Three-in-one non-invasive brain function monitoring system, method, medium, equipment and application

Abstract

The invention belongs to the technical field of brain function monitoring, and discloses a three-in-one non-invasive brain function monitoring system and method, a medium, equipment and application, and the monitoring method comprises the following steps: monitoring brain oxygen saturation of a critical patient; carrying out transcranial Doppler ultrasonic measurement on intracranial blood vessel blood flow velocity and ultrasonic measurement on the inner diameter of the optic nerve sheath; monitoring a quantitative electroencephalogram of the patient; various brain function monitoring series indexes and continuous changes of the indexes along with disease evolution are obtained, the pathogenic mechanism and pathogenesis of the indexes are found in time, and primary diseases and related complications of the primary diseases are treated. According to the invention, the brain oxygen saturation can avoid too low brain oxygen or low brain oxygen for a long time and too high brain oxygen due to oxygen utilization disorder; cerebral blood flow can be monitored through a transcranial Doppler (TCD) technology to find an optimal craniocerebral perfusion pressure (CPP), and a cerebral blood flow self-adjusting function is enabled to be in an optimal state; the electroencephalogram can avoid early warning of over-sedation or abnormal discharge and the like. Through organic combination of the three, noninvasive, real-time and dynamic monitoring of critical patients is realized.

Description

Technical Field

[0001] This invention belongs to the field of brain function monitoring technology, and in particular relates to a three-in-one non-invasive brain function monitoring system, method, medium, device and application. Background Technology

[0002] Currently, critically ill patients, especially those with sepsis, are receiving increasing attention due to their high mortality rate. However, critical illness-related brain dysfunction was previously easily overlooked in the ICU. Studies have shown that the incidence of critical illness-related brain dysfunction is as high as 15-50%, easily complicated by multiple organ dysfunction syndrome (MODS), with a high incidence of delirium, prolonged mechanical ventilation time, and increased mortality. However, due to the severity of the patients' conditions and the difficulty in moving them, invasive procedures involving the cranium or requiring transport for external examinations are risky and extremely difficult. Transcranial Doppler ultrasound and optic nerve sheath diameter examinations require physicians from both the neurology TCD team and the ophthalmology ultrasound team; bedside quantitative EEG monitoring requires physicians from both the neurology and electrophysiology teams. Current technology cannot provide bedside, continuous, and dynamic monitoring.

[0003] Based on the above analysis, the problems and defects of the existing technology are as follows: the patient's condition is serious and it is not easy to move. Invasive or out-of-town examinations involving the cranium are risky and very difficult, and bedside, continuous, and dynamic monitoring cannot be achieved.

[0004] The difficulty in solving the above problems and defects is as follows:

[0005] Since the significance of its integration is not yet clear, there is currently no mature technology that integrates the "three-in-one" non-invasive brain function monitoring. Existing technologies are all simply a collection of instruments containing the three technologies.

[0006] Clinical practice has revealed that these three factors are mutually causal and interrelated; however, due to the different granularities of time, it is quite difficult to display and manage these three data points in the same time dimension.

[0007] It needs to be combined with current cloud computing, big data and artificial intelligence analysis technologies, as there are deviations in the standardization of data acquisition by existing instruments.

[0008] The significance of solving the above problems and defects is as follows:

[0009] It has the potential to partially address the current serious disconnect between research findings and clinical practice in neurological function testing for critically ill patients;

[0010] Given its strong time dependence, the "three-in-one" non-invasive brain function monitoring method can provide critically ill patients with a more immediate, comprehensive, and systematic assessment of overall neurological dysfunction.

[0011] Develop a standardized process for integrating neurological function monitoring data from critically ill patients;

[0012] There is potential to further integrate various monitoring methods and multiple component suppliers to form a complete solution for neurological function assessment of critically ill patients in the future. Summary of the Invention

[0013] To address the problems existing in the prior art, this invention provides a three-in-one non-invasive brain function monitoring system, method, medium, device, and application.

[0014] This invention is implemented as follows: a three-in-one non-invasive brain function monitoring system, the three-in-one non-invasive brain function monitoring system comprising:

[0015] Brain oxygen saturation monitoring module; monitors the oxygen saturation in the patient's brain using a brain oxygen monitor;

[0016] Transcranial Doppler ultrasound and optic nerve sheath diameter measurement module: Uses a bedside portable ultrasound instrument to perform transcranial Doppler ultrasound monitoring and optic nerve sheath diameter measurement on patients;

[0017] Quantitative EEG module; using a bedside quantitative EEG monitor to monitor the patient's EEG.

[0018] Another objective of this invention is to provide a three-in-one non-invasive brain function monitoring method, which includes the following steps:

[0019] Step 1: Monitor the brain oxygen saturation and related indicators of critically ill patients;

[0020] Step two: Perform transcranial Doppler ultrasound and optic nerve sheath diameter measurement;

[0021] Step 3: Monitor the patient's quantitative electroencephalogram (EEG);

[0022] Step four: Obtain various brain function monitoring indicators and their dynamic changes, and identify their correlation with complications;

[0023] Step 5: By exploring the correlation between prognosis and disease outcome, we can uncover the possible underlying pathophysiological mechanisms of brain dysfunction in critically ill patients and construct a standardized management process based on non-invasive brain oxygenation, TCD / ONSD, and quantitative EEG.

[0024] Furthermore, step one specifically includes:

[0025] Near-infrared spectroscopy (NIRS) is used to continuously monitor regional brain oxygen saturation (rSO2) in patients. Brain oxygenation status is assessed by measuring the absorption and scattering of near-infrared light at different wavelengths in brain tissue. The data acquisition device is a sensor placed on the patient's forehead. Simultaneously, relevant indices such as tissue hemoglobin index (THI) or total hemoglobin concentration (HbT) are monitored.

[0026] Artifact detection: based on signal gradient, heartbeat variability, or standard difference constant.

[0027] Artifact removal algorithm:

[0028] Temporal Derivative Distribution Repair (TDDR): Corrects baseline drift and spike artifacts without requiring user parameters.

[0029] Spline interpolation: used to correct baseline drift and long-term slow motion artifacts.

[0030] Savitzky-Golay (SG) filtering or robust locally weighted regression smoothing (Rloess): used to remove high-frequency spike artifacts.

[0031] Principal Component Analysis (PCA) / Target PCA (tPCA): Identifies and removes artifact components by decomposing the signal.

[0032] Adaptive filters: These filters use a reference channel (such as accelerometer data) to remove motion artifacts and improve the signal-to-noise ratio.

[0033] Wavelet filtering is used to remove noise and artifacts.

