A spatial cognition digital drug effect evaluation method and device

Abstract

This invention discloses a method and apparatus for evaluating the efficacy of spatial cognitive digital drugs, comprising: conducting clinical assessments of drug users to obtain evaluation indicators for medical and nursing diagnoses; conducting spatial cognitive assessments of drug users, and simultaneously processing and analyzing the multimodal physiological data of drug users to obtain efficacy evaluation indicators for multimodal data analysis; collecting drug acceptance data from drug users to obtain their evaluation indicators; and comprehensively judging the effectiveness of the drug by combining the evaluation indicators for medical and nursing diagnoses, the efficacy evaluation indicators for multimodal data analysis, and the evaluation indicators for drug users. Subjective evaluation by patients can improve the credibility of the efficacy of spatial cognitive digital drugs, and clinical diagnosis of patients through medical and nursing channels, combined with multimodal physiological data analysis, can compensate for the current lack of objectivity in the efficacy evaluation of spatial cognitive digital drugs.

Description

Technical Field

[0001] This invention relates to the field of multimodal brain-computer interface technology, and in particular to a method and device for evaluating the effects of digital drugs in spatial cognition. Background Technology

[0002] Brain-computer interfaces (BCIs) measure the central nervous system and translate neural activity into a series of artificial commands, thereby replacing, restoring, supplementing, or enhancing the output of the central nervous system. BCI systems establish novel information interaction and control pathways between the central nervous system and its internal and external environment, achieving bidirectional interaction and real-time feedback between the brain and external devices without relying on conventional spinal cord or peripheral neuromuscular pathways. The initial motivation for BCI research was to help paralyzed and disabled individuals regain the ability to communicate with the outside world. Currently, the vast majority of BCI startups also focus on the healthcare field. It can be said that the medical field is one of the most important application areas for BCIs. The combination of digital intelligent technology with novel sensing systems and AI algorithms has opened up new avenues for digital intervention in life and health, improving disease prevention and treatment. Mobile technology, remote communication technology, smart devices, wearable devices, health information analysis technology, and personalized medicine can help patients understand their health status in a timely manner, and enable doctors to accurately understand patients' conditions and implement precise interventions. This emerging phenomenon in the field of healthcare information technology is called "digital therapy."

[0003] Based on digital therapy technologies, the core of digital medicine is to use software to intervene in and treat diseases. It can be used alone or in combination with drugs or devices.

[0004] The evaluation of the efficacy of digital drugs differs significantly from that of traditional drugs. Taking traditional small-molecule drugs as an example, there is extensive research on their active ingredients, chemical structures, mechanisms of action after entering the body, and pharmacokinetic aspects related to metabolism. Various diagnostic indicators, clinical gold standards, and improvements in clinical symptoms provide a good basis for assessing drug effectiveness and side effects. However, the efficacy of digital drugs is not as readily observable and tangible as the various diagnostic indicators used for traditional drugs, and potential side effects and harms are difficult to predict and assess. For instance, in the case of digital therapy products for mental health, treatment effectiveness in patients with depression is typically measured using clinical scales, with changes in scale scores indicating drug efficacy. However, this approach is highly subjective, and the thinking and state of depressed patients are inherently uncertain. Compared to relying on indicators and data for traditional small-molecule drugs, this approach introduces a degree of uncontrollability and carries many unknown long-term risks.

[0005] Currently, the lack of unified industry standards for evaluating the efficacy of digital drugs hinders their subsequent promotion. Digital therapies are a new field requiring understanding and acceptance from physicians. The lack of standards slows down market development during physician training and market cultivation. For example, in some popular diseases, there are numerous digital therapy products, and inconsistent clinical data cannot be compared. In existing technologies, physicians rely entirely on their clinical experience to judge patients' conditions, selecting evaluation areas based on prior knowledge for diagnosis. This approach easily overlooks brain changes outside the study area. Because it heavily relies on physicians' clinical experience and subjective judgment, it is prone to misdiagnosis. Therefore, existing technologies suffer from insufficient objectivity, controllability, and inconsistent evaluation standards in evaluating the efficacy of digital drugs based on spatial cognition. Summary of the Invention

[0006] This invention provides a method and apparatus for drug efficacy analysis. Addressing the shortcomings of current spatial cognitive digital drug efficacy evaluation methods, such as insufficient objectivity, lack of controllability, and inconsistent evaluation standards, this method comprehensively evaluates the efficacy of digital drugs from three dimensions: medical diagnosis, multimodal data analysis, and patient evaluation, based on scientifically objective multimodal clinical physiological data. The technical solution is as follows:

[0007] This invention provides a method for evaluating the effectiveness of digital drugs based on spatial cognition, comprising:

[0008] Clinical assessments were conducted on medication users to obtain evaluation indicators for medical and nursing diagnoses.

[0009] Spatial cognition was assessed on medication users, and multimodal physiological data of medication users were processed and analyzed to obtain efficacy evaluation indicators from multimodal data analysis;

[0010] The drug acceptance level of the medication users was collected to obtain the evaluation indicators of the medication users;

[0011] The effectiveness of the drug is determined by combining the evaluation indicators of medical diagnosis, the efficacy evaluation indicators of multimodal data analysis, and the evaluation indicators of the drug users.

