Freezing gait assessment method and system under Parkinson's disease toe beat field normal form based on fNIRS

Abstract

The invention discloses a method and a system for evaluating a frozen gait under a Parkinson's disease toe beat normal form based on a near infrared brain function imaging technology (fNIRS for short), and the method and the system are used for evaluating the frozen gait under the Parkinson's disease toe beat normal form. The invention provides a sitting toe beating movement normal form which gets rid of the influence of balance factors and can reflect the lower limb movement ability of healthy old people and patients with frozen gaits and non-frozen gaits of Parkinson's disease, and brain function data of different movement stages are collected and calculated by utilizing a near-infrared brain function imaging technology. The activation characteristics of the forehead cortex are reflected when the lower limbs of healthy old people and Parkinson patients move. And then the frozen gait prediction model is constructed by using the brain function data, so that the occurrence of the frozen gait of the Parkinson's disease patient can be effectively predicted, the severity of the frozen gait is accurately evaluated, and a brain function basis for early intervention is provided clinically.

Description

Technical Field

[0001] This invention relates to the field of gait assessment technology, and in particular to a frozen gait assessment method and system based on the toe-tapping paradigm in Parkinson's disease using functional near-infrared spectroscopy (fNIRS). Background Technology

[0002] Parkinson's disease (PD) is a common neurodegenerative disease. Its postural and gait abnormalities include trunk flexion syndrome, difficulty initiating movement, shuffling gait, dragging gait, scoliosis, and frozen gait. Among these, frozen gait occurs in 10%-20% of early-stage PD patients and is more common in the middle and late stages of the disease. It often leads to falls and disability, seriously affecting the patient's quality of life.

[0003] Impaired prefrontal executive function is one of the pathogenic mechanisms of frozen gait in Parkinson's disease. The prefrontal cortex is an important structure of the motor executive control network, but the correlation between the occurrence of frozen gait and prefrontal cortex function remains unclear. Near-infrared brain functional imaging technology can reveal changes in prefrontal cortex brain function under the most realistic and natural conditions. Studies have found that prefrontal cortex executive function affects the occurrence of frozen gait, compensating for decreased walking automation through executive control function.

[0004] Current fNIRS-based frozen gait paradigms for Parkinson's disease primarily focus on lower limb activities such as walking, dual-task walking, obstacle walking, and treadmill walking. It has been found that Parkinson's patients exhibit increased prefrontal cortex activation during lower limb activities. Compared to healthy individuals, Parkinson's patients show increased prefrontal cortex activation during walking, turning, and obstacle crossing. Furthermore, patients with frozen gait exhibit even higher prefrontal cortex activation during walking and turning compared to those without frozen gait. Parkinson's patients not only show increased prefrontal cortex activation during lower limb activities but also during posture control and even static standing. In reality, walking requires a balance between balance and motor functions. Previous studies have all been conducted in a standing position, including motor functions but failing to consider the influence of balance. Existing lower limb movement paradigms for frozen gait in Parkinson's disease encompass both posture control and walking, failing to examine brain function during purely lower limb movement.

[0005] Currently, there are no precise and reliable biomarkers for assessing the severity of frozen gait in Parkinson's disease. Assessment primarily relies on clinical symptoms and physical examination. However, gait is influenced by multiple factors. Tremors in the hands or legs affect range of motion, bradykinesia leads to decreased walking speed and shorter stride length, and increased muscle tone affects standing stability. Simultaneously, non-motor symptoms such as cognitive function, anxiety, and depression also affect frozen gait. Therefore, relying on clinical symptoms and physical examination to assess frozen gait is limited. Commonly used frozen gait assessment scales can retrospectively evaluate the manifestation, frequency, and severity of frozen gait, but due to the influence of patient cognitive function and subjective evaluation, they cannot provide quantitative assessment. Therefore, quantitative assessment of frozen gait in Parkinson's disease patients is particularly important. Summary of the Invention

[0006] To address the aforementioned problems, the purpose of this invention is to provide a method for assessing frozen gait in Parkinson's disease using a toe-tapping paradigm based on fNIRS. This method proposes a seated toe-tapping paradigm that eliminates the influence of balance factors and reflects lower limb motor abilities in both healthy elderly individuals and Parkinson's disease patients. Furthermore, it utilizes fNIRS to collect and calculate brain function data at different stages in both Parkinson's disease patients with and without frozen gait, ultimately reflecting the prefrontal cortex activation during lower limb movement in both healthy elderly individuals and Parkinson's disease patients. By combining brain function data with basic clinical data and gait data, a frozen gait prediction model can be constructed, effectively predicting the occurrence of frozen gait in Parkinson's disease patients and assessing its severity.

[0007] Therefore, the present invention adopts the following technical solution:

[0008] This invention discloses a frozen gait assessment method based on fNIRS in the toe-tapping paradigm for Parkinson's disease, comprising:

[0009] The brain activation changes of subjects were recorded using fNIRS while they performed a toe-tapping exercise paradigm to obtain brain function activation indicators. The toe-tapping exercise paradigm included: the subject sat comfortably in a chair with a straight back and armrests, placed his heel on the ground, and tapped the ground with his toes for 20 seconds with the maximum amplitude and speed.

[0010] The toe tapping exercise assessment scale was used to obtain the toe tapping exercise scale scores of the subjects.

[0011] Obtain basic clinical and gait data from the subjects;

[0012] A machine learning-based frozen gait prediction model was constructed. The model was trained using a dataset consisting of brain function activation indicators, toe-tapping exercise scale scores, clinical basic data, and gait data from three groups of people: Parkinson's disease patients with frozen gait, Parkinson's disease patients with non-frozen gait, and healthy elderly people when performing the toe-tapping exercise paradigm.