[0034] Furthermore, step two specifically includes:

[0035] Two key measurements were performed on the patient using a bedside portable ultrasound device:

[0036] By penetrating the skull with ultrasound, the system measures the blood flow velocity in major intracranial vessels (such as the middle cerebral artery, anterior cerebral artery, and posterior cerebral artery); the system provides flow velocity, pulsatility index (PI), and displays the spectral waveform at specific points. A TCD provides audible sound and spectral waveforms.

[0037] The diameter of the sheath surrounding the optic nerve is measured using a bedside portable ultrasound device. Measurements are typically performed 3 mm posterior to the eyeball. Two measurements are taken in different planes (sagittal and axial) for each eye.

[0038] Furthermore, step three specifically includes:

[0039] Continuous monitoring and analysis of the patient's brain activity is performed using a bedside quantitative electroencephalogram (qEEG). Electrodes are placed on the scalp, following the international 10-20 system or its variants. Raw EEG signals are preprocessed after acquisition, including bad channel identification and labeling, and data filtering according to the desired frequency band.

[0040] Furthermore, step four specifically includes:

[0041] The system continuously acquires and records all brain oxygen saturation, TCD / ONSD, and quantitative electroencephalogram (EEG) indicators monitored in steps one, two, and three. The system tracks the dynamic trends of these indicators over time.

[0042] Multimodal data integration and time synchronization:

[0043] The system continuously acquires and records all indicators from brain oxygen saturation, TCD / ONSD, and quantitative electroencephalography (EEG) monitoring. The data integration framework employs a real-time multimodal medical data processing framework to integrate different data sources, which can be represented as follows:

[0044] F(D) = M(G(I),H(V),Q(P)), where G, H, and Q are preprocessing functions, and M is a machine learning model.

[0045] Data standardization and feature engineering:

[0046] Data standardization uses industry standards (such as CDISC, HL7, FHIR) to standardize data elements, terminology, and coding systems, extracting useful features from various modalities. Convolutional neural networks (CNNs) are used for image and video analysis, recurrent neural networks (RNNs) are used for sequence data (such as text and speech), and autoencoders are used for dimensionality reduction and anomaly detection.

[0047] Dynamic trend analysis:

[0048] The system tracks the dynamic changes of various brain function monitoring indicators over time. By analyzing continuous data, abnormal fluctuations in brain function indicators can be detected in a timely manner, and these fluctuations can be correlated with the evolution of the patient's condition and the occurrence and development of complications, thereby revealing potential pathogenic mechanisms and causes.

[0049] Furthermore, step five specifically includes:

[0050] Advanced data analytics and machine learning applications:

[0051] AI systems rapidly identify patterns in large amounts of medical data, assisting clinical decision-making and predicting the prognosis of patients with traumatic brain injury. Deep learning algorithms extract high-level abstract features from multidimensional patient data, improving diagnostic speed and accuracy. Multimodal fusion deep learning improves model accuracy and robustness by integrating multiple data sources.

[0052] Discovery of pathophysiological mechanisms and causal inference:

[0053] By using AI / ML to perform correlation analysis on multimodal data, we aim to uncover the underlying pathophysiological mechanisms leading to brain dysfunction in critically ill patients. Causal inference models utilize unstructured data as proxy signals for unobserved confounding factors to improve the accuracy of causal effect estimation.

[0054] Building a process-oriented management system:

[0055] Based on three non-invasive monitoring technologies—brain oxygen saturation, transcranial Doppler (TCD) / on-slow emission detection (ONSD), and quantitative electroencephalography (EEG)—a standardized, integrated management process has been established. This system will guide clinicians on how to comprehensively utilize these three monitoring data for early diagnosis, risk assessment, optimization of treatment plans, and real-time evaluation of treatment effectiveness. Ultimately, it aims to achieve non-invasive, real-time, dynamic, and systematic management of brain function in critically ill patients.

[0056] Another object of the present invention is to provide a computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the three-in-one non-invasive brain function monitoring method.

[0057] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the three-in-one non-invasive brain function monitoring method.

[0058] Another objective of this invention is to provide an information data processing terminal for implementing the aforementioned three-in-one non-invasive brain function monitoring method.

[0059] Combining all the above technical solutions, the advantages and positive effects of this invention are as follows: The three-in-one non-invasive brain function monitoring system provided by this invention can prevent brain oxygen saturation from becoming too low or remaining at a low level for a prolonged period; cerebral blood flow can be monitored by TCD to find the optimal CPP and self-regulate the function to the optimal state; and electroencephalography can avoid excessive sedation and provide early warning of abnormal discharges. The organic combination of these three technologies enables non-invasive, real-time, and dynamic monitoring of patients.

[0060] Solving existing problems: This invention clearly overcomes the problems of inconvenience in moving critically ill patients, high risk of invasive examinations, difficulty in continuous dynamic monitoring at the bedside, and difficulty in data integration in the existing technology.

[0061] Technical innovations: The innovations of this invention lie in the seamless integration of multimodal data, precise time synchronization, intelligent data processing (including advanced artifact removal and AI / ML analysis), and the construction of a data-driven pathophysiological mechanism discovery and management process system, which are not available in existing "instrument stacking" methods.

[0062] Clinical benefits:

[0063] Early diagnosis and intervention: The system can provide early warning of brain dysfunction, thereby enabling timely intervention and reducing the incidence of complications (such as delirium, MODS, and prolonged mechanical ventilation).

[0064] Improved patient prognosis: Data demonstrates the positive effects of the system application, such as reduced patient mortality and improved neurological function recovery.

[0065] Optimize healthcare resources: Through accurate prediction and personalized treatment, optimize the allocation of healthcare resources and reduce healthcare costs.

[0066] Promoting precision medicine: Providing personalized neurological function assessment and management plans for critically ill patients, and promoting the application of precision medicine in the field of neurocritical care. Attached Figure Description

[0067] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 This is a schematic diagram of the structure of the three-in-one non-invasive brain function monitoring system provided in an embodiment of the present invention;

[0069] Figure 1 The module includes: 1. Brain oxygen saturation monitoring module; 2. Transcranial Doppler ultrasound and optic nerve sheath diameter measurement module; 3. Quantitative electroencephalography (EEG) monitoring module.

[0070] Figure 2 This is a schematic diagram of the three-in-one non-invasive brain function monitoring system provided in an embodiment of the present invention.

[0071] Figure 3 This is a schematic diagram of optimized brain perfusion provided in an embodiment of the present invention.

[0072] Figure 4 This is a flowchart of the three-in-one non-invasive brain function monitoring system provided in an embodiment of the present invention. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0074] To address the problems existing in the prior art, the present invention provides a three-in-one non-invasive brain function monitoring system, method, medium, device and application. The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0075] like Figures 1-2 As shown, the three-in-one non-invasive brain function monitoring system provided in this embodiment of the invention includes:

[0076] Brain oxygen saturation monitoring module 1; monitors the oxygen saturation in the patient's brain using a brain oxygen monitor;

[0077] Transcranial Doppler ultrasound and optic nerve sheath diameter measurement module 2: Uses a bedside portable ultrasound instrument to perform transcranial Doppler ultrasound monitoring and optic nerve sheath diameter measurement on the patient;

[0078] Quantitative EEG monitoring module 3: Use a bedside quantitative EEG monitor to monitor the patient's EEG.

