CN105310695B - Dyskinesia Assessment Equipment - Google Patents

Abstract

Present invention discloses a kind of unusual fluctuation disease assessment equipments, including external sensing device, after the equipment acquires external sensing device data, noise is removed with high-pass filtering and low-pass filtering, choose suitable time window data intercept, exercise intensity is extracted from interception section, the characteristic of frequency and times of exercise, or, interception segment data is integrated to extract exercise intensity again after obtaining speed or angle, the characteristic of frequency and times of exercise, it is again that assessment models are established in reference with unusual fluctuation disease marking scales, according to the assessment models, the characteristic of extraction is corresponded into unusual fluctuation disease menace level.

Description

Dyskinesia Assessment Equipment

technical field

The invention belongs to the technical field of medical devices, and in particular relates to a monitoring device capable of evaluating dyskinesia and a signal analysis method thereof.

Background technique

Dyskinesia is an abnormal involuntary movement, including chorea, throwing movement, choreoathetosis and movement disorders, etc., often manifested as involuntary or dance-like hyperactivity or dystonia of the trunk and limbs , often appear in diseases such as Huntington's disease, Parkinson's disease treated with long-term dopamine drugs, and delayed dyskinesia treated with neurological agents for a long time. This abnormal action cannot be controlled by the patient itself, so it brings great inconvenience and even pain to the patient.

The current clinical assessment scales for dyskinesias are not perfect. For example, the AIMS assessment scale mainly evaluates the degree of dyskinesia in patients, and the Goetz assessment scale mainly assesses the impact of dyskinesias on patients' lives. In addition, dyskinesias may not necessarily appear during clinical evaluation. For example, dyskinesias in Parkinson's disease only appear during the peak dosage period of medication and the fading period of drug effects. Therefore, the evaluation mainly relies on the patient's memory, and the evaluation results are subject to the subjective opinions of patients and doctors. Great impact.

The shortcomings of the current assessment algorithm for dyskinesias are: in daily life, the accuracy of assessment is low when patients perform active movements, such as walking; poorly separable from other symptoms of patients, such as slowness of movement; the assessment scale is not perfect, Evaluation models are often established with a single evaluation scale as the "gold standard".

Contents of the invention

The purpose of the present invention is to allow patients with dyskinesias to obtain accurate and timely treatment.

In order to achieve the purpose of the above invention, the present invention provides a device for evaluating dyskinesia, including an in vitro sensing device, the device is configured to perform the following steps:

Collect data from external sensor devices;

High-pass filtering and low-pass filtering to remove noise; the cut-off frequency of the low-pass filter is 3.5-4Hz to remove rest tremor components; the cut-off frequency of the high-pass filter is 0.05Hz-1Hz to remove linear drift;

Continuously collect sensor data with a time window of S;

Use the sliding time window S1 to intercept the sensor data whose time window length is S, S1<S, extract the characteristic data of motion intensity, frequency and number of motions from the intercepted section, or, integrate the intercepted section data to obtain the speed or angle and then Extract the characteristic data of exercise intensity, frequency and exercise times;

Establishing an assessment model with reference to the dyskinesia scoring scale, associating the characteristic data with the severity of dyskinesias;

Corresponding the extracted feature data to the severity level of dyskinesia according to the assessment model;

Wherein, the in vitro sensing device includes a single-axis or multi-axis accelerometer and/or gyroscope;

The frequency feature data of the movement includes line length features, half-wave features, or performing fast Fourier transform on the intercepted data segment to obtain its spectrum information, and extracting the main frequency or main frequency bandwidth Wp corresponding to the maximum spectrum;

The characteristic data of the frequency of motion in the sliding time window S1 is the ratio of the frequency energy in the main frequency bandwidth Wp to the overall energy; or, the step of extracting the characteristic data of the frequency of motion in the sliding time window S1 includes, averaging the S1 points Divide into sub-data segments of point S2, calculate the mean value of each sub-data segment, and then calculate the ratio or difference between the mean value of the S2 sub-data segment and the mean value of S1.

As a further improvement of an embodiment of the present invention, the step of corresponding the extracted feature data to the severity level of dyskinesia includes using a classifier to analyze the feature data according to the evaluation model.

As a further improvement of an embodiment of the present invention, the step of corresponding the extracted feature data to the severity level of dyskinesia includes performing unsupervised cluster analysis on the feature data, or performing multivariate regression on the feature data and setting a fixed threshold or dynamic Threshold, or set a fixed threshold or dynamic threshold for a single feature data, and then use the AND gate, OR gate, and NOT gate to make logical judgments.