[0013] Using the subjects' brain function activation indicators, toe tapping exercise scale scores, basic clinical data, and gait data as model inputs, the trained frozen gait prediction model was used to predict the occurrence of frozen gait in Parkinson's disease patients and to assess the severity of frozen gait.

[0014] Furthermore, the toe-slapping motion evaluation scale assesses the speed of the toe-slapping motion, the amplitude of the toe's distance from the floor, whether there is any slowness or pause in the motion, and whether there is a trend of the amplitude of the toe-slapping motion decreasing over time. It is divided into 5 levels: normal, slight, mild, moderate, and severe, with a score of 0-4.

[0015] Furthermore, the basic clinical data include: UPDRS III score, HY stage, disease duration, daily levodopa equivalent dose, MMSE score, MoCA score, animal speech fluency test score, 10-minute drawing test score, HAMA score, and HAMD score.

[0016] Furthermore, the gait data includes: gait speed, stride length, stride length, support phase duration, swing phase duration, maximum forward swing angle of the left and right lower legs, maximum backward swing angle of the left and right lower legs, peak angular velocity of the left and right lower legs, maximum forward swing angle of the left and right arms, maximum backward swing angle of the left and right arms, and peak angular velocity of the left and right arms in the timed standing-up walking test.

[0017] Furthermore, the brain function activation indicators include: baseline period, right toe tapping task period, rest period, left toe tapping task period, rest period, and average HbO and Hb concentrations in each phase.

[0018] Furthermore, prior to model training, the following steps are also included:

[0019] We used fNIRS to collect data on brain activation changes when healthy elderly individuals, those with frozen gait due to Parkinson's disease, and those without frozen gait performed a toe-tapping movement paradigm.

[0020] By calculating brain activation changes at different stages, the correlation between prefrontal cortex activation during lower limb movement was obtained for healthy elderly individuals, those with Parkinson's disease exhibiting frozen gait, and those without frozen gait.

[0021] Furthermore, it also includes: using brain activation change data collected by fNIRS of healthy older adults performing the toe-tapping exercise paradigm to verify the reliability of the toe-tapping exercise paradigm.

[0022] Furthermore, prior to model training, the following steps are also included:

[0023] A correlation analysis was conducted on the frozen gait severity scale and changes in brain activation to obtain the correspondence between changes in brain activation and the severity of frozen gait.

[0024] Furthermore, the frozen gait prediction model is based on the Gradient Boosting Decision Tree (GBDT) model;

[0025] Brain function activation indicators, toe-tapping exercise scale scores, clinical baseline data, and gait data were collected from three groups of people: Parkinson's disease patients with frozen gait, Parkinson's disease patients with non-frozen gait, and healthy elderly individuals, when they performed the toe-tapping exercise paradigm. Brain function activation indicators were standardized, toe-tapping exercise scale scores were coded, missing value imputation and outlier handling were performed on clinical baseline data, and gait data were normalized to form a comprehensive feature set.

[0026] Multiple gradient boosting decision trees are constructed and trained using the comprehensive feature set, and importance scores of each feature are extracted from the gradient boosting decision trees; the importance of the feature is measured by calculating its contribution to the reduction of the loss function when it is a split node in all trees;

[0027] Each feature in the comprehensive feature set is ranked according to its importance in the model to assess and validate the severity of frozen gait;

[0028] The original dataset is randomly and evenly divided into five subsets;

[0029] Cross-validation is used to evaluate the model's performance.

[0030] In another aspect, the present invention also provides a frozen gait assessment system based on fNIRS in the toe-tapping paradigm for Parkinson's disease, comprising:

[0031] The first data acquisition unit uses fNIRS to record brain activation changes when subjects perform the toe-tapping exercise paradigm, and obtains the subjects' brain function activation indicators. The toe-tapping exercise paradigm includes: the subject sits comfortably in a chair with a straight back and armrests, places his heels on the ground, and taps the ground with his toes for 20 seconds with the maximum amplitude and speed.

[0032] The second data acquisition unit uses the toe tapping exercise evaluation scale to obtain the subject's toe tapping exercise scale score.

[0033] The third data acquisition unit acquires the subject's basic clinical data and gait data;

[0034] The prediction model building unit constructs a machine learning-based frozen gait prediction model; the model is trained using a dataset consisting of brain function activation indicators, toe-tapping exercise scale scores, basic clinical data, and gait data from three groups of people: Parkinson's disease patients with frozen gait, Parkinson's disease patients with non-frozen gait, and healthy elderly people when performing the toe-tapping exercise paradigm.

[0035] The frozen gait assessment unit uses the subject's brain function activation index, toe tapping exercise scale score, clinical basic data, and gait data as model inputs. It uses the trained frozen gait prediction model to predict the occurrence of frozen gait in the subject's Parkinson's disease and assess the severity of frozen gait.

[0036] Advantages and positive effects of the present invention:

[0037] This invention aims to eliminate balance factors and analyze changes in brain function during lower limb movement, focusing on prefrontal cortex activation during seated limb movement. It establishes a seated toe-tapping movement paradigm, first validating its reliability in healthy elderly individuals, and then applying it to Parkinson's disease patients. Furthermore, it utilizes brain function data combined with basic clinical and gait data to construct a frozen gait prediction model, effectively predicting the occurrence and severity of frozen gait in Parkinson's disease patients. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 A flowchart illustrating the method for evaluating lower limb motor function in healthy elderly individuals and patients with Parkinson's disease, as described in this invention.

[0040] Figure 2 This is a schematic diagram illustrating the seated toe-tapping task paradigm of an embodiment of the present invention;

[0041] Figure 3 This is a flowchart illustrating the seated toe-tapping task paradigm of an embodiment of the present invention;

[0042] Figure 4 This is a diagram showing the distribution of fNRIS optical poles and channels in an embodiment of the present invention.