[0079] like Figure 3 As shown, the core of the "three-in-one" non-invasive brain function monitoring theory provided by the embodiments of the present invention is: brain perfusion is always within the scope of hemodynamics, the core of neurocritical care management is brain protection, and treatment begins with optimizing brain perfusion.

[0080] like Figure 4 As shown, the three-in-one non-invasive brain function monitoring method provided in this embodiment of the invention includes the following steps:

[0081] S101, used to monitor brain oxygen saturation in critically ill patients;

[0082] S102, perform transcranial Doppler ultrasound and optic nerve sheath diameter measurement;

[0083] S103, monitoring the patient's quantitative electroencephalogram;

[0084] S104: Acquire various brain function monitoring indicators and their dynamic changes, and discover their relationship with complications;

[0085] S105, through the correlation of prognosis, explores the possible intrinsic pathophysiological mechanisms of brain dysfunction in critically ill patients and constructs a "three-in-one" management process based on non-invasive brain oxygenation-TCD / ONSD-quantitative EEG.

[0086] The three-in-one non-invasive brain function monitoring method of the present invention includes the following five steps, aiming to achieve non-invasive, real-time, and dynamic monitoring of critically ill patients, and to deeply explore the intrinsic pathophysiological mechanisms of brain dysfunction and build a process-oriented management system.

[0087] Step 1: Monitor brain oxygen saturation and related indicators in critically ill patients

[0088] This step utilizes near-infrared spectroscopy (NIRS) to continuously monitor regional brain oxygen saturation (rSO2) in the patient's brain. NIRS is a non-invasive technique that assesses cerebral oxygenation by measuring the absorption and scattering of near-infrared light at different wavelengths in brain tissue. The data acquisition device is typically a sensor placed on the patient's forehead. In addition to rSO2, related indices such as tissue hemoglobin index (THI) or total hemoglobin concentration (HbT) are also monitored; these indicators reflect the balance between oxygen supply and consumption in brain tissue.

[0089] NIRS can continuously and non-invasively monitor rSO2, which is a significant advantage over intermittent or invasive monitoring methods, enabling real-time detection of subtle changes in cerebral oxygenation that may otherwise be overlooked. Traditional monitoring methods may fail to detect changes in cerebral oxygenation even when routine intraoperative monitoring (such as heart rate, blood pressure, and blood oxygen saturation) shows no change. Therefore, continuous NIRS monitoring can detect early or occult cerebral deoxygenation events, which is crucial for timely intervention in rapidly changing critical conditions and may prevent irreversible brain damage.

[0090] Calculation and explanation of core indicators:

[0091] Regional brain oxygen saturation (rSO2): directly reflects the balance of oxygen supply and demand in brain tissue. rSO2 values ​​below baseline or below specific thresholds (such as 75% or 80% of baseline, or an absolute value below 60%) are associated with postoperative neurocognitive dysfunction, higher mortality, increased intracranial pressure, and impaired cerebral perfusion.

[0092] Total hemoglobin concentration (HbT): Calculated by measuring the sum of the concentrations of oxyhemoglobin and deoxyhemoglobin. It reflects changes in cerebral blood volume. Calculations are typically based on a modified Beer-Lambert law, incorporating the extinction coefficient of hemoglobin.

[0093] Total hemoglobin reactivity index (THx): derived from the correlation between slow wave activity of HbT and arterial blood pressure (ABP).

[0094] Tissue oxygen responsiveness index (TOx): derived from the slow wave activity correlation between rSO2 and arterial blood pressure (ABP).

[0095] Positive THx and TOx values ​​typically indicate impaired cerebrovascular autoregulation. rSO2 provides a snapshot of oxygenation, while the inclusion of THI / HbT and the development of reactivity indices (THx, TOx) signify a shift in monitoring from purely static measurements to dynamic assessments of cerebrovascular autoregulation. By calculating and continuously monitoring THx and TOx, the system can dynamically assess the brain's ability to maintain stable blood flow amidst changes in systemic blood pressure. This goes beyond simple oxygenation level measurements, providing a deeper understanding of the adaptive capacity of the cerebrovascular system, offering a more complex and clinically relevant insight for optimizing cerebral perfusion pressure (CPP) and preventing secondary brain injury. This directly supports the present invention's goal of optimizing cerebral blood flow.

[0096] Data quality control and artifact removal:

[0097] NIRS data is susceptible to motion artifacts (such as spikes and baseline drift). Therefore, the system integrates a variety of advanced artifact removal techniques to ensure data reliability.

[0098] Artifact detection: based on signal gradient, heartbeat variability, or standard difference constant.

[0099] Artifact removal algorithm:

[0100] Temporal Derivative Distribution Repair (TDDR): Corrects baseline drift and spike artifacts without requiring user parameters.

[0101] Spline interpolation: used to correct baseline drift and long-term slow motion artifacts.

[0102] Savitzky-Golay (SG) filtering or robust locally weighted regression smoothing (Rloess): used to remove high-frequency spike artifacts.

[0103] Principal Component Analysis (PCA) / Target PCA (tPCA): Identifies and removes artifact components by decomposing the signal.

[0104] Adaptive filters: These filters use a reference channel (such as accelerometer data) to remove motion artifacts and improve the signal-to-noise ratio.

[0105] Wavelet filtering is used to remove noise and artifacts.

[0106] Clearly defined advanced artifact removal techniques for NIRS data are crucial for system reliability, as the raw NIRS signal is highly susceptible to patient motion noise. Without robust artifact removal capabilities, the accuracy and clinical utility of NIRS data will be severely compromised, potentially leading to false alarms or missed critical events. Therefore, detailed descriptions of these algorithms demonstrate a deep understanding of the practical implementation challenges and enhance system reliability and data integrity.

[0107] Table 1: Core Indicators, Calculation, and Clinical Significance of Brain Oxygen Saturation Monitoring

[0108]

[0109]

[0110] Step 2: Perform transcranial Doppler ultrasound and optic nerve sheath diameter measurement.

[0111] This step involves using a bedside portable ultrasound device to perform two key measurements on the patient.