[0012] Preferably, the evaluation indicators for clinical assessment of medication users to obtain medical diagnoses include:

[0013] S101: Collect the Guildford-Zimmerman Spatial Orientation Test Score A1, the Perspective Spatial Positioning Test Score B1, and the Corsi Block Tapping Task Score C1 of the drug user before medication, and calculate the total score of the scales T1 of the drug user before medication, where T1=A1+B1+C1.

[0014] S102: Collect the Guildford-Zimmerman Spatial Orientation Test Score A2, the Perspective Spatial Positioning Test Score B2, and the Corsi Block Tapping Task Score C2 of the drug user after medication, and calculate the total score of the scales T2 after medication, where T2 = A2 + B2 + C2.

[0015] S103: Calculate the evaluation index Z1 of the medical diagnosis, where Z1 = T2 - T1.

[0016] Preferably, spatial cognition assessment is performed on medication users, and multimodal physiological data of medication users are processed and analyzed to obtain efficacy evaluation indicators for multimodal data analysis, including:

[0017] S201: When the patient performs a spatial cognition assessment task before taking the medication, the portable device simultaneously collects the patient's physiological data before taking the medication.

[0018] S202: When the user performs a spatial cognition assessment task after taking the medication, the portable device simultaneously collects the user's physiological data after taking the medication;

[0019] S203: Perform data preprocessing, feature extraction, feature fusion, and data classification on the physiological data collected in step S201 and step S202 in sequence.

[0020] Preferably, the physiological data includes electroencephalogram (EEG) signal data, electrooculogram (EOG) signal data, and eye movement data;

[0021] The data preprocessing of the physiological data collected in step S201 and step S202 includes: drift data removal, bandpass filtering, artifact removal, and baseline correction.

[0022] The data preprocessing methods for eye movement data before and after medication include: outlier removal method, missing value handling method, and filtering and noise reduction method.

[0023] Data preprocessing methods for electrooculogram (EOG) signal data before and after medication include: filtering and noise reduction.

[0024] Preferably, the method for feature extraction of the physiological data collected in step S201 and the physiological data collected in step S202 includes: extracting the coupling features of the EEG signal using the sorting conditional mutual information method;

[0025] The feature extraction methods for the eye movement data before and after medication include: extracting eye movement behavior and extracting pupil diameter;

[0026] The feature extraction method for the electrooculogram (EOG) signal data before and after medication includes: calculating the low-to-high frequency average power spectrum ratio.

[0027] Preferably, the method for feature fusion of the physiological data collected in step S201 and the physiological data collected in step S202 includes: fusing features of different modalities through a phased feature splicing method, wherein the features to be fused are the features obtained through feature extraction.

[0028] Preferably, the data classification method includes: classifying the features of the fused data using a binary classification method.

[0029] Preferably, the method of collecting data on the drug acceptance level of medication users to obtain the evaluation indicators for the medication users includes:

[0030] A questionnaire survey was conducted among the medication users, and the scores from the questionnaire survey were used as evaluation indicators for the medication users.

[0031] Preferably, the questionnaire includes the following:

[0032] The evaluation criteria included: the patient's experience and comfort with digital medication; the safety of digital medication; the therapeutic effect of digital medication; the portability of digital medication; and the level of acceptance of the new treatment method.

[0033] Each item is scored out of 10 points.

[0034] A digital drug analysis device, comprising:

[0035] Clinical assessment module: Conduct clinical assessments of medication users to obtain evaluation indicators for medical and nursing diagnoses;

[0036] Multimodal data assessment module: It assesses the spatial cognition of drug users and processes and analyzes their multimodal physiological data to obtain efficacy evaluation indicators from multimodal data analysis;

[0037] Patient assessment module: Collects data on the medication acceptance level of patients and obtains evaluation indicators for these patients;

[0038] Comprehensive analysis module: Receives the evaluation indicators of the medical diagnosis, the efficacy evaluation indicators of the multimodal data analysis, and the evaluation indicators of the medication user, and determines whether the medication used by the medication user is effective.

[0039] The beneficial effects of the technical solutions provided by the embodiments of the present invention include at least the following:

[0040] This invention employs a multi-dimensional evaluation system. Patient subjective evaluation enhances the credibility of spatial cognitive digital drug efficacy assessments. Clinical diagnosis by healthcare professionals incorporates traditional clinical methods, addressing the current lack of objectivity in spatial cognitive digital drug efficacy evaluations. This invention focuses on scientifically objective clinical multimodal physiological data analysis, improving both the objectivity and credibility of spatial cognitive digital drug efficacy evaluations. The multi-dimensional comprehensive evaluation system significantly enhances the controllability of spatial cognitive digital drug efficacy evaluations, providing a reliable basis for efficacy evaluation in the current field.

[0041] Healthcare professionals are the primary caregivers for patients with spatial cognitive impairment, playing a supportive and supervisory role during their treatment. Digital medicine, as a medical aid, aims to simplify traditional treatments, improve convenience and effectiveness, and reduce the burden on healthcare professionals. Therefore, acceptance and recognition from healthcare professionals are essential prerequisites for the widespread adoption of digital medicine. Based on these considerations, we incorporate current traditional clinical assessment methods for spatial cognitive impairment into our digital medicine evaluation scheme to enhance the objectivity of digital medicine evaluation.

