferably, the relevance *c* of the feedback information is used to process the feedback data based on the assessment items and the predicted data to obtain the relevance *c* of the feedback information. The relevance *c* of the feedback information refers to the discrepancy between the attributes of the obtained feedback information and the attributes of the item's answer. This attribute discrepancy does not mean that the feedback information does not contain the correct answer, but rather that there is no relevance between the two. For example, regarding item 7, "Can you count down the months from December?", the attribute of the item's answer is a logically continuous number, while the patient might answer with their own birthdate or confusedly repeat the question posed by the assistant. The attribute of the feedback information is personal information or comprehension impairment, neither of which constitutes a logically continuous number. In the above situation, an attribute discrepancy exists, and the relevance *c* of the feedback information is judged to fluctuate, indicating that the person being assessed exhibits unclear thinking, irrelevant answers, or an inability to keep up with the topic being discussed during the assessment process.

[0129] Preferably, the eye movement analysis unit is used to process the feedback data according to the assessment items and the prediction data to obtain the blink status b. By monitoring and calculating the patient's blink frequency during the assessment process, it is possible to determine whether the blink status b fluctuates. If the blink status b is determined to fluctuate, it indicates that the subject under assessment has changes in reaction speed, a tendency to fall asleep, or low alertness during the assessment process.

[0130] Preferably, the speech rate analysis unit processes the feedback data based on the assessment items and predicted data to obtain the speech rate d. Here, speech rate d refers to the frequency of hand movements. For example, for item 7, "Can you count down the months starting from December?", the display shows numbers 1-20 vertically arranged along with several answer items unrelated to the question. The person being assessed can select different numbers or items by scrolling the mouse wheel, and pressing the trigger button will input the currently selected option into the display. During this process, the frequency of the patient scrolling the mouse wheel and pressing the trigger button is recorded. If fluctuations in speech rate d are detected, it indicates that the person being assessed has changes in speech speed, reaction speed, or difficulty keeping up with the topic being discussed during the assessment.

[0131] Preferably, the third-party judgment information on patient behavior obtained during the assessment process through secondary processing by the data processing module includes at least one parameter value determined by the primary analysis information. By pre-setting, the data processing module sets fluctuations in the primary analysis information to a numerical value of 1, and the absence of fluctuations to 0. For example, when the reaction time analysis unit analyzes the changing trends of reaction times corresponding to each item in a pre-set order and obtains a fluctuation in the attribute of the behavioral information of reaction time a—a—then the parameter value corresponding to the third-party judgment information includes at least a1. Correspondingly, for other primary analysis information such as blinking behavior b, the relevance of feedback information c, and speech rate d, the parameter value corresponding to the third-party judgment information includes at least one or more of b1, b0, c1, c0, d1, and d0.

[0132] S3: The identification module uses a delirium assessment model to calculate, based at least on third-party judgment information, an assessment value for at least one delirium feature obtained from a rapid assessment of delirium confusion in the subject to be assessed. Based on the assessment values ​​of several delirium features, it can be determined whether the subject to be assessed has been labeled with delirium or not.

[0133] Therefore, more specifically regarding the "delirium assessment model" in step S3, step S3 includes at least one or more of the following steps:

[0134] S31: Obtain several parameters and their corresponding parameter values ​​determined by the generation module 102, including at least Aij, Bij, ζj, and Cij;

[0135] Aij refers to the parameter of the patient's answer, where i indicates the item i and j indicates the feedback information as "yes, incorrect, no, or correct".

[0136] Bij refers to the parameter of the answer given by the assistant, i refers to the item i, and j indicates the feedback information as "yes, incorrect, no or correct".

[0137] ζj refers to the parameter of patient behavior during the assessment process, ζ refers to at least one behavioral piece of information, and j indicates the feedback information of "fluctuation or no fluctuation".

[0138] Cij refers to the parameter of patient behavior during the assessment process, C refers to the third-party judgment contained in at least one behavioral information, i refers to the feedback information for item i, and j indicates "yes, no, no or correct".

[0139] i∈{01,02....09,10....21,22}, j∈{1,0}, ζ∈{a,b,c,d....};

[0140] S32: Based on the association between at least one Aij and at least one Cij, match Aij with Cij. If Aij and its corresponding Cij match successfully, output Aij. If Aij and its corresponding Cij fail to match, update the j value in Aij with the j value in Cij, and then output the updated Aij.