[0112] Data acquisition and preprocessing (transcranial Doppler ultrasound, TCD):

[0113] By penetrating the skull with ultrasound, the system measures blood flow velocity in major intracranial vessels such as the middle cerebral artery, anterior cerebral artery, and posterior cerebral artery. The system provides flow velocity, pulsatility index (PI), and displays spectral waveforms at specific points. TCD provides audible sound and spectral waveforms, not just images, offering unique temporal resolution and real-time feedback for assessing cerebral hemodynamics. This real-time auditory and visual feedback allows clinicians to immediately assess dynamic changes in cerebral blood flow, such as the initial slope of systolic acceleration or end-diastolic velocity, which is crucial for rapid decision-making in acute neurological events such as vasospasm. It provides a continuous, dynamic assessment that static imaging cannot achieve.

[0114] Calculation and interpretation of core indicators (transcranial Doppler ultrasound, TCD):

[0115] Peak contraction velocity (PSV): The highest peak value in the spectrum.

[0116] End-diastolic flow (EDV): The point at the end of a cardiac cycle, before the peak of systole.

[0117] Average Flow Velocity (MFV): The formula is MFV = (PSV + 2 * EDV) / 3 12 This indicator reflects cerebral blood flow and perfusion status.

[0118] Pulsatility index (PI): A commonly used TCD parameter for measuring blood flow resistance. The formula is PI = (PSV - EDV) / MFV. PI is directly proportional to the pulse amplitude of systemic arterial blood pressure and inversely proportional to cerebral perfusion pressure (CPP). An elevated PI (especially >2) suggests low CPP under normal systemic hemodynamics, which may further indicate elevated intracranial pressure (ICP). PI is also affected by factors such as PaCO2, heart rate, cardiac output, and cerebrovascular resistance.

[0119] Resistance Index (RI): RI = (PSV - EDV) / PSV. Similar to PI, an increase in RI also indicates a low CPP.

[0120] Clinical applications: TCD can be used to assess cerebral blood flow, cerebral vascular reactivity, cerebral perfusion status, and predict cerebral vasospasm (e.g., an increase in mean flow velocity >50 cm / s).

[0121] While PI and RI are valuable, their interpretation requires comprehensive consideration of systemic hemodynamics (such as MAP and CPP) and other physiological factors (such as PaCO2). TCD parameters alone are insufficient for definitive diagnosis of ICP or specific brain states. The "trinity" system of this invention achieves a more comprehensive and accurate assessment by combining rSO2 (oxygenation) and qEEG (electrical activity) with TCD. For example, elevated PI accompanied by decreased rSO2 and abnormal EEG patterns is a more powerful indicator of impaired brain perfusion and dysfunction than any single parameter, highlighting the synergistic effect of multimodal monitoring.

[0122] Data acquisition and preprocessing (optic nerve sheath diameter, ONSD):

[0123] The diameter of the sheath surrounding the optic nerve is measured using a bedside portable ultrasound device. Measurements are typically performed 3 mm posterior to the eyeball. It is recommended to perform two measurements in different planes (sagittal and axial) for each eye to improve reliability.

[0124] The shift from manual ONSD measurement to automated image processing algorithms directly addresses the challenges of operator dependence and variability, thereby improving the reliability and consistency of ICP estimation. Manual ONSD measurement is cumbersome, time-consuming, and susceptible to human error due to operator experience. Automating this process significantly reduces operator variability and improves diagnostic consistency, making ONSD a more reliable tool for bedside ICP assessment. This represents a key improvement in clinical practice, addressing the need for rapid and consistent assessment in acute care settings.

[0125] Core Indicator Calculation and Interpretation (Optic Nerve Sheath Diameter, ONSD):

[0126] ONSD value: considered a non-invasive, rapid alternative indicator for assessing intracranial pressure (ICP).

[0127] Interpretation criteria:

[0128] <5mm: Indicates normal intracranial pressure.

[0129] 5-6mm: Gray area.

[0130] 6mm: This indicates elevated intracranial pressure.

[0131] Image processing algorithms:

[0132] Optimal frame selection strategy: The optimal frame in the ultrasound sequence is automatically identified using the Kernel Correlation Filter (KCF) tracking algorithm and the SimpleLinear Iterative Clustering (SLIC) segmentation algorithm.

[0133] Optic nerve sheath localization and measurement: Accurate mapping and measurement were performed using a Gaussian mixture model (GMM) combined with the KL divergence method.

[0134] Verification: The automated algorithm showed no significant difference from manual measurement results, and demonstrated high accuracy and reliability. 16 .

[0135] Applying advanced image processing techniques to ONSD measurement and combining it with its correlation with ICP provides a non-invasive alternative to this traditionally invasive and high-risk parameter. Direct ICP measurement is invasive and risky. ONSD, as a non-invasive alternative to ICP, offers further improved reliability through automated measurement. This non-invasive, automated ICP estimation, combined with TCD (Transcranial Doppler) and NIRS (National Institute for Health and Respiratory Syndrome), provides comprehensive hemodynamics and oxygenation status of the brain without invasive procedures, representing a significant breakthrough in patient safety and the feasibility of intensive care monitoring.

[0136] Table 2: Core Indicators, Calculation, and Clinical Significance of Transcranial Doppler Ultrasound and Optic Nerve Sheath Diameter Measurement

[0137]

[0138]

[0139] Step 3: Monitor the patient's quantitative electroencephalogram (EEG)

[0140] This step involves continuous monitoring and analysis of the patient's brain activity using a bedside quantitative electroencephalogram (qEEG). Electrodes are typically placed on the scalp, following the international 10-20 system or its variants. Raw EEG signals are preprocessed after acquisition, including bad channel identification and labeling, and data filtering according to the desired frequency band.

[0141] The shift from traditional visual EEG interpretation to quantitative electroencephalography (qEEG) represents a fundamental change, enabling objective, data-driven assessments of brain function—crucial for automated analysis and integration. Visual interpretation, being subjective and time-consuming, limits its continuous, real-time application in intensive care. qEEG transforms complex raw signals into quantifiable metrics (such as power and frequency), making them suitable for automated processing, comparison with standardized databases, and integration with other digital data streams. This is essential for the "dynamic monitoring" and "data integration" aspects of this invention.

[0142] Calculation and explanation of core indicators:

[0143] Spectral Analysis: Mathematical processing (such as Fourier transform) of raw EEG signals to extract quantitative indicators such as power, frequency, and asymmetry of various frequency components.

[0144] Frequency bands:

[0145] Delta (0.5–4 Hz): Usually associated with deep sleep or brain dysfunction.

[0146] Theta (4–7 Hz): associated with drowsiness or certain brain dysfunctions.

[0147] Alpha (8–13 Hz): associated with a relaxed and conscious state.

[0148] Beta (13–30Hz): associated with alertness or mental activity.

[0149] Amplitude integrated electroencephalography (aEEG / CFM): A simplified and continuous brain function monitoring technique that compresses EEG signals and displays them as a semi-logarithmic graph of peak-to-peak amplitude over time, facilitating long-term monitoring and real-time interpretation.