[0042] Before and after digital drug therapy, traditional clinical assessments of patients with spatial cognitive impairment typically use scales to measure their spatial cognitive abilities. Therefore, in the medical and nursing diagnostic section, commonly used clinical scales such as the Guildford-Zimmerman Spatial Orientation Test, the Perspective-Based Spatial Positioning Test, and the Corsi Block Tagging Task Scale were employed to assess changes in spatial cognitive abilities in patients with spatial cognitive impairment before and after digital drug therapy. These scales include tests related to spatial cognitive abilities, involving factors such as spatial memory, spatial perspective, and spatial positioning, allowing for a more comprehensive assessment of spatial cognitive abilities. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart of the spatial cognition digital drug efficacy evaluation method provided in an embodiment of the present invention;

[0045] Figure 2 This is a schematic diagram of the spatial cognition digital drug efficacy evaluation device provided in an embodiment of the present invention;

[0046] Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0047] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0048] This invention is based on scientific and objective multimodal clinical physiological data, and comprehensively designs a spatial cognition digital drug efficacy evaluation method from three dimensions: medical diagnosis, multimodal data analysis, and patient evaluation.

[0049] like Figure 1 As shown, this invention provides a method for evaluating the effectiveness of digital drugs based on spatial cognition, comprising the following steps:

[0050] S1. Conduct clinical assessments of medication users to obtain evaluation indicators for medical and nursing diagnoses.

[0051] S2. Conduct spatial cognition assessment on medication users, and process and analyze the multimodal physiological data of medication users to obtain efficacy evaluation indicators from multimodal data analysis;

[0052] S3. Collect data on the drug acceptance level of drug users to obtain their evaluation indicators;

[0053] S4. A comprehensive analysis is conducted, combining the evaluation indicators of the medical and nursing diagnoses, the efficacy evaluation indicators of the multimodal data analysis, and the evaluation indicators of the medication users, to determine the effectiveness of the medication.

[0054] Steps S1 through S3 can be executed sequentially, in a random order as needed (e.g., in the order of S2, S1, and S3), or simultaneously. Healthcare professionals are the primary caregivers for patients with spatial cognitive impairment, playing a supportive and supervisory role during their treatment. Digital medicine, as a medical aid, aims to simplify traditional treatments, improve convenience and effectiveness, and reduce the burden on healthcare professionals. Therefore, acceptance and recognition from healthcare professionals are essential prerequisites for its widespread adoption. Based on these considerations, we incorporate current traditional clinical assessment methods for spatial cognitive impairment into our digital medicine evaluation scheme to enhance the objectivity of digital medicine evaluation.

[0055] Among them, S1, clinical assessment of medication users, and evaluation indicators for obtaining medical and nursing diagnoses include:

[0056] S101: Collect the Guildford-Zimmerman Spatial Orientation Test Score A1, the Perspective Spatial Positioning Test Score B1, and the Corsi Block Tapping Task Score C1 of the drug user before drug administration, and calculate the total score of the scales before drug administration T1, where T1=A1+B1+C1.

[0057] S102: Collect the Guildford-Zimmerman Spatial Orientation Test Score A2, the Perspective Spatial Positioning Test Score B2, and the Corsi Block Tapping Task Score C2 of the drug user after medication administration, and calculate the total score of the scales T2 after medication administration, where T2=A2+B2+C2;

[0058] S103: Calculate the evaluation index Z1 for medical diagnosis, where Z1 = T2 - T1.

[0059] Specifically, before and after patients with spatial cognitive impairment received digital drug therapy, they were assessed using a scale. Higher scale scores indicated stronger spatial cognitive abilities, and the changes in scores between the two assessments reflected changes in the patient's spatial cognitive abilities. The difference between the sum of the three scale scores after treatment and the total score before treatment was used as an evaluation indicator for medical diagnosis.

[0060] Before and after digital drug therapy, traditional clinical assessments of patients with spatial cognitive impairment typically use scales to measure their spatial cognitive abilities. Therefore, in the medical and nursing diagnostic section, commonly used clinical scales such as the Guildford-Zimmerman Spatial Orientation Test, the Perspective-Based Spatial Positioning Test, and the Corsi Block Tagging Task Scale were employed to assess changes in spatial cognitive abilities in patients with spatial cognitive impairment before and after digital drug therapy. These scales include tests related to spatial cognitive abilities, involving factors such as spatial memory, spatial perspective, and spatial positioning, allowing for a more comprehensive assessment of spatial cognitive abilities.

[0061] This invention uses multimodal physiological data analysis as the core of digital drug efficacy evaluation. Before and after digital drug treatment, patients with spatial cognitive impairment undergo spatial cognitive assessment tasks through a spatial cognitive assessment system. During the spatial cognitive assessment tasks, the system simultaneously collects the patient's electroencephalogram (EEG), electrooculogram (EOG), and eye movement (EMG) signals using a portable device, and stores the collected multimodal data locally for subsequent processing.

[0062] Based on this, S2, spatial cognition assessment of medication users is conducted, and multimodal physiological data of medication users is processed and analyzed to obtain efficacy evaluation indicators from multimodal data analysis. Specific operations include:

[0063] S201: When a patient is performing a spatial cognitive assessment task before taking medication, a portable device simultaneously collects the patient's physiological data before medication.

[0064] S202: When a patient performs a spatial cognitive assessment task after taking medication, a portable device simultaneously collects the patient's physiological data after taking medication.

[0065] S203: Perform data preprocessing, feature extraction, feature fusion, and data classification on the physiological data collected in step S201 and step S202 in sequence.