[0043] Figure 5 This is a three-dimensional schematic diagram of the fNIRS channel location according to an embodiment of the present invention;

[0044] Figure 6This is a diagram showing the activation of the prefrontal cortex in a healthy elderly person during a seated toe-tapping exercise, according to an embodiment of the present invention.

[0045] Figure 7 This invention provides a topological map of the prefrontal cortex activation of a Parkinson's disease patient with frozen gait, a Parkinson's disease patient with non-frozen gait, and a healthy elderly person during a seated right toe tapping motion.

[0046] Figure 8 The figures represent the relative changes in oxyhemoglobin in the right prefrontal cortex of Parkinson's disease patients with frozen gait, Parkinson's disease patients with non-frozen gait, and healthy elderly individuals during the right toe-tapping exercise phase, as described in this invention.

[0047] Figure 9 Examples of the present invention include hemoglobin activation curves of the frontal lobe in patients with frozen gait, patients with non-frozen gait, and healthy elderly individuals during seated right toe tapping movements.

[0048] Figure 10 This invention relates the relative change in oxygenated hemoglobin in the right prefrontal cortex of patients with frozen gait to the severity of frozen gait.

[0049] Figure 11 This is the verification result of the gradient boosting decision tree model in the embodiment of the present invention. Detailed Implementation

[0050] 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.

[0051] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0052] like Figure 1As shown in the embodiment of the present invention, a method for assessing Parkinson's disease toe-tapping paradigms and frozen gait based on fNIRS is provided. The method includes the following steps:

[0053] S101. Using fNIRS to record brain activation changes in subjects when performing the toe-tapping movement paradigm, the subjects' brain function activation indicators were obtained.

[0054] Postural balance and gait are inherently linked. Maintaining postural stability and walking function relies on common neural circuits involving the basal ganglia, cortex, brainstem, and spinal cord. Insufficient postural control can impair walking function, leading to decreased walking speed, while postural balance rehabilitation training can improve gait. Existing walking, obstacle-crossing walking, and treadmill walking paradigms all incorporate both postural balance and walking aspects, failing to assess brain function in cases of lower limb movement alone. We believe that experimental paradigms should consider the functions of both aspects separately as much as possible. An effective lower limb movement paradigm should first be effective, i.e., able to detect lower limb functional abnormalities and differences from normal subjects. Secondly, the paradigm needs to be reproducible, not only able to be repeatedly applied but also maintaining consistency and stability across different studies. Simultaneously, the paradigm should be simple, with strong reproducibility in clinical application, minimizing discomfort for patients. Therefore, the paradigm designed in this study adheres to the above principles of effectiveness, reproducibility, and simplicity, establishing a seated toe-tapping exercise paradigm that minimizes balance factors.

[0055] This invention presents the establishment and verification of a toe-tapping exercise paradigm. The experiment reliably recorded brain activation changes during toe-tapping exercises in healthy elderly individuals using fNIRS technology. The study demonstrates the reliability of the toe-tapping exercise paradigm and reveals the influence of the prefrontal cortex on lower limb motor execution in healthy elderly individuals. Both prefrontal cortex regions in healthy elderly individuals participated in the execution of left and right toe-tapping exercises, respectively. Furthermore, the activation level of the dorsolateral prefrontal cortex was negatively correlated with cognitive level; the lower the cognitive reserve, the greater the role of the motor execution control network. This invention establishes a toe-tapping exercise paradigm for the first time and uses task-based fNIRS technology to verify that the toe-tapping exercise paradigm in a seated state reflects the prefrontal cortex's execution control of lower limb activities, exploring the neurophysiological mechanism of the prefrontal cortex's involvement in lower limb movements in healthy elderly individuals in a seated state.

[0056] Toe tapping is a classic exercise for evaluating lower limb motor control. The most commonly used assessment scales for Parkinson's disease, the International Parkinson and Movement Disorder Society (MDS) and the Unified Parkinson's Disease Rating Scale (UPDRS), include toe tapping in their motor function assessment sections. The MDS-UPDRS has been extensively clinically proven to be an effective measure of lower limb motor ability, demonstrating good validity and repeatability. The specific assessment method for toe tapping in this invention is essentially the same as that in the MDS-UPDRS, except that the MDS-UPDRS involves 10 consecutive taps, while this invention involves approximately 40 taps over 20 seconds. In rhythmic movements such as walking, toe tapping, and finger tapping in healthy individuals, following the 2Hz human resonance theory, the comfortable time interval between two repetitions is 500-600ms, or 100-120 times per minute, or 2 times per second. Therefore, it takes approximately 5 seconds for a healthy individual to complete 10 toe tapping movements in the MDS-UPDRS. The brain's blood oxygenation response time to movement is 4-7 seconds. To effectively detect changes in cortical blood oxygenation during toe tapping movements using fNRIS, this invention increases the task duration to ensure the stability of experimental results. Simultaneously, to maintain the simplicity of the task paradigm, and considering current fNIRS-based lower limb movement studies in Parkinson's disease patients, most paradigms show significant differences in activation during the first 20 seconds of the movement phase compared to healthy individuals. A 20-second movement phase is sufficient to distinguish between patients and healthy individuals; therefore, the task duration in this paradigm is 20 seconds. Based on existing paradigms, the baseline period typically involves a quiet standing rest for 20-30 seconds, with the rest period also lasting 20-30 seconds. This invention designs both the rest and rest periods to be 20 seconds each. This constitutes one trial; the paradigm is repeated three times, and the change in hemoglobin concentration is taken as the average of the three trials.

[0057] Among them, the brain activation changes in the toe-tapping exercise paradigm are mainly data on changes in the concentrations of oxyhemoglobin and deoxyhemoglobin in the prefrontal cortex.

[0058] S102: The toe tapping exercise assessment scale was used to obtain the subject's toe tapping exercise scale score.