[0150] Clinical applications: Particularly suitable for neonatal intensive care, used to diagnose hypoxic-ischemic encephalopathy (HIE) and monitor / diagnose epileptic activity.

[0151] Characteristics include baseline, upper limit, sleep-wake cycle, epileptic seizures (high amplitude, repetitive waveforms), low amplitude or inhibitory waveforms (poor prognosis), and burst suppression pattern (poor prognosis).

[0152] Clinical significance: These indicators can reflect the overall functional state of the brain, cortical excitability, depth of sedation, and the presence of abnormal discharges (such as epileptiform activity). Real-time monitoring with qEEG can provide early warning of brain dysfunction such as excessive sedation, ischemic brain injury, and epileptic seizures.

[0153] The combination of qEEG (detailed spectral analysis) and aEEG (simplified continuous monitoring) provides a flexible and comprehensive approach to EEG activity assessment, adaptable to different levels of interpretability and clinical scenarios in intensive care. qEEG provides detailed quantitative indicators, while aEEG simplifies EEG data for real-time interpretation in the ICU, particularly suitable for neonates. This dual approach enables the system to provide in-depth neurophysiological analysis (qEEG for researchers / experts) and rapid, actionable bedside clinical information (aEEG). This ensures the system's usability across a wider range of users and clinical urgency levels, maximizing its practical application and impact in busy ICU environments.

[0154] Data quality control and artifact removal:

[0155] EEG signals are sensitive to various artifacts, such as electrooculography (EOG), electromyography (EMG), electrocardiography (ECG), and motion artifacts.

[0156] Artifact suppression techniques:

[0157] Independent Component Analysis (ICA): Decomposes EEG data into statistically independent components, then identifies and removes artifact components.

[0158] Principal Component Analysis (PCA): Converts correlated variables into uncorrelated principal components to remove eye-movement artifacts.

[0159] Empirical Mode Decomposition (EMD): The EEG signal is decomposed into Intrinsic Mode Functions (IMFs), combined with Detrended Fluctuation Analysis (DFA) for mode selection, and then a cleaner signal is extracted through Wavelet Packet Transform (WPT).

[0160] Wavelet method: used for noise reduction and artifact removal.

[0161] Adaptive filters: used to remove unwanted noise.

[0162] Advanced artifact removal algorithms for qEEG are emphasized as crucial for maintaining signal integrity and ensuring the accuracy of quantitative metrics, directly impacting the reliability of clinical decisions provided by the system. EEG is sensitive to irrelevant signal sources and artifacts (such as EOG and EMG), which distort the signal and reduce the signal-to-noise ratio. Accurate quantitative EEG analysis (e.g., power spectral density) heavily relies on clean data. Without robust artifact removal capabilities, derived quantitative metrics will be unreliable, potentially leading to misinterpretations and potentially harmful clinical decisions. Therefore, a detailed description of these complex algorithms highlights the system's commitment to data quality and clinical reliability.

[0163] Table 3: Core Indicators of Quantitative Electroencephalography, Calculation, and Clinical Significance

[0164]

[0165] Table 4: Common Artifacts in Each Modality and Their Removal Techniques

[0166]

[0167]

[0168] Step 4: Obtain various brain function monitoring indicators and their dynamic changes, and identify their correlation with complications.

[0169] This step is crucial for data integration and preliminary analysis. The system will continuously acquire and record all brain oxygen saturation, TCD / ONSD, and quantitative EEG indicators monitored in steps one, two, and three. More importantly, the system will track the dynamic trends of these indicators over time.

[0170] Multimodal data integration and time synchronization:

[0171] This system continuously acquires and records all indicators from brain oxygen saturation, TCD / ONSD, and quantitative electroencephalography (EEG) monitoring. The data integration framework employs a real-time multimodal medical data processing framework to integrate different data sources, which can be represented as follows:

[0172] F(D) = M(G(I),H(V),Q(P)), where G, H, and Q are preprocessing functions, and M is a machine learning model.

[0173] Precise time synchronization is crucial for multimodal signal fusion and is essential for downstream applications such as validation and supplemental information utilization. Aligning signals acquired by different devices through direct cross-correlation of temporal amplitudes makes it universally applicable to various signal types. Advanced systems utilize multi-channel EEG and eye-tracking technology to ensure precise synchronization, enhance artifact removal, and enable detailed analysis. This explicit focus on time synchronization and multimodal data fusion directly addresses the problems of "different temporal granularities" and "difficulty in managing data in the same time dimension" identified in background technologies. This is a core innovative step beyond simple data collection. It demonstrates a complex architectural design that goes beyond simply displaying three independent data streams. Precise time synchronization is essential for identifying causal relationships, correlating cross-modal events (e.g., rSO2 decreases occurring simultaneously with EEG slowdowns), and enabling advanced AI / ML analysis that relies on synchronized input. Without this, a "trinity" system will remain merely an "instrument stack."

[0174] Data standardization and feature engineering:

[0175] Data standardization employs industry standards (such as CDISC, HL7, and FHIR) to standardize data elements, terminology, and coding systems to achieve data aggregation, sharing, and interoperability. This helps eliminate variability arising from different research groups and data recording practices. Data standardization is not merely a technical detail; it is fundamental to system scalability, interoperability with existing healthcare systems (EHRs), and the effective training of powerful AI models. Existing instruments suffer from "data standardization bias," hindering integration with "cloud computing, big data, and artificial intelligence." Standardized data ensures consistency, accuracy, and efficiency. This is crucial because AI / ML models require high-quality, consistent data for effective training and reliable predictions. Without standardization, the accuracy and universality of the "big data" and "AI" components in this invention across different clinical settings or patient populations would be severely limited. It also facilitates data exchange and aggregation for large-scale studies.

[0176] Extracting useful features from various modalities is key to feature engineering. Deep learning methods play an important role in this, such as convolutional neural networks (CNNs) for image and video analysis, recurrent neural networks (RNNs) for sequence data (such as text and speech), and autoencoders for dimensionality reduction and anomaly detection.

[0177] Dynamic trend analysis:

[0178] The system tracks the dynamic changes of various brain function monitoring indicators over time. By analyzing this continuous data, abnormal fluctuations in brain function indicators can be detected in a timely manner, and their correlation with the evolution of the patient's condition and the occurrence and development of complications (such as delirium, MODS, prolonged mechanical ventilation, etc.) can be established, thereby revealing the potential pathogenic mechanisms and causes.

[0179] Shifting from static snapshots to dynamic trend analysis is crucial for the early detection and understanding of the progression of brain dysfunction in intensive care, as patient conditions can change rapidly. Critically ill patients exhibit "rapidly evolving physiological states." This invention aims to provide "real-time, dynamic monitoring." Analyzing trends over time can detect subtle, gradual changes or acute deteriorations that might be missed by measurements at a single time point. This dynamic perspective is essential for predicting impending complications, assessing the effectiveness of interventions in real time, and understanding the trajectory of brain injury, thereby enabling proactive rather than reactive clinical management.