[0066] The system used for spatial cognition assessment of medication users is called the Spatial Cognition Assessment System. This system is a hybrid system integrating numerous hardware and software components. The hardware includes: an OpenBCI EEG signal acquisition device, an EEG signal acquisition device, an HTC VivePro virtual reality headset, and a desktop PC. The OpenBCI device uses 16 semi-dry electrodes with an impedance of less than 10kΩ, sampling at 1000Hz to acquire EEG signals from subjects performing spatial cognition assessment tasks. The data is transmitted in real-time to the PC via the OpenBCI's built-in Wi-Fi signal source. The HTC VivePro displays the virtual reality assessment scene, with a resolution of 2880*1600 for each eye and a refresh rate of 90Hz, providing a high-definition and smooth display. The HTC VivePro integrates a Tobii eye-tracking data acquisition device, capable of acquiring eye movement information such as pupil diameter, gaze direction, and eye closure degree at a sampling rate of 120Hz. To avoid interference from the VR headset on the acquisition of electrooculogram (EOG) signals, we used the Fp2 electrodes of the OpenBCI device to acquire the EOG signals. The desktop PC has a 10th-generation i5 processor, 8GB of RAM, a GTX 1650 dedicated graphics card, and other hardware to support the operation of various software.

[0067] The software portion of this system relies on programming languages ​​such as C# and Python, as well as code editing and runtime environments such as Unity3D, Visual Studio Code, and Anaconda. The C# programming language is used with Unity3D to construct the scene for the virtual reality spatial cognition assessment task. The Visual Studio Code development environment utilizes Python to analyze and process offline multimodal data, including EEG signals, eye-tracking signals, and electrooculogram signals, and is also used to construct the user interface for the spatial cognition assessment system.

[0068] The spatial cognition assessment system integrates spatial cognition assessment tasks, multimodal data acquisition and storage, and multimodal data analysis and processing. During the patient's spatial cognition assessment, multimodal physiological data is collected using a portable device. The OpenBCI device integrates a Wi-Fi signal source, allowing a desktop PC to connect to the OpenBCI's Wi-Fi signal for data communication. An eye tracker is integrated into the VR headset; the collected eye-tracking data is transmitted to the data acquisition unit via serial communication for data parsing, and finally, the collected multimodal data is stored on the computer's local disk. Finally, machine learning and deep learning algorithms are used to process the multimodal data.

[0069] Spatial cognition assessment tasks are mostly assessment games developed based on virtual reality technology, designed to evaluate changes in patients' spatial cognitive abilities before and after receiving digital drug treatment. The development and implementation process of spatial cognition assessment tasks will be illustrated below with specific examples.

[0070] To minimize the impact of external factors on spatial cognitive assessment, the spatial cognitive assessment task was presented in virtual reality. Leveraging the immersive experience and high degree of freedom and controllability of VR scenarios, participants could focus less on the task itself and less on distractions. The scenario was constructed using the Unity3D game engine. To assess whether participants' spatial cognitive abilities changed before and after digital drug treatment, a scenario such as Virtual City Walking (VCW) was used for testing. Specifically, a spatial cognitive ability assessment was conducted before and after digital drug treatment. Each assessment consisted of two steps: a memorization phase and a repetition phase. The memorization phase involved participants learning and memorizing the surrounding environment and landmarks at intersections while walking in the virtual city, aiding in route memorization. The repetition phase involved participants retracing the pre-arranged route in the virtual city using their spatial memory and sense of direction.

[0071] During the spatial cognition assessment task, multimodal data of patients are collected simultaneously. The following will detail the methods for collecting and synchronizing multimodal data, including steps 1 and 2:

[0072] Operation 1: EEG and EEG Signal Acquisition Method: This system uses an OpenBCI EEG acquisition device with a Wi-Fi module for data acquisition. During acquisition, the impedance of each channel must be below 10kΩ. The real-time acquired EEG and EEG signals are transmitted via Wi-Fi and then stored on a desktop PC. The system uses wet and semi-dry electrodes to acquire data from 16 brain regions. The electrode placement method follows the international standard lead 10-20 standard for electrode positioning. The brain regions acquired include: Fp1, Fp2, F7, F8, F3, F4, Fz, FCz, C3, C4, Cz, P7, P8, Pz, O1, and O2. Electrode Fp2 is used to acquire EEG signals, and the areas at the bilateral earlobes are used as reference electrodes.

[0073] Operation 2: Eye Tracking Signal Acquisition Method: Eye tracking data acquisition is primarily accomplished by the eye tracker built into the HTC VIVEPro. The eye tracker provides up to 120 eye movement data points per second, with a measurement accuracy between 0.5° and 1.1°, and supports data acquisition from a 110° field of view. Before data acquisition, the eye tracker needs to be calibrated. Specifically, locate the eye tracking data button in the SteamVR control panel. The steps are as follows: Connect the VIVEProEye device to your desktop PC via cable; connect any VR controller; wear the head-mounted display; press the system button on the VR controller to open the control panel; select VIVEProEye from the SteamVR control panel; ensure the eye tracking data function is enabled, focus the VR controller on the calibration button, and press the trigger button to enter the calibration interface; adjust the head-mounted display height and interpupillary distance according to the on-screen prompts; follow the movement of the dots on the screen with your eyes; after calibration, press the system button to close the main control panel.

[0074] The eye-tracking acquisition system was built within the Unity3D environment using the SRaipal software development kit provided by HTC. The data collected by the eye tracker includes information such as 3D gaze direction, pupil diameter, pupil position, and eye opening degree. The coordinate axes for the gaze direction data are centered between the eyes, with the positive z-axis perpendicular to the screen forward, the positive x-axis to the left, and the positive y-axis upward, and the values ​​are normalized to between -1 and 1. Pupil position data is normalized to between 0 and 1.