[0059] Specifically, a self-developed toe-tapping exercise assessment scale was used, referencing the toe-tapping exercise in the Unified Parkinson's Disease Rating Scale of the International Parkinson's Disease and Movement Disorders Association. Subjects were asked to sit comfortably in a chair with a straight back and armrests, placing their heels on the ground, and tapping their toes with maximum amplitude and speed for 20 seconds. The speed of the movement, the amplitude of the toes off the floor, any slowing or pauses in the movement, and whether there was a trend of decreasing amplitude in the toe-tapping motion over time were assessed. The severity was graded into five levels: normal, mild, slight, moderate, and severe, with a score of 0-4.

[0060] like Figure 3 As shown, the brain function test consists of 5 phases: baseline phase, right toe tapping task phase, rest phase, left toe tapping task phase, and rest phase, totaling 130 seconds. This constitutes one trial, which is repeated three times.

[0061] S201. Baseline period: Subjects sit quietly in a comfortable position, looking at the "+" mark on the computer screen, minimizing eye movements and actions such as coughing, chewing, and yawning. The fNIRS data collected is used as baseline data for a total of 30 seconds.

[0062] S202, Right Toe Tapping Task: Participants tapped their right toes on the ground continuously for 5 seconds at maximum amplitude and speed, repeated 4 times for a total of 20 seconds, while a "+" mark continued to be displayed on the screen.

[0063] S203. During the rest period, the subjects sat quietly for 30 seconds, looking at the "+" mark on the computer screen and avoiding movement.

[0064] S204, Left Toe Tapping Task: Participants performed a continuous left toe tapping motion for 5 seconds at maximum amplitude and speed, repeated 4 times for a total of 20 seconds, while a "+" mark continued to be displayed on the screen.

[0065] S205, Rest period, 30 seconds in total. Subjects sit quietly and look at the "+" mark on the computer screen, avoiding movement.

[0066] S103: Obtain basic clinical data and gait data of the subjects;

[0067] The basic clinical data include: UPDRS III score, HY stage, disease course, daily levodopa equivalent dose, MMSE score, MoCA score, animal speech fluency test score, 10-minute drawing test score, HAMA score, and HAMD score.

[0068] Gait data includes: gait speed, stride length, stride length, support phase duration, swing phase duration, maximum forward swing angle of left and right lower legs, maximum backward swing angle of left and right lower legs, peak angular velocity of left and right lower legs, maximum forward swing angle of left and right arms, maximum backward swing angle of left and right arms, and peak angular velocity of left and right arms in the timed stand-up walking test.

[0069] S104: Construct a machine learning-based frozen gait prediction model; train the model using a dataset consisting of brain function activation indicators, toe-tapping exercise scale scores, clinical basic data, and gait data from three groups of people: Parkinson's disease patients with frozen gait, Parkinson's disease patients without frozen gait, and healthy elderly people when performing the toe-tapping exercise paradigm.

[0070] Specifically, S401: The study was divided into three groups: a Parkinson's disease gait freezing group, a Parkinson's disease non-gait freezing group, and healthy older adults. Basic information of the participants in all three groups was recorded, including gender, age, education level, height, anxiety and depression scale, cognitive scale, UPDRS III score, HY stage, disease duration, daily levodopa equivalent dose, MMSE score, MoCA score, animal speech fluency test score, 10-minute drawing test score, HAMA score, and HAMD score. For patients in the Parkinson's disease gait freezing and non-gait freezing groups, Part III of the Unified Parkinson's Disease Rating Scale (URRS), disease duration, and levodopa equivalent dose were recorded.

[0071] Among them, Parkinson's disease patients were classified into the Parkinson's disease frozen gait group and the Parkinson's disease non-frozen gait group according to the internationally recognized frozen gait questionnaire. The frozen gait questionnaire assesses the severity of frozen gait and its impact on daily life by evaluating the scenarios, frequency and duration of frozen gait. It has a total score of 24 points, with ≥1 point indicating the presence of frozen gait and <1 point indicating the absence of frozen gait.

[0072] The specific evaluation methods are as follows:

[0073] The Hamilton Anxiety Rating Scale (HAMA): The most classic anxiety scale, it is a peer-rating scale. It includes 14 items, divided into two main factors: physical and psychological anxiety. HAMA scoring criteria: <7 points indicate no anxiety symptoms; 7-13 points indicate possible mild anxiety; 14-20 points indicate some degree of anxiety; 21-28 points indicate significant anxiety; ≥29 points indicate severe anxiety.

[0074] The Hamilton Depression Rating Scale (HAMD): The most classic depression scale, it is a self-rating scale. It contains 24 items. HAMD scoring criteria: <8 points indicate no depressive symptoms; 8-20 points indicate possible mild depressive symptoms; 21-34 points indicate possible moderate depressive symptoms; >35 points indicate possible severe depressive symptoms.

[0075] Part III of the Unified Parkinson's Disease Rating Scale (UKR) is used to assess patients' motor function. It includes 18 motor items such as speech, facial expression, rigidity, finger tapping, hand movements, forearm rotation, and toe tapping. Each item is graded on a scale of 0 to 4, from mild to severe.

[0076] Levodopa daily equivalent dose: Based on the standard conversion factor established by the International Movement Disorders Association, the anti-Parkinson's drugs are converted into levodopa doses to calculate the total daily oral levodopa dose for the patient.

[0077] S402: Healthy elderly individuals, Parkinson's disease patients with frozen gait, and Parkinson's disease patients with non-frozen gait complete cognitive function assessment and gait quantitative evaluation.

[0078] Analyzing the relationship between cognitive function and frozen gait from a behavioral perspective reveals that Parkinson's disease patients with frozen gait experience a decline in cognitive function, especially executive function, which is correlated with gait parameters.

[0079] Cognitive function assessment includes overall cognitive function assessment and executive function assessment; overall cognitive function assessment uses the Mini-Mental Intelligence Scale and the Montreal Cognitive Scale; executive function assessment uses the categorized word animal fluency task and the 10-minute drawing test in the verbal fluency test.