[0141] S33: Based on the association between at least one Bij and at least one ζj, match Bij with at least one ζj. If a match is successful between Bij and any one of the corresponding at least one ζj, output Bij. If a match fails between Bij and all of the corresponding ζj, update the j value in Bij with the j value in ζj, and then output the updated Bij.

[0142] The identification module 103 pre-stores the association relationship between at least one Aij and at least one Cij, and the association relationship between at least one Bij and at least one ζj;

[0143] The association between at least one Aij and at least one Cij is determined by the fact that the value between the first non-numeric digit and the last non-numeric digit can be regarded as the corresponding i value; the association between at least one Bij and at least one ζj is determined by a pre-defined method.

[0144] The matching method between Bij and at least one ζj refers to comparing the j value in Bij with the j value in at least one ζj. If the two values ​​are the same, the match is successful; otherwise, the match fails.

[0145] The matching method between Aij and Cij refers to comparing the j value in Aij with at least one j value in Cij. If the two values ​​are the same, the match is successful; otherwise, the match fails.

[0146] S34: Based on the pre-defined delirium feature k, k∈{1,2,3,4}, retrieve Aij and Bij corresponding to the entries ii∈{01,02....19,20} for at least one delirium feature k one by one, and determine whether the evaluation value of the delirium feature k is negative or positive according to the pre-defined evaluation value judgment conditions of the delirium feature k;

[0147] "The pre-set evaluation criteria for delirium feature k" refers to retrieving all Aij and / or Bij corresponding to delirium feature k. If the parameter value j of any Aij or any Bij is 1, then the evaluation value of delirium feature k is positive; if the parameter value j of all Aij and / or Bij is 0, then the evaluation value of delirium feature k is negative.

[0148] As above, the parameters corresponding to the attributes of the object to be evaluated and / or the feedback information of the assistants are set to 1 if they are "present or incorrect" and 0 if they are "absent or correct".

[0149] S35: Based on the evaluation value of the currently determined delirium feature k, k∈{1,2,3,4}, determine whether it meets the preset selective evaluation conditions. If it is determined that it does not meet the preset selective evaluation conditions, output the evaluation value of the above-mentioned determined delirium feature k, k∈{1,2,3,4}; otherwise, prompt the assistant to perform selective evaluation, and update the evaluation value of the currently determined delirium feature k, k∈{1,2,3,4} based on the evaluation result of selective evaluation, and output the updated evaluation value.

[0150] "Selective assessment conditions" refer to the following: after Aij and / or Bij corresponding to items 1 to 20 are retrieved from the cloud platform 106 for calculation, if the currently determined delirium feature 1 is negative, delirium feature 2 is positive, delirium feature 3 is positive and / or delirium feature 4 is positive, the assistant is prompted to conduct selective assessments including items 21 to 22.

[0151] The process of “selective evaluation” refers to re-determining delirium feature 1 based on Biji∈{21,22} corresponding to delirium feature 1, and updating the evaluation value of delirium feature 1 according to the determination result.

[0152] S4: After the delirium assessment module 103 obtains the assessment value of at least one delirium feature obtained from the rapid assessment of delirium confusion of the subject to be assessed, the assessment processing module 104 combines the above assessment values ​​of at least one delirium feature and determines whether the combination meets the predetermined assessment conditions.

[0153] "Combining the above assessment values ​​for at least one delirium feature" means summarizing the assessment values ​​for at least one delirium feature into a combination Σ, such as {k1+,k2+,k3+,k4+}.

[0154] "Pre-defined assessment conditions" refer to three pre-set combinations of conditions for determining whether the assessed object has delirium: Ψ1{k1+,k2+,k3+,k4+}, Ψ2{k1+,k2+,k3+,k4-}, and Ψ3{k1+,k2+,k3-,k4+}.

[0155] The method of “determining whether the combination meets the predetermined evaluation conditions” refers to comparing the combination Σ with the combination {Ψ1, Ψ2, Ψ3} respectively.

[0156] S41: When the combination of several evaluation values ​​for at least one delirium feature obtained from the rapid assessment of delirium consciousness confusion of the subject to be evaluated satisfies the predetermined evaluation conditions, the delirium assessment result of the subject to be evaluated is output as follows: the subject to be evaluated has delirium.

[0157] Specifically, if the combination Σ matches one of the combinations {Ψ1, Ψ2, Ψ3}, that is, if the combination of several evaluation values ​​for at least one delirium feature obtained from the rapid assessment of delirium consciousness confusion of the subject to be evaluated satisfies the predetermined evaluation conditions, then the delirium evaluation result of the subject to be evaluated is output as follows: the subject to be evaluated has delirium.