As a further improvement of an embodiment of the present invention, the time window length S is 30 seconds to 10 minutes.

As a further improvement of an embodiment of the present invention, the feature data of exercise intensity includes root mean square, Hilbert transform envelope, cumulative energy, mean square sum, line length feature or area feature.

As a further improvement of an embodiment of the present invention, the dyskinesia evaluation device further includes an implanted device in the body, and the device is configured to perform another step: transmitting a stimulation parameter adjustment signal to the body according to the evaluation result of the dyskinesia severity implanted device.

Compared with the prior art, the dyskinesia evaluation equipment provided by the present invention can extract more stable and separable feature data to distinguish dyskinesias from slowness of movement, muscle rigidity, and active movements, and can be used to evaluate the effectiveness of dyskinesia treatment programs. Sex, to guide doctors to adjust medication, or to adjust the stimulation program of implanted devices in the body.

Description of drawings

FIG. 1 is a schematic structural view of an embodiment of the dyskinesia assessment device of the present invention;

Fig. 2 is an evaluation flow chart of the dyskinesia evaluation device of the present invention;

3 is a schematic diagram of a time window for extracting features associated with dyskinesias in an embodiment of the present invention;

FIG. 4 is a schematic diagram of a main frequency Fp and a main frequency bandwidth Wp in an embodiment of the present invention.

Detailed ways

The present invention will be described in detail below in conjunction with specific embodiments shown in the accompanying drawings. However, these embodiments do not limit the present invention, and any structural, method, or functional changes made by those skilled in the art according to these embodiments are included in the protection scope of the present invention.

The difficulty in evaluating dyskinesias lies in the fact that the frequency range of their movements overlaps with those of slowness of action, myotonia, and active movements, and it is difficult to distinguish them simply by filtering. Moreover, the difference in the range of motion between dyskinesias and active movements is small, so it is necessary to extract stable, rigid, and active movements. The features with good separability are analyzed. The basis of the present invention is that the main frequency ranges of dyskinesia and slowness of action, muscle rigidity and active movement are different, so the data of different frequency ranges can be extracted respectively, such as 1-1.5Hz, 1.5-2Hz, 2-2.5 Hz, etc., can be used but not limited to the ratio of energy in the two frequency ranges. Within a fixed time window, the number of movements of dyskinesias is more than that of normal movements, for example, 5 times of active movements within 2 minutes, while dyskinesias are uncontrolled, and may occur 10 or more times.

Specifically, the present invention provides a device for evaluating dyskinesia, which includes an in vitro sensing device. The in vitro sensing device may be worn on, but not limited to, the wrist, ankle, torso or shoulder. Depending on where the dyskinesias are predominant, single or multiple external sensing devices may be worn.

In order to effectively distinguish dyskinesias from slowness of movement, muscle rigidity, and active movements, the motion data collected by the present invention include the following four categories: in common actions of daily life, such as sitting, walking, eating, drinking, etc., abnormal movements of different severities The movement data of dyskinesia, the movement data of slowness of movement and muscle rigidity without dyskinesia, the movement data of normal control group, and the movement data of different severity of dyskinesias during quiet rest.

As shown in FIG. 1 , the in vitro sensing device includes a sensor, an A/D converter, a processor, a memory, and a communication module that are electrically connected in sequence. Wherein, the memory may be a non-volatile memory or a volatile memory. The processor is preferably coupled with the sensor to monitor the state of the patient and perform signal acquisition and processing according to the programming program stored in the memory.

The communication module is used for information transmission between the external sensing device and a computer, a patient controller, an implanted device in the body or other external devices. Information transmission can be one-way or two-way. The communication module includes a wireless communication module and/or a wired communication module. The wireless communication module can be used for information transmission with implanted devices in the body, and can also be connected with other external devices through a wireless network. The wired communication module is connected with a computer or other carriers through a wired network.

The invention uses a built-in or an external power supply to supply power to the external sensing device. The built-in power supply is a battery pack or batteries, which can be dry batteries or rechargeable batteries. Preference is given to using lithium batteries (eg lithium ion or lithium polymer). If rechargeable, the battery is coupled to a charging circuit so that the battery is periodically coupled to an external power source for charging.

The sensors are single or multi-axis accelerometers and/or gyroscopes.