[0080] Mini-Intelligence Scale: Used to assess the cognitive function of subjects, including tests on orientation, memory, attention and calculation, language, etc. The total score is 30 points, and a score greater than 27 points is considered normal. The criteria for dementia are: illiterate ≤17 points, primary school education ≤20 points, and secondary school education ≤24 points.

[0081] The Montreal Cognitive Inventory includes test items in eight cognitive domains: executive function, language, visuospatial function, attention, abstract thinking, calculation ability, and orientation. The maximum score is 30 points. If the subject's years of education are ≤12 years, 1 point is added to the total score. A score of 26 or above is defined as normal, and a score of less than 21 is defined as dementia.

[0082] Animal Speech Fluency Test: A method for measuring language fluency, requiring test takers to name as many animals as possible within one minute, counting the number of correct names. Scoring method: 1 point for each correct name; animals appearing in imagination or mythology, such as dragons and qilin, are counted correctly. Alternative names are counted as duplicates, and only one is counted.

[0083] 10-Point Clock Drawing Test: A method for assessing executive function, this test requires participants to draw a clock, fill in all the numbers, and indicate 11:10. The Sunderland scoring method is used, with a maximum score of 10: 9-6: Clock face and hands are mostly complete, with slight errors in hand position. 5-1: Clock face, numbers, and / or hand position are mostly incorrect.

[0084] Gait quantification assessment utilizes a wearable motion and gait quantification assessment system. Based on wearable motion sensing technology, it employs high-precision sensors to capture real-time human temporal-spatial parameters and kinematic data, transmitting this data to a computer system for further model construction and analysis of relevant spatiotemporal motion characteristics. Data collection took place in a spacious clinic room, maintaining a quiet and comfortable environment. The clinic room floor was equipped with backrested chairs, and markings indicated the starting walking, turning back, and 360° rotation test positions. The distance from the starting point to the turning back was 5 meters, and the 360° rotation test position was located in the middle of the 5-meter mark, forming a 0.6-meter square. Before data collection, following the system operation manual, data sensors were worn at 10 points: both wrists, chest, waist, both thighs, both calves, and both feet. During data collection, a high-definition camera was connected to the computer containing the system. The lens angle was adjusted according to the actual experimental conditions to ensure the camera could capture the entire experimental scene. After the experiment began, staff and family members present minimized movement, and other personnel avoided entering the area to prevent interference with the acquisition of important image segments. Before the experiment, the subjects were fully informed of the task content and practiced 2-3 times to ensure a smooth testing process.

[0085] Preferably, gait quantification assessment includes a gait task of a timed stand-up walking test. Timed stand-up walking test: The subject sits in a chair with a backrest, approximately 45cm high, with armrests approximately 20cm high. After hearing the "start" command, the subject stands up, stabilizes, and walks forward in a straight line at a comfortable pace and gait to a 5-meter marker. Then, the subject turns back and sits down in the chair. The test obtains gait temporal and spatial parameters and kinematic characteristics, including gait speed, stride length, stride length, stance phase duration, swing phase duration, maximum forward and backward swing angles of both lower legs, peak angular velocities of both lower legs, maximum forward and backward swing angles of both arms, and peak angular velocities of both arms.

[0086] In other implementations, gait tasks such as the narrow-path walking test and the 360° stationary turn test can also be used. Narrow-path walking test: Before the experiment, a narrow passage is constructed using two chairs at the midpoint of a 5-meter test distance. Subjects sit in chairs with backrests, following the same procedure as the stand-up-walk test. After hearing the "start" command, they stand up, stabilize themselves, and walk forward in a straight line at a comfortable pace and gait, traversing the narrow passage to the 5-meter marker, returning, traversing the narrow passage again, and finally walking back to the chairs to sit down. The test obtains relevant gait parameters during the cognitive task of walking with the narrow passage, including gait speed, stride length, stride length, stance phase duration, swing phase duration, maximum forward and backward angles of the left and right lower legs, peak angular velocities of the left and right lower legs, maximum forward and backward angles of the left and right arms, and peak angular velocities of the left and right arms.

[0087] 360° circling test: Subjects stand within a 0.6-meter diameter square marked on the ground at a 5-meter test distance. They rotate continuously two times in place, first to their left (360° constitutes one rotation), then stand still for 10 seconds. The same method is then repeated, rotating continuously two times in place to their right. The test obtains relevant gait parameters during the circling task, including gait speed, stride length, stride length, stance phase duration, swing phase duration, maximum forward and backward angles of both lower legs, peak angular velocities of both lower legs, maximum forward and backward angles of both arms, and peak angular velocities of both arms.

[0088] SPSS 22.0 statistical software was used for data processing. Partial correlation analysis was used to analyze the correlation between gait parameters and cognitive function. After opening the software, click Analyze - Correlation - Partial Correlation. The overall cognitive function assessment was performed using the Mini-Mental State Examination (MMS) score, Montreal Cognitive Inventory (MCI) score, Animal Fluency Task and 10-Minute Drawing Test score, and compared with gait speed, stride length, stride length, stance phase duration, swing phase duration, maximum forward swing angle of both lower legs, maximum backward swing angle of both lower legs, peak angular velocity of both lower legs, maximum forward swing angle of both arms, and maximum forward swing angle of both arms. Gait parameters such as the maximum backswing angle and peak angular velocity of the left and right arms were input into the statistical software as pairwise variables. Demographic factors and clinical symptoms such as age, gender, disease duration, UPDRS III, HY stage, daily levodopa equivalent dose, Hamilton Anxiety Rating Scale score, and Hamilton Depression Rating Scale score were used as control variables. Partial correlation analysis was then performed, and a two-tailed test was used for significance testing to obtain the correlation coefficient between cognitive function and gait parameters. The correlation coefficient ranged from -1 to 1, with the closer to -1 indicating a stronger negative correlation and the closer to 1 indicating a stronger positive correlation.