[0180] Step 5: By analyzing the correlation with prognosis, explore the possible underlying pathophysiological mechanisms of brain dysfunction in critically ill patients, and construct a standardized management process based on non-invasive brain oxygenation, TCD / ONSD, and quantitative EEG.

[0181] This is the highest level of analysis and application. After acquiring and analyzing the dynamic changes of various brain function indicators and their association with complications, this step will further explore the correlation between these indicators and patient prognosis (such as mortality, length of hospital stay, and recovery of neurological function).

[0182] Advanced data analytics and machine learning applications:

[0183] Artificial intelligence / machine learning (AI / ML) plays a crucial role at this stage. AI systems can quickly identify patterns in large amounts of medical data, assisting clinical decision-making and predicting the prognosis of patients with traumatic brain injury (TBI). Deep learning algorithms can extract high-level abstract features from multidimensional patient data (such as neuroimaging, EEG, genes, and blood biomarkers), improving diagnostic speed and accuracy. Multimodal fusion deep learning improves model accuracy and robustness by integrating multiple data sources (images, text, sensor data, etc.). Early fusion (feature splicing) or late fusion (model output combination) strategies can be employed.

[0184] The application of AI / ML, especially multimodal deep learning, elevates systems from simple monitoring tools to complex diagnostic and prognostic aids, capable of discovering hidden patterns and making predictions that surpass human cognitive abilities. The complexity of severe illnesses and brain dysfunction makes it difficult for clinicians to identify subtle patterns and predict outcomes. AI / ML algorithms can identify patterns that are difficult for humans to perceive and deconstruct complex, multidimensional patient data. Multimodal deep learning improves accuracy and robustness by leveraging the strengths of different data sources. This marks a leap from passive monitoring to proactive, predictive medicine. The AI ​​component, acting as the "brain" of the "three-in-one" system, transforms raw, integrated data into actionable intelligence for personalized medicine and early intervention—this is the ultimate goal of this invention.

[0185] Discovery of pathophysiological mechanisms and causal inference:

[0186] Correlation analysis of multimodal data using AI / ML can delve deeper into the underlying pathophysiological mechanisms leading to brain dysfunction in critically ill patients. Causal inference models utilize unstructured data (such as wearable sensor measurements, clinical notes, and medical images) as proxy signals for unobserved confounding factors to improve the accuracy of causal effect estimation. This helps determine whether a specific intervention truly leads to prognostic improvement.

[0187] The ability to make causal inferences using multimodal data is a key advanced step in understanding the "why" behind brain dysfunction, going beyond simple correlations to identify the true pathophysiological mechanisms. Understanding the "intrinsic pathophysiological mechanisms" of brain dysfunction is challenging. This invention proposes using "prognostic correlation" and "advanced data analysis / ML." In particular, causal inference models can utilize unstructured data to explain unobserved confounding factors. This demonstrates that this invention focuses not only on clinical applications but also on scientific discovery. By identifying causal relationships between specific monitoring parameters and outcomes, the system can help pinpoint the exact mechanism of the injury (e.g., specific hemodynamic changes leading to ischemia) or the effectiveness of interventions. This deep understanding is crucial for developing targeted therapies and truly personalized medicine, thus fulfilling the invention's promise to "uncover potential intrinsic pathophysiological mechanisms."

[0188] Building a process-oriented management system:

[0189] Ultimately, based on three non-invasive monitoring technologies—brain oxygen saturation, TCD / ONSD, and quantitative electroencephalography (EEG)—a standardized, integrated management process system was constructed. This system will guide clinicians on how to comprehensively utilize these three monitoring data for early diagnosis, risk assessment, optimization of treatment plans, and real-time evaluation of treatment effectiveness. Ultimately, it aims to achieve non-invasive, real-time, dynamic, and systematic management of brain function in critically ill patients.

[0190] Building a standardized management process system signifies a shift from ad-hoc interpretations relying on expert experience to a more systematic, evidence-based approach to neurocritical care. Current neurological monitoring is fragmented and lacks integration. This invention aims to construct a "three-in-one" management system. This standardized process will reduce variability in clinical practice, ensure consistent application of monitoring data, and facilitate training. It transforms complex medical data into actionable clinical pathways, enabling advanced neurological monitoring to be used and utilized more broadly by healthcare providers, thereby improving the overall quality of patient care.

[0191] Table 5: Key Links and Decision Support in the Tripartite Non-invasive Brain Function Monitoring and Management Process System

[0192]

[0193]

[0194] 1. Detailed implementation of system modules

[0195] The three-in-one non-invasive brain function monitoring system of the present invention has been carefully designed at both the hardware and software levels to ensure its efficient and reliable operation.

[0196] Hardware integration: The brain oxygen monitor, bedside portable ultrasound instrument, and bedside quantitative EEG monitor are connected to the central processing unit through a unified data interface (such as USB, Ethernet, Bluetooth). This integration method ensures seamless data transmission and centralized processing from each module, avoiding the drawbacks of traditional distributed monitoring.

[0197] Software Architecture: The system's software architecture adopts a modular design, including a data acquisition layer, a data preprocessing layer, a data fusion layer, a data analysis and visualization layer, and a decision support layer. This layered design improves the system's scalability and maintainability, allowing for the future integration of more monitoring modalities or the upgrading of algorithms.

[0198] Database Design: The system employs an efficient database structure to store raw data, processed data, analysis results, and patient clinical information. The database design considers the storage and retrieval efficiency of real-time data streams, ensuring high performance even when processing large amounts of continuous monitoring data.

[0199] User Interface: The system provides an intuitive user interface that can display multimodal data in real time and dynamically. The interface design emphasizes readability and ease of use, using graphical displays and highlighting key indicators to help medical staff quickly understand the patient's brain function status and provide clear decision support information.

[0200] 2. Specific procedures for monitoring methods

[0201] The specific process of the monitoring method of this invention strictly follows the steps of S101-S105, and the data flow and algorithm call order have been optimized:

[0202] 1. Data Acquisition Synchronization Startup: Monitoring of NIRS, TCD / ONSD, and qEEG devices is simultaneously started through a unified control module. All data acquisitions are accurately timestamped, laying the foundation for subsequent time synchronization.

[0203] 2. Real-time Data Transmission and Preprocessing: Raw data collected by each monitoring module is transmitted to the central processing unit in real time via a unified interface. During data transmission, artifact removal and preliminary index calculations are performed in parallel. For example, NIRS data undergoes TDDR or spline interpolation processing, TCD / ONSD data is processed using image processing algorithms such as KCF tracking and GMM, and qEEG data undergoes artifact suppression via ICA or EMD.