[0075] The following example illustrates a method for synchronizing multimodal data: To ensure time alignment of data from different modalities, data acquisition can be initiated and terminated using an event-driven approach. The data acquisition module creates an independent thread for each modality and enters a standby state, awaiting the event triggered by the "Start Acquisition" button. Clicking the "Start Acquisition" button triggers the start acquisition event, and all standby data acquisition threads simultaneously begin recording EEG, Eoptometry, and Eye Movement (EMG) data. The acquired data includes timestamps accurate to milliseconds, and each data point is separated by a specific character and stored in a CSV file on the local disk.

[0076] Physiological data includes electroencephalogram (EEG) signal data, electrooculogram (EOG) signal data, and eye movement (EMT) data. Data preprocessing for the physiological data collected in steps S201 and S202 includes: drift data removal, bandpass filtering, artifact removal, and baseline correction. Preprocessing methods for EMT data before and after medication include: outlier removal, missing value handling, and filtering / denoising. Preprocessing methods for EMT signal data before and after medication include: filtering / denoising.

[0077] The method for feature extraction of the physiological data collected in step S201 and step S202 includes: extracting the coupling features of EEG signals using the sorting conditional mutual information method;

[0078] The feature extraction methods for pre- and post-medication eye movement data include: extracting eye movement behavior and extracting pupil diameter; the feature extraction methods for pre- and post-medication electrooculogram (EOG) signal data include: calculating the low-to-high frequency average power spectrum ratio. The method for feature fusion of the physiological data collected in step S201 and step S202 includes: fusing features from different modalities through a staged feature stitching method, wherein the features being fused are those obtained through feature extraction.

[0079] Data classification methods include: classifying the features of fused data using a binary classification method.

[0080] After completing the acquisition and storage of multimodal data, the following section will detail the methods for processing multimodal data. The multimodal data processing workflow includes: data preprocessing, feature extraction, feature fusion, and data classification.

[0081] Data preprocessing:

[0082] EEG signal preprocessing: To reduce the interference of noise in EEG signals on the experimental results, the MNE library in the Python programming language was used for data preprocessing. This mainly included:

[0083] ① Remove drift data. Use MNE to load all acquired EEG signals into memory and use its built-in functions to plot the EEG signals. If a significant offset is found in the array range of the EEG signal waveform, this part of the data needs to be removed based on channel information and timestamp information.

[0084] ② Bandpass filtering. Based on previous research, 40Hz is the highest frequency of EEG signals related to spatial cognitive behavior, and data below 1Hz is noise data that is prone to data drift and needs to be filtered out. Therefore, a bandpass filter of 1–40Hz is used to filter the EEG signals.

[0085] ③ Artifact Removal. During the experiment, subjects are prone to head movements, which can cause friction between the electrodes and the scalp, resulting in additional motion artifacts in the EEG signal. Furthermore, EEG signals are easily interfered with by signals from electrooculography (EOG) and electromyography (EMG). For this type of interference data, the Independent Component Correlation Algorithm (PCA) built into the MNE can be used to remove interfering components from the EEG signal, improving the signal-to-noise ratio.

[0086] ④ Baseline correction. During data acquisition, the subject's EEG is also affected by spontaneous noise. Therefore, the resting-state EEG can be used as a baseline, and the mean of the resting-state EEG signal can be subtracted from the data of each channel to reduce noise and data drift.

[0087] Electrooculogram (EOG) signal preprocessing: EOG signals are easily affected by the external environment and are weak signals with no obvious pattern, thus limiting their time-domain analysis. Frequency-domain analysis methods are widely used in EOG signal preprocessing, such as Fast Fourier Transform (FFT), power spectrum estimation, and bispectral estimation. We filter the acquired raw EOG signals to remove glitches and weak noise, and simultaneously perform frequency-domain analysis using FFT.

[0088] Eye-tracking data preprocessing: The acquired eye-tracking data needs to be processed to reduce the impact of noise on the analysis results. This mainly includes:

[0089] ① Abnormal Sample Removal. During eye-tracking data acquisition, spontaneous and involuntary eye-closing behaviors can lead to abnormal data collection. During eye-closing, the eye tracker may misjudge data such as pupil size and gaze direction, and may be unable to collect normal data, resulting in empty data sets and missing values. When prolonged eye-closing causes the percentage of missing data in a sample to exceed a certain threshold, the sample is considered invalid. This paper removes samples with a missing data percentage greater than 25% to avoid noise affecting the experimental results.

[0090] ② Missing Value Handling. After filtering outlier samples, some data may still be missing due to rapid blinking. In this case, missing data needs to be imputed. Taking pupil size as an example, the collected pupil size data is empty after the eyes are closed. Therefore, methods need to be used to impute missing values, reducing data bias and improving data quality. There are many methods for imputing missing values, including mean interpolation, forward and backward interpolation, linear interpolation, and polynomial interpolation. We use polynomial imputation to fill in the missing data.

[0091] ③ Filtering and Denoising. After processing for missing values, the data largely recreates the true eye movement behavior. However, due to factors such as blinking, saccades, microsaccades, and changes in lighting, the data inevitably contains many artifacts, which still need to be filtered out. For pupil diameter data, a Savitsky-Golay filter is used to smooth high-frequency spikes while preserving extreme points and edge information. For gaze coordinate information, which is based on eye movements, the raw gaze coordinate data is affected by various factors. Among them, saccades and microsaccades have a significant impact on the gaze point, causing artifacts in the eye movement coordinate data and thus affecting the extraction of eye movement behavior information. Kalman filtering can be used to reduce interference from noise.