[0089] S403: Healthy elderly individuals complete the seated toe tapping exercise scale and brain function test, and observe the different activation patterns of the prefrontal cortex at different stages of the exercise to verify the effectiveness and reliability of the paradigm.

[0090] like Figure 6 As shown, the brain activation in healthy elderly individuals during different phases of toe-tapping exercises was analyzed. It was found that the activation of the prefrontal cortex increased during left and right toe-tapping exercises, and the brain activation gradually returned to the pre-activity level during the rest period. This verifies the effectiveness and reliability of the toe-tapping exercise paradigm in reflecting lower limb activity and brain function.

[0091] S404: Parkinson's disease patients complete the toe tapping exercise scale and brain function test to assess the different activation of the prefrontal cortex in patients with frozen gait and non-frozen gait, and analyze the relationship between prefrontal cortex activation and the severity of frozen gait from the perspective of brain function.

[0092] To control for the effects of demographic factors and clinical symptoms on brain activation, a linear mixed model was used to control for covariates including disease duration, UPDRS III, HY stage, MMSE score, and MoCA score. Age and sex were used as factors, and group was used as the dependent variable. The model was compared between groups under both fixed and main effects. In SPSS software, the linear mixed model was opened by clicking Analyze - Mixed Models - Linear. Brain function data were placed in the "Dependent Variable" field, and group was placed in the "Factor" field. Demographic factors such as sex, age, height, MMSE, HAMA, HAMD, HY stage, UPDRS III, disease duration, and LEDD, as well as the severity of clinical symptoms, were placed in the "Covariates" field to control for their effects on brain function.

[0093] The brain function testing equipment is configured as follows: This invention uses a quantitative brain function imaging device with near-infrared light source wavelengths of 695nm and 830nm, a time resolution of 0.1 seconds, and a sampling rate of 10Hz. The probe cap has 33 probes in total: 17 transmitting probes and 16 receiving probes, arranged in a 3×11 grid. The area between the transmitting and receiving probes, where the emitted light reflects inside the brain, is the detection area, called a channel. There are a total of 52 channels. Figure 4 As shown. The distance between each pair of emitting and receiving photodiodes is 3cm, which can detect changes in optical density within a depth range of 2-3cm from the scalp. The raw optical density data is converted into HbO and Hb concentration data (path length correction factors DPF695 = 6.51, DPF830 = 5.86) by the ETG-4000 system's built-in software according to default parameters, based on Lambert-Beer's law.

[0094] The probe distribution and brain region localization are as follows: based on international EEG standards. Figure 10-2The system electrode placement marks the scalp probe position. After locating Cz and Fz, the midlines of Cz and Fz are aligned parallel to the center of the probe cap, and the probe cap is then placed. Subsequently, a magnetic navigation three-dimensional positioning system is used to accurately locate the brain region. Figure 5 First, the positions of Cz, nasal root, occipital protuberance, and bilateral preauricular fossa were marked using a navigation system. Then, the positions of each channel were registered to the Montreal Neurological Institute and Hospital (MNI) coordinate system. The registration probabilities of each channel in the MNI coordinate system and its corresponding anatomical position are shown in Table 1. The bilateral dorsolateral prefrontal lobes and bilateral frontal polar regions were selected as regions of interest (FPAs), rdlPFC (Ch3-5, Ch14, Ch24), ldlPFC (Ch6-9, Ch18, Ch19, Ch29, Ch40), rFPA (Ch15, Ch25, Ch26, Ch35, Ch36, Ch46, Ch47), and lFPA (Ch17, Ch27, Ch28, Ch38, Ch39, Ch48, Ch49).

[0095] Based on probe distribution and brain region localization, fNIRS data obtained using the toe-tapping paradigm assessment method were preprocessed using the NIRS-KIT toolkit based on the MATLAB toolbox. This preprocessing mainly consisted of three steps:

[0096] 1) Drift Reduction: fNIRS signals exhibit chronic drift over time. To eliminate these systematic trend changes, drift reduction is first performed. NIRS-KIT uses a polynomial fitting algorithm to estimate a linear or non-linear trend and removes it from the original blood oxygenation signal. The default polynomial order is first order. The calculation process for activation level indicators of prefrontal cortex brain regions at various stages is as follows:

[0097] 2) Head movement correction: In the experimental paradigm, subjects inevitably experience head movements, which leads to relative displacement between the optical probe and the scalp. Motion artifacts in the fNIRS signal are obvious unidirectional jumps in oxyhemoglobin and deoxyhemoglobin in the time series. Spatially, head movements may cause relative displacement between the entire optical plate and the scalp. Therefore, this invention uses the time derivative distribution repair head movement correction method to reduce the impact of head movements on the signal, which can effectively remove various jump noises and baseline drift.

[0098] 3) Filtering: Since the acquired fNIRS signals contain physiological signals such as respiration and heartbeat in addition to those in the region of interest (ROI), filtering is necessary to reduce their impact on the true signal. In this study, a 0.01–0.08 Hz bandpass filter was used to remove low-frequency and high-frequency noise to reflect true neural activity. Because the obtained fNIRS data are relative changes, this invention performed Z-score correction on the fNIRS data. The baseline used was the average of the baseline 30 seconds of data, yielding the relative changes in fNIRS data during the task period. After the above preprocessing, the average oxyhemoglobin and deoxyhemoglobin concentrations for each subject in each channel during the task and rest periods were obtained. To increase the signal-to-noise ratio, the oxyhemoglobin and deoxyhemoglobin concentrations within the same ROI were summed and averaged to obtain the average oxyhemoglobin and deoxyhemoglobin concentrations within each subject's ROI.