Dyskinesia assessment devices also include internally implanted devices. The implanted device in the body includes a stimulation module, which can send electrical stimulation to the patient through electrodes to treat or alleviate dyskinesia. The implanted device in the body also includes a processor responsible for data processing. For example, a classifier can be programmed into an implanted device to characterize the severity of dyskinesias based on the analyzed data. The processor can also send instructions to the stimulation module according to the result of data processing. The implanted device in the body is also provided with a wireless communication module for two-way signal transmission with the external sensing device.

When necessary, the dyskinesia assessment equipment may also include a data processing device, such as a PC. Although the external sensing device and the internal implanted device can complete the data collection and analysis process, the advantage of setting up a separate data processing device for analysis and calculation is that it can reduce the power consumption of the external sensing device and the internal implanted device, prolong the The battery life time is also convenient for medical staff to monitor and adjust operations.

The dyskinesia assessment device of the present invention also includes a filter, which can be selectively arranged in an external sensing device or an internal implant device or a data processing device, or be electrically connected to the external sensing device as a separate external device. Preferably, the filter is a high-pass filter and a low-pass filter arranged in pairs, or a band-pass filter.

Fig. 2 schematically shows a flow chart of the dyskinesia assessment device of the present invention. The method uses an external sensing device to collect sensor data, including single or multiple single-axis or multi-axis acceleration data and/or gyroscope data.

Sensor data may include linear drift or other DC components, which are removed by high-pass filtering. For Parkinson's disease, the frequency range of rest tremor is 4-6Hz, while the frequency of dyskinesia is lower, so the sensor data also needs to be low-pass filtered to remove the rest tremor component. High-pass and low-pass filters can be used, or band-pass filters can be used instead, to extract motion data in the frequency range of dyskinesias. Filter types include, but are not limited to, Chebyshev filters, Butterworth filters, elliptic filters, empirical mode decomposition, or wavelet transforms.

As shown in Figure 3, the sensor data of the observation time window S is collected continuously for analysis. Among them, if the window length is too short, it will contain less symptom-related information, which will lead to an increase in the false positive rate; on the other hand, due to the influence of active motion, if the window length is too long, the detection accuracy and accuracy will be reduced.

Sliding time window S1 is used to intercept the data of observation time window S. Extract the characteristic data related to exercise intensity, frequency and number of times of movement in a specific time in the intercepted data segment, or integrate the intercepted data to obtain speed or angle, and then extract the data related to exercise intensity, frequency and number of times of movement in a specific time, etc. related feature data. The characteristic data can be selected but not limited to root mean square, main frequency, the ratio of main frequency energy to overall energy, half-wave algorithm, area algorithm, line length algorithm, etc. can be used. If multiple sensors are used, cross-correlation analysis can be performed on each sensor, energy or velocity ratios in different frequency ranges can be calculated, or coherent analysis can be performed.

Then, correlate the collected feature data with the disease state to evaluate the severity of dyskinesia. The associating step includes measuring the extracted feature data with a scoring scale to obtain a corresponding relationship between the feature data and the severity of dyskinesia. Specifically, multiple evaluation scales are used to establish evaluation models, and different weights are assigned to each evaluation result according to the meaning of each evaluation scale to obtain a regression model, thereby establishing a robust evaluation model. The scoring scales are various international dyskinesia assessment scales, including but not limited to AIMS, Goetz, UPDRS, etc. Using the rating scale to establish an evaluation model can make the evaluation results output by this method consistent with the evaluation results of the current standard, which facilitates the promotion of this method in the existing medical system and avoids self-built standards.

Then, the extracted data are evaluated using the evaluation model described above. Evaluation can use classifiers, unsupervised cluster analysis can be performed, multivariate regression can be performed on feature variables, and fixed thresholds or dynamic thresholds can be set, or fixed thresholds or dynamic thresholds can be set for a single feature variable, and finally the AND gate is used , OR gate, and NOT gate for logical judgment. Wherein, the classifier type includes but not limited to support vector machine, artificial neural network, and extreme learning machine. The classifier program can be set in the processor of the sensor device outside the body, or in the implanted device or the data processing device inside the body. Preferably, the classifier is set in an implanted device or a data processing device in the body.

The method for evaluating dyskinesia by the device of the present invention is illustrated below with an example.

Set the high-pass filter to remove linear drift, and the cut-off frequency of the high-pass filter can be 0.05Hz-1Hz.