[0092] After preprocessing the multimodal data, EEG features, electrooculogram (EOG) features, and eye movement (EMG) features related to spatial cognitive ability are extracted using feature extraction algorithms. The specific implementation method is as follows:

[0093] EEG signal feature extraction: The Permutation Conditional Mutual Information (PCMI) method is used to extract the coupling features of EEG signals. Specifically, for EEG signals, two time series X and Y from different EEG channels are selected, and the two sequences have the same number of sampling points N.

[0094] Electroocular signal feature extraction: The blink frequency feature is extracted by calculating the number of blinks per unit time through peak detection. The spectrum is obtained by fast Fourier transform of the filtered electroocular signal. Based on previous verification and extensive experimental analysis, 0-2.5Hz is selected as the low frequency band and 2.5-10Hz as the high frequency band. The ratio of the average power spectrum of low and high frequencies is calculated as the electroocular feature.

[0095] Eye movement signal feature extraction:

[0096] Feature extraction based on eye movement behavior: Common eye movement behaviors include blinking, gazing, and saccades. Blink frequency and blink speed reflect cognitive load. Gazing indicates that the subject's gaze remains fixed on an object for a period of time, indicating relatively concentrated attention and active thinking. Saccades, on the other hand, involve the gaze moving rapidly between different objects, indicating less concentrated attention. Therefore, representative features can be extracted based on blinking, gazing, and saccades. Blinking is automatically acquired by eye-tracking equipment, while gazing and saccades require analysis based on the user's fixation coordinates. After obtaining blinking, gazing, and saccades, features such as blink duration, gazing duration, saccade duration, and saccade amplitude can be extracted. Furthermore, statistical features such as maximum, minimum, mean, and variance within a specific time period can be extracted. When a user engages in different eye movement behaviors due to thought processes, the amplitude of changes in eye fixation coordinates varies; therefore, the amplitude and rate of change of fixation coordinates, as well as statistical features of these features within a specific time period, can be extracted.

[0097] Feature extraction based on pupil diameter: During cognitive activities, pupil size changes with cognitive load. Researchers have found that in reading, pupil diameter is larger when gazing under high cognitive load and smaller when cognitive load is low. When an object requires more attention, the pupil needs to dilate to acquire more information more quickly. Therefore, changes in pupil diameter can serve as an indicator of the intensity of mental processing; that is, changes in pupil diameter can reflect the current cognitive load. Features extracted based on changes in pupil diameter have strong discriminative power in cognitive processes. Common features based on pupil diameter include: pupil size during gazing, amplitude of pupil change, speed of pupil change, duration of pupil change, number of pupil dilations, number of pupil constrictions, and their statistical values ​​under multiple stimulus conditions, such as maximum, minimum, and mean values.

[0098] After feature extraction from EEG, EOS, and eye-tracking data, a feature fusion algorithm is used to fuse the extracted multimodal features. The fusion order is determined by calculating the correlation coefficients between pairs of modalities; features with high correlation coefficients are fused first, followed by fusion of the fused features with the third type of feature. Taking the fusion of EEG and eye-tracking features as an example, a classic cascading approach can be used to concatenate the EEG and eye-tracking feature vectors to form a total feature vector. Alternatively, an attention-based feature fusion algorithm can be used, assigning different weights to different modalities to improve the fusion effect.

[0099] A feature fusion algorithm is used to obtain fused multimodal features. Based on these fused multimodal features, the changes in the patient's spatial cognitive ability are assessed. The specific assessment approach involves using data mining and deep learning methods to perform binary classification on assessment data before and after digital drug treatment. High classification accuracy indicates a high degree of discriminatory power between the pre-test and post-test data, suggesting a significant change in spatial cognitive ability. The specific classification method is illustrated below: An XGBoost ensemble learning classifier is used for binary classification of multimodal data before and after digital drug treatment. The fused multimodal feature data is used as input to the XGBoost classifier to construct a classification model. The data is divided into training and test sets in a 9:1 ratio. A grid search method is used to optimize the model parameters. K-fold cross-validation is used to train the model on the training set. After parameter tuning, the model is used to predict on the test set, and its ability is evaluated. K-fold cross-validation can be used to prevent model overfitting and improve the model's generalization ability. For example, if K is set to 5, the training set is divided into 5 folds and trained 5 times. In each training, the model learns 80% of the data and predicts 20% of the data. Finally, the average of the 5 validation prediction results is used to obtain the final classification accuracy. The classification accuracy is used as an indicator for evaluating the efficacy of spatial cognitive digital drugs.

[0100] S3. Collect data on drug acceptance levels among medication users to obtain user evaluation indicators; the user evaluation indicators for drug acceptance levels include:

[0101] A questionnaire survey was conducted among medication users, and the survey scores were used as evaluation indicators for medication users.

[0102] like Figure 3 As shown, the questionnaire includes the following:

[0103] The evaluation criteria include: the patient's experience and comfort with digital drugs; the safety of digital drugs; the therapeutic effect of digital drugs; the portability of digital drugs; and the level of acceptance of new treatment methods. Each criterion is scored out of 10.