[0099] The changes in oxyhemoglobin concentration in the prefrontal cortex were calculated during the baseline period, the right toe tapping task period, the rest period, the left toe tapping task period, and the rest period. The relative changes in hemoglobin concentration during the exercise period compared to the baseline and rest periods reflected the prefrontal cortex's executive control over the toe tapping movement. One-way ANOVA was used to compare the differences in bilateral prefrontal cortex activation during the task period among Parkinson's disease patients with frozen gait, Parkinson's disease patients without frozen gait, and healthy elderly individuals. The results showed differences in the activation level of the right prefrontal cortex among the three groups during the task period. Post-hoc tests revealed that Parkinson's disease patients with frozen gait > Parkinson's disease patients without frozen gait, and Parkinson's disease patients with frozen gait > healthy elderly individuals, with statistically significant differences. This indicates that the activation level of the right prefrontal cortex during right toe tapping was higher in Parkinson's disease patients with frozen gait than in Parkinson's disease patients without frozen gait and healthy elderly individuals. Figure 7 The study presents a topological map of oxyhemoglobin levels in the bilateral prefrontal cortex during a right toe tapping task in patients with frozen gait, non-frozen gait, and healthy elderly individuals. It shows that all three groups of subjects exhibited increased bilateral prefrontal cortex activation, with the degree of increase being: patients with frozen gait > patients with non-frozen gait > healthy elderly individuals. Figure 8 The figure shows the relative changes in oxyhemoglobin in the right prefrontal cortex of three groups during the task period, with the most significant changes observed in patients with Parkinson's disease and frozen gait. Figure 10 As shown, further correlation analysis between the frozen gait questionnaire and the activation level of the right prefrontal cortex revealed a correlation between the two. The more severe the frozen gait, the more the right prefrontal cortex was activated. That is, in patients with severe frozen gait, performing the toe-tapping task requires more prefrontal cortex executive function support.

[0100] S405: Using UPDRS scores, cognitive function assessments, gait data, and indicators such as toe tapping exercise scales, cognitive function scales, and brain function activation, a gradient boosting decision tree model GBDT is established, and the sensitivity and specificity of the model are verified by five-fold cross-validation to assess and validate the severity of frozen gait.

[0101] The specific steps of using the Gradient Boosting Decision Tree (GBDT) model with five-fold cross-validation are as follows:

[0102] First, a GBDT model is constructed. GBDT is an ensemble model that uses multiple decision trees for prediction. A training dataset is used to build a GBDT model. Next, feature selection is performed: the constructed GBDT model is used to select important features. For each feature, we can calculate its frequency across all trees and rank the feature importance based on this frequency. Then, the dataset is split, randomly and evenly dividing the original dataset into five subsets. Cross-validation is used to evaluate the model's performance. One subset is selected sequentially as the test set, and the other four subsets are used as the training set. The model is trained and evaluated on the test set. This process is repeated five times, ensuring that each subset is used as the test set once. Finally, the average of the five performance evaluation results is taken to obtain the final evaluation result, such as... Figure 11 As shown.

[0103] The GBDT model was subjected to five-fold cross-validation, and the results are shown in Table 1.

[0104] Table 1

[0105]

[0106] S105: Using the subject's brain function activation index, toe tapping exercise scale score, clinical basic data and gait data as model inputs, the trained frozen gait prediction model is used to predict the occurrence of frozen gait in the subject's Parkinson's disease and assess the severity of frozen gait.

[0107] The assessment method described in the above embodiments used patients with frozen gait, non-frozen gait, and healthy elderly individuals as subjects. First, from a behavioral perspective, the gait of patients with frozen gait was quantitatively analyzed, observing their gait characteristics under timed standing-walking and cognitive-related motor task paradigms. The correlation between frozen gait and overall cognitive and executive functions was analyzed. Second, leveraging the high temporal resolution and ecological validity of the fNIRS, a seated toe-tapping movement paradigm suitable for studying prefrontal executive control function in patients with frozen gait was created, and its reliability was validated in healthy elderly individuals. Finally, the seated toe-tapping exercise paradigm was applied to patients with frozen gait and those without, and the differences in brain function under the toe-tapping exercise paradigm were compared. Using UPDRS scores, gait data, and indicators such as the toe-tapping exercise scale, cognitive function scale, and brain function activation, GBDT and random forest models were established, and the sensitivity and specificity of the five-fold cross-validation model were tested. This model can be used to assess the severity of frozen gait.

[0108] Corresponding to the assessment methods in the above embodiments, this embodiment of the invention also provides a frozen gait assessment system based on fNIRS in the toe-tapping paradigm for Parkinson's disease, comprising:

[0109] The first data acquisition unit uses fNIRS to record brain activation changes when subjects perform the toe-tapping exercise paradigm, and obtains the subjects' brain function activation indicators. The toe-tapping exercise paradigm includes: the subject sits comfortably in a chair with a straight back and armrests, places his heels on the ground, and taps the ground with his toes for 20 seconds with the maximum amplitude and speed.

[0110] The second data acquisition unit uses the toe tapping exercise evaluation scale to obtain the subject's toe tapping exercise scale score.

[0111] The third data acquisition unit acquires the subject's basic clinical data and gait data;

[0112] The prediction model building unit constructs a machine learning-based frozen gait prediction model; the model is trained using a dataset consisting of brain function activation indicators, toe-tapping exercise scale scores, basic clinical data, and gait data from three groups of people: Parkinson's disease patients with frozen gait, Parkinson's disease patients with non-frozen gait, and healthy elderly people when performing the toe-tapping exercise paradigm.

[0113] The frozen gait assessment unit uses the subject's brain function activation index, toe tapping exercise scale score, clinical basic data, and gait data as model inputs. It uses the trained frozen gait prediction model to predict the occurrence of frozen gait in the subject's Parkinson's disease and assess the severity of frozen gait.