[0204] 3. Timestamp Alignment and Fusion: Precise timestamp alignment is performed on the preprocessed data, and methods such as time amplitude cross-correlation are used to achieve accurate synchronization between different modalities. The synchronized data enters the multimodal data fusion module, where feature extraction and fusion are performed using deep learning models (such as CNNs and RNNs).

[0205] 4. Dynamic analysis and pattern recognition: The fused multimodal data is input into the AI / ML model for dynamic trend analysis and abnormal pattern recognition, and is correlated with the patient's clinical events (such as delirium, MODS, and prolonged mechanical ventilation time).

[0206] 5. Mechanism Discovery and Decision-Making Recommendations: The AI ​​model further analyzes the correlation between fused data and patient prognosis (such as mortality rate, length of hospital stay, and neurological function recovery). Through causal inference models, it uncovers the underlying pathophysiological mechanisms leading to brain dysfunction in critically ill patients. Finally, based on preset rules or learned patterns, the system generates personalized decision support recommendations to guide clinicians in optimizing treatment plans.

[0207] 3. Proof section

[0208] This section will provide specific embodiments, experiments, simulations, or positive experimental data demonstrating the inventiveness of the invention to verify its advantages in solving prior art problems, improving monitoring accuracy, and improving patient outcomes. The ultimate proof of the inventiveness and clinical value of this invention lies in demonstrating the synergistic effect of multimodal integration, which can bring insights and results unattainable through single monitoring methods. A core deficiency of existing solutions is that they are merely “a stack of three instruments,” lacking a “clear meaning of integration.” Therefore, this proof section must clearly demonstrate that the combination of NIRS, TCD / ONSD, and qEEG, after being fused and processed by the described AI algorithms, can provide a superior qualitative understanding of brain states and better quantitative patient outcomes than any single modality or its unintegrated use. This can be achieved by demonstrating improved diagnostic accuracy (e.g., higher AUC in complex diseases), earlier detection of key events, or a stronger correlation with long-term neurological function recovery. This will be the decisive evidence of its “inventiveness” and “clinical value.”

[0209] 3.1 Experimental Design and Data Collection

[0210] Study type: Prospective, multicenter, controlled clinical trials or retrospective cohort studies are recommended. Prospective studies can better assess the intervention's effectiveness, while multicenter studies can improve the generalizability of the results.

[0211] Study subjects: Recruit patients in the intensive care unit (ICU), especially those with acute brain injury (such as subarachnoid hemorrhage, intracranial hemorrhage, ischemic brain injury, traumatic brain injury, epilepsy, meningitis, encephalitis, etc.).

[0212] Data collection:

[0213] Baseline data: Collect patient demographic information, medical history, underlying diseases, Glasgow Coma Scale (GCS) scores on admission, etc.

[0214] Monitoring data: The brain oxygen saturation, TCD / ONSD, and quantitative electroencephalogram data of the "three-in-one" system of this invention are continuously collected and recorded synchronously with routine vital signs (heart rate, blood pressure, blood oxygen saturation, and end-tidal carbon dioxide ETCO2).

[0215] Clinical events and interventions: Record in detail the occurrence of patient complications (such as delirium, MODS, duration of mechanical ventilation, secondary brain injury, cerebral infarction, etc.), treatment interventions and their timing.

[0216] Prognostic data: Record short-term prognoses (such as length of hospital stay, length of stay in the ICU, and 90-day survival days to discharge DAOH90) and long-term prognoses (such as mortality rate, neuropsychiatric diagnosis, long-term sick leave, and income loss as proxy indicators of quality of life).

[0217] Data standardization: Emphasis is placed on using standards such as the Clinical Data Interchange Standards Consortium (CDISC) to standardize collected clinical data, ensuring data quality and interoperability, and laying the foundation for subsequent analysis.

[0218] 3.2 Experimental Results and Analysis

[0219] Single-modal performance verification:

[0220] NIRS: Validating the accuracy of indicators such as rSO2, THx, and TOx in early warning of cerebral hypoxia and assessment of brain autoregulation. For example, demonstrating the correlation between decreased rSO2 and postoperative neurocognitive dysfunction.

[0221] TCD / ONSD: Validate the effectiveness of MFV, PI, and RI in assessing cerebral blood flow perfusion, vascular resistance, and predicting vasospasm. Validate the accuracy of ONSD as a surrogate indicator for non-invasive ICP (compared to invasive ICP measurements or clinical diagnosis), for example, the high consistency between automated ONSD algorithms and manual measurements.

[0222] qEEG: Validate the sensitivity and specificity of qEEG (including aEEG) in assessing sedation depth, detecting abnormal discharges (such as epilepsy), and brain functional status.

[0223] Validation of the advantages of multimodal fusion:

[0224] Improved diagnostic accuracy: Comparing the accuracy, sensitivity, and specificity of monomodal monitoring with “triad” fusion monitoring in diagnosing specific brain dysfunctions (such as ischemia, edema, epilepsy, and excessive sedation) (e.g., using ROC curve analysis, AUC value).

[0225] Early warning capability: Demonstrates that the "trinity" system can detect trends of brain dysfunction or complications earlier than traditional methods.

[0226] Prognostic prediction capability: Demonstrates that AI / ML models based on multimodal fusion data have a better predictive ability for patient prognosis (such as mortality and neurological function recovery) than single indicators or clinical experience.

[0227] Decision support effectiveness: Evaluate the impact of the decision support suggestions provided by the system on the timeliness and effectiveness of clinicians' interventions.

[0228] Statistical analysis methods: Multivariate regression analysis, time series analysis, and machine learning classification / regression models (such as support vector machine SVM, random forest, and deep learning networks) are used to evaluate the correlation, predictive ability, and classification accuracy among various indicators.

[0229] 3.3 Simulation Verification

[0230] Physiological model construction: Using computational models to simulate neural, vascular and metabolic processes in the brain, such as multivariate vector autoregression (VAR) models or state-space models, to capture the dynamic relationships between complex physiological signals.

[0231] System response simulation: Simulate the response of the "three-in-one" system under different physiological and pathological conditions (such as hypoxia, intracranial pressure, vasospasm, etc.) to verify its robustness and accuracy in complex scenarios.

[0232] Algorithm optimization and validation: Test and optimize the performance of data processing, artifact removal, multimodal fusion and AI algorithms in a simulation environment, especially when real data is difficult to obtain or ethical restrictions exist.