[0104] S4. Determining the effectiveness of a drug by combining evaluation indicators from medical and nursing diagnoses, efficacy evaluation indicators from multimodal data analysis, and evaluation indicators from medication users includes:

[0105] Patients with spatial cognitive impairment, as users of digital medications, participate fully in the treatment process and are direct beneficiaries of digital medications, possessing firsthand experience regarding their safety, convenience, and effectiveness. Based on subjective evaluations of digital medications for spatial cognitive impairment, a questionnaire survey was conducted during patient treatment to understand their acceptance and level of recognition of the digital medications. Patient scores on the questionnaires were used as evaluation indicators for the medication administrators.

[0106] Finally, the efficacy evaluations derived from three different dimensions—medical diagnosis, patient feedback, and multimodal physiological data analysis—were integrated. The evaluation index values ​​from these three dimensions showed a significant correlation with the patient's spatial cognitive ability. This invention uses multimodal physiological data analysis as its core. Due to the strong objectivity of clinical physiological data, its analysis results are used as the primary evaluation index. The scale, being the gold standard for clinical diagnosis, is used as an important reference, while the patient's subjective evaluation serves as an auxiliary and secondary reference. Based on the efficacy evaluations from these three dimensions, a comprehensive evaluation report on the effectiveness of the spatial cognitive digital drug is generated. This report includes evaluation index values ​​from all three dimensions, and the magnitude of each index value reflects the changes in the patient's spatial cognitive ability after receiving digital drug treatment, thereby verifying the effectiveness of the spatial cognitive digital drug.

[0107] like Figure 2 As shown, a digital drug analysis device is provided, including: a clinical assessment module 100, a patient assessment module 200, a multimodal data assessment module 300, and a comprehensive analysis module 400, wherein the clinical assessment module 100, the patient assessment module 200, and the multimodal data assessment module 300 are respectively connected to the comprehensive analysis module 400.

[0108] Clinical Assessment Module 100: Conduct clinical assessments on medication users to obtain evaluation indicators for medical and nursing diagnoses. Collect the Guildford-Zimmerman Spatial Orientation Test (A1), the Transcranial Doctrine Test (B1), and the Corsi Block Tapping Task (C1) scores of the medication users before medication use, and calculate the total scale scores T1 before medication use, where T1 = A1 + B1 + C1; collect the Guildford-Zimmerman Spatial Orientation Test (A2), the Transcranial Doctrine Test (B2), and the Corsi Block Tapping Task (C2) scores of the medication users after medication use, and calculate the total scale scores T2 after medication use, where T2 = A2 + B2 + C2;

[0109] The evaluation index Z1 for medical and nursing diagnosis was calculated, where Z1 = T2 - T1. Specifically, patients with spatial cognitive impairment were assessed using a scale before and after receiving digital drug therapy. Higher scale scores indicate stronger spatial cognitive abilities, and the change in scores between the two assessments reflects the change in the patient's spatial cognitive abilities. The difference between the sum of the three scale scores after treatment and the total score before treatment was used as the evaluation index for medical and nursing diagnosis.

[0110] Multimodal Data Assessment Module 300: This module assesses the spatial cognition of medication users and processes and analyzes their multimodal physiological data to obtain efficacy evaluation indicators. The questionnaire includes ratings for: the user's experience and comfort with digital medication, the safety of digital medication, the therapeutic effect of digital medication, the portability of digital medication, and the level of acceptance of the new treatment method; each rating is scored out of 10.

[0111] Patients with spatial cognitive impairment, as users of digital medications, participate fully in the treatment process and are direct beneficiaries of digital medications, possessing firsthand experience regarding their safety, convenience, and effectiveness. Based on subjective evaluations of digital medications for spatial cognitive impairment, a questionnaire survey was conducted during patient treatment to understand their acceptance and level of recognition of the digital medications. Patient scores on the questionnaires were used as evaluation indicators for the medication administrators.

[0112] Patient assessment module 200: Collects data on the patient's acceptance of the medication, obtaining evaluation indicators. Before medication administration, a portable device simultaneously collects the patient's physiological data during a spatial cognitive assessment task; after medication administration, the portable device simultaneously collects the patient's physiological data during a similar task. The physiological data collected in steps S201 and S202 are sequentially preprocessed, feature-extracted, feature-fused, and classified. Based on the classification accuracy, this accuracy is used as an indicator of the efficacy of the spatial cognitive digital drug.

[0113] Comprehensive Analysis Module 400: Receives evaluation indicators from medical and nursing diagnoses, efficacy evaluation indicators from multimodal data analysis, and evaluation indicators from medication users to determine whether the medication used by the users is effective.

[0114] This invention uses multimodal physiological data analysis as its core. Due to the strong objectivity of clinical physiological data, its analysis results are used as the primary evaluation indicator. Rating scales, considered the gold standard for clinical diagnosis, are used as an important reference, while patient subjective evaluations serve as supplementary and secondary references. The evaluation indicators from medical and nursing diagnoses, efficacy evaluation indicators from multimodal data analysis, and evaluation indicators from medication users are provided to doctors for reference, assisting them in making more accurate judgments about patients' conditions and reducing misdiagnosis rates. If doctors can determine that a patient's condition has improved based on the values ​​of these indicators, then the spatial cognitive digital therapy is proven to be effective.