[0158] S42: When the combination of several assessment values ​​for at least one delirium characteristic obtained from the rapid assessment of delirium confusion of the subject to be assessed does not meet the predetermined assessment conditions, the historical information related to the subject to be assessed in the medical information management system 107 is retrieved, and a secondary assessment is performed in combination with the medical information and / or assessment information in the historical information to determine the potential risk level Q of delirium of the subject to be assessed.

[0159] In cases where combination Σ does not match any of the combinations {Ψ1, Ψ2, Ψ3}, it is considered that "the combination of several evaluation values ​​for at least one delirium characteristic obtained from the rapid assessment of delirium consciousness confusion of the subject to be evaluated does not meet the predetermined evaluation conditions." Here, "evaluation information" refers to behavioral information and feedback information.

[0160] It should be noted that the specific embodiments described above are exemplary, and those skilled in the art can devise various solutions inspired by the disclosure of this invention. These solutions all fall within the scope of this invention and its protection. Those skilled in the art should understand that this specification and its accompanying drawings are illustrative and not intended to limit the scope of the claims. The scope of protection of this invention is defined by the claims and their equivalents.

Claims

1. A delirium risk monitoring device, characterized in that, The delirium risk monitoring device includes: At least one mobile electronic device acquires feedback information about the subject to be assessed and acquires behavioral information of the subject to be assessed during a rapid assessment of delirium and confusion. The feedback information refers to the subject's selection of several items on the rapid assessment scale for delirium and confusion. The feedback information includes yes, no, correct, and incorrect. The behavioral information includes reaction time, blinking, and speech rate. The delirium factor processing module generates the parameters required for the delirium assessment model based on the attributes of behavioral information and feedback information, respectively. The delirium assessment module calculates the value of at least one delirium feature obtained from the rapid assessment of delirium confusion of the subject under assessment based on the generated parameters using the delirium assessment model. The at least one delirium feature includes acute onset, inattention, confusion, and altered level of consciousness. An evaluation processing module is used to combine several evaluation values ​​of at least one delirium feature obtained by the delirium evaluation module, and to determine whether the combination meets predetermined evaluation conditions. The delirium factor processing module is also used to retrieve the dominant and latent factors related to the subject of assessment from the medical information management system after the subject of assessment has completed at least one rapid assessment of delirium confusion. Based on the attributes of the dominant and latent factors, it generates the tags required by the delirium risk monitoring module. The dominant factors refer to the primary and secondary risk factors of delirium, which are information that can be determined without assessment. The dominant factors include the subject of assessment's medical history and medication information. The latent factors refer to the delirium assessment data determined during the rapid assessment of delirium confusion, which are information determined through assessment. The delirium assessment data includes the trend of delirium potential risk level and at least one delirium feature. The trend of delirium potential risk level refers to the information on the change of delirium potential risk level over time based on each assessment of the subject of assessment. The delirium risk monitoring module is used to obtain multiple case information groups in the cloud platform that match the subject to be evaluated based on the generated tags, through information interaction with the cloud platform. After obtaining the tags, the delirium risk monitoring module first compares the latent factor tags of the subject to be evaluated with the latent factor tags of multiple case information in the cloud platform based on a pre-set similarity interval, thereby determining multiple case information in the cloud platform that meet the similarity interval and are used to form case information groups. The delirium risk monitoring module filters out multiple case information that meet the same tag as the tag of the subject to be evaluated based on the tags of each case information and the tags of the subject to be evaluated, and forms case information groups. The delirium risk monitoring module obtains the delirium risk prediction of the subject to be evaluated based on the proportion of each case information group in the obtained multiple case information groups, the trend of delirium potential risk level change, and the trend of delirium feature assessment value change, and determines the proportion of each case information group based on the number of case information in each case information group.

2. The delirium risk monitoring apparatus of claim 1, wherein, When the combination of several evaluation values ​​for at least one delirium feature obtained from the rapid assessment of delirium confusion of the subject to be evaluated meets the predetermined evaluation conditions, the delirium evaluation result of the subject to be evaluated is output, indicating that the subject to be evaluated has delirium. If the combination of several assessment values ​​for at least one delirium characteristic obtained from the rapid assessment of delirium confusion of the subject to be assessed does not meet the predetermined assessment conditions, then the historical information related to the subject to be assessed in the medical information management system is retrieved, and a secondary assessment is performed in combination with the assessment information in the historical information to determine the potential risk level of delirium of the subject to be assessed.

Images