[0233] Simulation validation is an important complement to clinical trials, especially considering the ethical and practical challenges of collecting real-world data from critically ill patients. It allows for controlled testing of complex interactions and algorithmic performance. Clinical trials in neurocritical care often yield negative results or have limitations in assessing neurological outcomes. Real-world data collection can be complex and involve ethical sensitivities. Physiological simulation models can capture the complex interrelationships between brain parameters. Simulation provides a controlled environment to test the system's response to various physiological perturbations, such as severe hypoxia and uncontrolled ICP spikes, which may be difficult or unethical to induce in human subjects. This allows for thorough validation of the theoretical utility of the algorithm and the "trinity" system before widespread clinical deployment, thus strengthening the proof-of-concept.

[0234] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0235] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A three-in-one non-invasive brain function monitoring system, characterized in that, include: Brain oxygen saturation monitoring module; The oxygen saturation in the patient's brain is monitored using a brain oxygenation monitor; Transcranial Doppler ultrasound and optic nerve sheath diameter measurement module: Uses a bedside portable ultrasound instrument to perform transcranial Doppler ultrasound monitoring and optic nerve sheath diameter measurement on patients; Quantitative EEG monitoring module; uses a bedside quantitative EEG monitor to monitor the patient's EEG.

2. A three-in-one non-invasive brain function monitoring method implementing the three-in-one non-invasive brain function monitoring system of claim 1, characterized in that, The three-in-one non-invasive brain function monitoring method includes the following steps: Step 1: Monitor the brain oxygen saturation and related indicators of critically ill patients; Step two: Perform transcranial Doppler ultrasound and optic nerve sheath diameter measurement; Step 3: Monitor the patient's quantitative electroencephalogram (EEG); Step four: Obtain various brain function monitoring indicators and their dynamic changes, and identify their correlation with complications; Step 5: By exploring the correlation between prognosis and disease outcome, we can uncover the possible underlying pathophysiological mechanisms of brain dysfunction in critically ill patients and construct a standardized management process based on non-invasive brain oxygenation, TCD / ONSD, and quantitative EEG.

3. The three-in-one non-invasive brain function monitoring method as described in claim 2, characterized in that, Step one specifically includes: Near-infrared spectroscopy is used to continuously monitor regional brain oxygen saturation in patients. The brain oxygenation status is assessed by measuring the absorption and scattering of near-infrared light of different wavelengths in brain tissue. The data acquisition device is a sensor placed on the patient's forehead. At the same time, relevant indices such as tissue hemoglobin index or total hemoglobin concentration are monitored. Artifact detection: based on signal gradient, heart rate variability, or standard difference constant; Artifact removal algorithm: Time derivative distribution repair: Corrects baseline drift and spike artifacts without requiring user parameters; Spline interpolation: used to correct baseline drift and long-term slow motion artifacts; Savitzky-Golay filtering or robust locally weighted regression smoothing: used to remove high-frequency spike artifacts; Principal Component Analysis / Target PCA: Identifies and removes artifact components by decomposing the signal; Adaptive filter: Uses a reference channel to remove motion artifacts and improve the signal-to-noise ratio; Wavelet filtering: used to remove noise and artifacts.

4. The three-in-one non-invasive brain function monitoring method as described in claim 2, characterized in that, Step two specifically includes: Two key measurements were performed on the patient using a bedside portable ultrasound device: The system measures blood flow velocity in major intracranial vessels by penetrating the skull with ultrasound; it provides flow velocity, pulsatility index, and displays spectral waveforms at specific points; TCD provides audible sound and spectral waveforms. Using a bedside portable ultrasound device, the diameter of the sheath around the optic nerve is measured by ultrasound; the measurement is usually performed 3 mm behind the eyeball; two measurements are taken at different planes in each eye.

5. The three-in-one non-invasive brain function monitoring method as described in claim 2, characterized in that, Step three specifically includes: The patient's brain activity is continuously monitored and analyzed using a bedside quantitative EEG monitor; electrodes are placed on the scalp following the international 10-20 system or its variants; the raw EEG signals are preprocessed after acquisition, including bad channel identification and labeling, and data filtering according to the required frequency band.

6. The three-in-one non-invasive brain function monitoring method as described in claim 2, characterized in that, Step four specifically includes: The system continuously acquires and records all brain oxygen saturation, TCD / ONSD, and quantitative electroencephalogram (EEG) indicators monitored in steps one, two, and three; the system tracks the dynamic trends of these indicators over time. Multimodal data integration and time synchronization: The system continuously acquires and records all indicators from brain oxygen saturation, TCD / ONSD and quantitative electroencephalography monitoring; the data integration framework adopts a real-time multimodal medical data processing framework to integrate different data sources, which can be expressed as F(D)=M(G(I),H(V),Q(P)), where G, H and Q are preprocessing functions and M is a machine learning model; Data standardization and feature engineering: Data standardization uses industry standards to standardize data elements, terminology, and coding systems, extracts useful features from various modalities, uses convolutional neural networks for image and video analysis, recurrent neural networks for sequence data, and autoencoders for dimensionality reduction and anomaly detection. Dynamic trend analysis: The system tracks the dynamic changes of various brain function monitoring indicators over time; through the analysis of continuous data, it promptly detects abnormal fluctuations in brain function indicators and correlates them with the evolution of the patient's condition and the occurrence and development of complications, thereby revealing potential pathogenic mechanisms and causes.

7. The three-in-one non-invasive brain function monitoring method as described in claim 2, characterized in that, Step five specifically includes: Advanced data analytics and machine learning applications: AI systems can quickly identify patterns in large amounts of medical data to assist clinical decision-making and predict the prognosis of patients with traumatic brain injury; deep learning algorithms can extract high-level abstract features from multidimensional patient data to improve diagnostic speed and accuracy; multimodal fusion deep learning can improve model accuracy and robustness by integrating multiple data sources. Discovery of pathophysiological mechanisms and causal inference: By using AI / ML to perform correlation analysis on multimodal data, we can uncover the underlying pathophysiological mechanisms leading to brain dysfunction in critically ill patients; causal inference models utilize unstructured data as proxy signals for unobserved confounding factors to improve the accuracy of causal effect estimation. Building a process-oriented management system: Based on three non-invasive monitoring technologies—brain oxygen saturation, TCD / ONSD, and quantitative electroencephalography—a standardized "three-in-one" management process system has been constructed. This system will guide clinicians on how to comprehensively utilize these three monitoring data for early diagnosis, risk assessment, optimization of treatment plans, and real-time evaluation of treatment effects. Ultimately, it will achieve non-invasive, real-time, dynamic, and systematic management of brain function in critically ill patients.

8. A computer device, characterized in that, The computer device includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the three-in-one non-invasive brain function monitoring method as described in any one of claims 2 to 7.

9. A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the three-in-one non-invasive brain function monitoring method as described in any one of claims 2 to 7.

10. An information data processing terminal, characterized in that, The information data processing terminal is used to implement the three-in-one non-invasive brain function monitoring system as described in claim 1.

Images