[0115] Figure 3 This is a schematic diagram of the structure of an electronic device 600 provided in an embodiment of the present invention. The electronic device 600 may vary considerably due to different configurations or performance. It may include one or more central processing units (CPUs) 601 and one or more memories 602. The memory 602 stores at least one instruction, which is loaded and executed by the processor 601 to implement the steps of the above-mentioned spatial cognition digital drug effect evaluation device.

[0116] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including instructions that can be executed by a processor in a terminal to complete the aforementioned spatial cognitive digital drug efficacy evaluation device. For example, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0117] The above embodiments are not limited to the technical solutions of the embodiments themselves, and the embodiments can be combined with each other to form new embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of the technical solutions of the present invention.

Claims

1. A spatial cognitive digital drug efficacy evaluation method, characterized in that, include: Clinical assessments were conducted on medication users to obtain evaluation indicators for medical and nursing diagnoses. Spatial cognition was assessed on medication users, and multimodal physiological data of medication users were processed and analyzed to obtain efficacy evaluation indicators from multimodal data analysis; The drug acceptance level of the medication users was collected to obtain the evaluation indicators of the medication users; The effectiveness of the drug is comprehensively analyzed by combining the evaluation indicators of medical and nursing diagnosis, the efficacy evaluation indicators of multimodal data analysis, and the evaluation indicators of the drug users. The evaluation indicators for obtaining medical and nursing diagnoses through clinical assessment of medication users include: S101: Collect the Guildford-Zimmerman Spatial Orientation Test Score A1, the Perspective Spatial Positioning Test Score B1, and the Corsi Block Tapping Task Score C1 of the drug user before medication, and calculate the total score of the scales T1 of the drug user before medication, where T1=A1+B1+C1. S102: Collect the Guildford-Zimmerman Spatial Orientation Test Score A2, the Perspective Spatial Positioning Test Score B2, and the Corsi Block Tapping Task Score C2 of the drug user after medication, and calculate the total score of the scales T2 after medication, where T2 = A2 + B2 + C2. S103: Calculate the evaluation index Z1 of the medical diagnosis, where Z1 = T2 - T1; The spatial cognition assessment of medication users, along with the processing and analysis of their multimodal physiological data, yields efficacy evaluation indicators based on multimodal data analysis, including: S201: When the patient performs a spatial cognition assessment task before taking the medication, the portable device simultaneously collects the patient's physiological data before taking the medication. S202: When the user performs a spatial cognition assessment task after taking the medication, the portable device simultaneously collects the user's physiological data after taking the medication; S203: Perform data preprocessing, feature extraction, feature fusion, and data classification on the physiological data collected in step S201 and step S202 in sequence.

2. The spatial cognitive digital drug efficacy evaluation method according to claim 1, characterized in that, The physiological data includes electroencephalogram (EEG) signal data, electrooculogram (EOG) signal data, and eye movement data; The data preprocessing of the physiological data collected in step S201 and step S202 includes: drift data removal, bandpass filtering, artifact removal, and baseline correction. The data preprocessing methods for eye movement data before and after medication include: outlier removal method, missing value handling method, and filtering and noise reduction method. Data preprocessing methods for electrooculogram (EOG) signal data before and after medication include: filtering and noise reduction.

3. The spatial cognitive digital drug efficacy evaluation method according to claim 2, characterized in that, The method for feature extraction of the physiological data collected in step S201 and the physiological data collected in step S202 includes: extracting the coupling features of the electroencephalogram (EEG) signals using the sorting conditional mutual information method. The feature extraction methods for the eye movement data before and after medication include: extracting eye movement behavior and extracting pupil diameter; The feature extraction method for the electrooculogram (EOG) signal data before and after medication includes: calculating the low-to-high frequency average power spectrum ratio.

4. The spatial cognitive digital drug efficacy evaluation method according to claim 3, characterized in that, The method for feature fusion of the physiological data collected in step S201 and the physiological data collected in step S202 includes: fusing features of different modalities by a phased feature splicing method, wherein the features to be fused are the features obtained through feature extraction.

5. The spatial cognitive digital drug efficacy evaluation method according to claim 1, characterized in that, The data classification method includes classifying the features of the fused data using a binary classification method.

6. The spatial cognitive digital drug efficacy evaluation method according to claim 1, characterized in that, The process of collecting data on drug acceptance among medication users, and obtaining evaluation indicators for these users, includes: A questionnaire survey was conducted among the medication users, and the scores from the questionnaire survey were used as evaluation indicators for the medication users.

7. The spatial cognitive digital drug efficacy evaluation method according to claim 6, characterized in that, The questionnaire included the following: The evaluation criteria included: the patient's experience and comfort with digital medication; the safety of digital medication; the therapeutic effect of digital medication; the portability of digital medication; and the level of acceptance of the new treatment method. Each item is scored out of 10 points.

8. A digital drug analysis device, characterized in that, The digital drug analysis device is applied to the spatial cognitive digital drug efficacy evaluation method according to claim 1, and the device comprises: Clinical assessment module: Conduct clinical assessments of medication users to obtain evaluation indicators for medical and nursing diagnoses; Multimodal data assessment module: Spatial cognition assessment of medication users, and processing and analysis of multimodal physiological data of medication users to obtain efficacy evaluation indicators of multimodal data analysis; Patient assessment module: Collects data on the medication acceptance level of patients and obtains evaluation indicators for these patients; Comprehensive analysis module: Receives the evaluation indicators of the medical diagnosis, the efficacy evaluation indicators of the multimodal data analysis, and the evaluation indicators of the medication user, and determines whether the medication used by the medication user is effective.

Images