[0114] The evaluation system of this invention is described simply because it corresponds to the evaluation method in the above embodiments. For any similarities, please refer to the description of the evaluation method section in the above embodiments, which will not be elaborated here.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating freezing gait in Parkinson's disease under the fNIRS-based toe-tapping paradigm, characterized in that, The method comprises the following steps: record the brain activation changes of the subject during the execution of the toe-tapping movement paradigm by using fNIRS, and obtain the brain function activation index of the subject; the toe-tapping movement paradigm comprises: the subject comfortably sits on a chair with a straight back and handles, and places the heel on the ground, and taps the toes on the ground with the maximum amplitude and the fastest speed for 20 seconds; obtain the toe-tapping movement scale score of the subject by using the toe-tapping movement evaluation scale; obtain the clinical basic data and gait data of the subject; construct a frozen gait prediction model based on machine learning; use the brain function activation index, toe-tapping movement scale score, clinical basic data and gait data of the Parkinson's disease frozen gait patient, Parkinson's disease non-frozen gait patient and healthy elderly person to train the model; use the brain function activation index, toe-tapping movement scale score, clinical basic data and gait data of the subject as the input of the model, and use the trained frozen gait prediction model to predict the occurrence of the frozen gait of the Parkinson's disease patient and evaluate the severity of the frozen gait.

2. The fNIRS-based method for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm according to claim 1, characterized in that, The toe-tapping movement evaluation scale evaluates the speed of the toe-tapping movement, the amplitude of the toe distance from the floor, whether there is movement retardation or pause, and whether there is a trend that the amplitude of the toe tapping becomes smaller and smaller, and is divided into five levels of normal, slight, mild, moderate and severe, with scores of 0-4.

3. The fNIRS-based method for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm according to claim 1, characterized in that, The clinical basic data includes: UPDRS III score, H-Y staging, disease duration, daily levodopa equivalent dose, MMSE score, MoCA score, animal speech fluency test score, 10-minute walk test score, HAMA score and HAMD score.

4. The fNIRS-based method for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm according to claim 1, characterized in that, The gait data includes: step speed, step length, stride length, support phase duration, swing phase duration, left and right calf front swing maximum angle value, left and right calf back swing maximum angle value, left and right calf angular velocity peak value, left and right arm front swing maximum angle value, left and right arm back swing maximum angle value, and left and right arm angular velocity peak value.

5. The fNIRS-based method for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm according to claim 1, characterized in that, The brain function activation index includes: baseline period, right toe-tapping task period, rest period, left toe-tapping task period, rest period, and average HbO and Hb concentration in each period.

6. The fNIRS-based method for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm according to claim 1, characterized in that, Before model training, it also includes: collect the brain activation change data of the healthy elderly person, Parkinson's disease frozen gait and non-frozen gait person during the execution of the toe-tapping movement paradigm by using fNIRS; calculate the brain activation change data in different periods to obtain the corresponding relationship of the prefrontal cortex activation of the healthy elderly person, Parkinson's disease frozen gait and non-frozen gait person during the lower limb movement.

7. The fNIRS-based method for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm according to claim 6, characterized in that, It also includes: the brain activation change data of the healthy elderly person during the execution of the toe-tapping movement paradigm collected by using fNIRS, which verifies the reliability of the toe-tapping movement paradigm.

8. The fNIRS-based method for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm according to claim 6, characterized in that, Before model training, it also includes: correlation analysis of the frozen gait severity scale and the brain activation change is performed to obtain the corresponding relationship between the brain activation change and the severity of the frozen gait.

9. The fNIRS-based method for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm according to claim 1, characterized in that, The frozen gait prediction model is based on the gradient boosting decision tree model GBDT. The brain function activation indicators, the toe tapping scale scores, the clinical basic data and the gait data of three types of people, i.e., Parkinson's disease patients with freezing gait, Parkinson's disease patients without freezing gait and healthy elderly people, are collected when the people perform a toe tapping paradigm; the brain function activation indicators are standardized, the toe tapping scale scores are coded, the clinical basic data are filled with missing values and processed for abnormal values, and the gait data are normalized to form a comprehensive feature set; The comprehensive feature set is used to construct and train multiple gradient boosting decision trees, and the importance scores of each feature are extracted from the gradient boosting decision trees; the importance of the feature is measured by calculating its contribution to the reduction of the loss function as a splitting node in all trees; Each feature in the comprehensive feature set is sorted according to its importance in the model to evaluate and verify the severity of freezing gait; The original data set is randomly and uniformly divided into five subsets; Cross-validation is used to evaluate the performance of the model.

10. A system for assessing freezing of gait in Parkinson's disease under toe-tapping paradigm based on fNIRS, characterized in that, It comprises: A first data acquisition unit that records the brain activation changes of a subject performing a toe tapping paradigm using fNIRS to obtain the brain function activation indicators of the subject; The toe tapping paradigm includes: the subject comfortably sits on a chair with a straight back and handles, and places the heel on the ground, and taps the toe on the ground with the maximum amplitude and the fastest speed for 20 seconds; A second data acquisition unit that uses a toe tapping evaluation scale to obtain the toe tapping scale scores of the subject; A third data acquisition unit that obtains the clinical basic data and gait data of the subject; A prediction model construction unit that constructs a freezing gait prediction model based on machine learning; a data set formed by the brain function activation indicators, the toe tapping scale scores, the clinical basic data and the gait data of three types of people, i.e., Parkinson's disease patients with freezing gait, Parkinson's disease patients without freezing gait and healthy elderly people, when the people perform a toe tapping paradigm is used to train the model; A freezing gait evaluation unit that uses the brain function activation indicators, the toe tapping scale scores, the clinical basic data and the gait data of the subject as model inputs, and uses the trained freezing gait prediction model to predict the occurrence of freezing gait in Parkinson's disease patients and evaluate the severity of freezing gait.

Images