Set the low-pass filter to remove tremor components. For example, the highest frequency of dyskinesia in Parkinson's disease is 4Hz, while the frequency range of rest tremor is 4-6Hz. Therefore, low-pass filtering the acceleration data can remove the rest tremor component, and the cut-off frequency of low-pass filter Can be 3.5-4Hz.

The above high-pass filtering and low-pass filtering steps do not need to be performed in a specific order.

The high-pass filter and low-pass filter can be replaced by band-pass filters with the cut-off frequency set as above to extract motion data within the dyskinesia frequency range.

Optimize the length of the observation time window, and use the observation time window S of the same length to extract features. Preferably, according to different clinical manifestations of patients, the window length can be adopted but not limited to 30s, 1min, 2min, 5min or 10min, etc. There can be a certain overlap rate between the continuous observation time windows S. The length of the above-mentioned time window can reduce the false positive rate, and contains more information, which removes the influence of slow action or active movement to a certain extent, and increases the detection precision and accuracy.

Extract features associated with disease states. If the data within the observation time window S is directly analyzed, information on the time scale may be lost. The present invention adopts the sliding time window S1 of the same length (for example, the number of sampling points S1=500), cuts the observation data into data segments, as shown in Figure 3, and extracts the movement intensity, main frequency and movement within a specific time in each data segment Frequency and other related features, this approach can increase temporal precision and more accurately distinguish dyskinesias from active movements.

Extracting motion intensity features may use time domain features such as root mean square (RMS), Hilbert transform envelope, cumulative energy, or mean square sum, or line length features or area features. where the root mean square expression for two or three axes is:

The frequency features of motion can be extracted using time-domain features such as line length and half-wave, or as shown in Figure 4, fast Fourier transform is performed on the data segment to obtain its spectrum information, and the main frequency Fp and main frequency bandwidth corresponding to the maximum spectrum are extracted. Wp.

To extract the number of motions within the sliding time window S1, the ratio of the frequency energy within the main frequency bandwidth Wp to the overall energy can be used. Another method for extracting the number of movements in the sliding time window S1 is to first divide point S1 into sub-data segments of point S2, calculate the average value of each sub-data segment to reduce background noise, and then obtain the sub-data segment of S2 The ratio or difference between the mean and the S1 mean. A larger ratio or difference indicates a greater likelihood of movement. Then, in order to determine active movement and dyskinesia, different thresholds are set for active movement and dyskinesia, and the sub-data segments in the above S1 that exceed the threshold are counted, and the number of sub-data segments is recorded as Rj, or its percentage of time in S1 for Pj.

The aforementioned features associated with dyskinesias are extracted within the observation time window S to obtain multiple feature matrices. In order to select features with high sensitivity or high correlation with dyskinesia, the features are sorted. The sorting method can use each feature matrix as the input of the subsequent classifier, using k-fold cross-validation or leave-one-out cross-validation, or Use methods such as maximum correlation and minimum redundancy criteria, correlation or distance.

The dyskinesia rating scale AIMS, Goetz, or UPDRS was used to verify the onset of dyskinesias combined with the selected correlation features, and an evaluation model that could correspond to the selected correlation features and the severity of dyskinesias 0-4 or 0-3 was obtained.

For example, correlating rapid objective movements of the upper extremities to AIMS scales of 0 (normal), 1 (slight, possibly normal), 2 (slight), 3 (moderate), 4 (severe), dyskinesias Percentage of time present in the waking state of the day correlates to 0 (normal), 1 (1%-25%), 2 (26%-50%), 3 (51%-75%), 4 on the UPDRS scale (76%~100%). Based on this, the criteria for evaluating the associated features are formulated.

Based on this evaluation model, program classifiers such as support vector machines, artificial neural networks, or extreme learning machines, and input the selected associated features into the classifier, and the output result will be the severity of dyskinesia 0-4 or 0-3. Alternatively, set a fixed threshold or a dynamic threshold for the associated features, define the severity of dyskinesias within the corresponding threshold range according to the evaluation model established in the previous step, and use AND gates, OR gates, and NOT gates to logically judge the severity of the output dyskinesias degree evaluation.

According to the evaluation result of dyskinesia severity, the external sensor device or patient controller or PC transmits the stimulation parameter adjustment signal to the implanted device in the body, or the implanted device in the body completes the final data processing step and sends the stimulation parameters to the stimulation module Adjusting the signal can automatically select the preset program control mode in the implanted device in the body, and can also adjust the stimulation program of the implanted device in the body in real time.

The present invention can save motion information as data, and can detect dyskinesias automatically in real time, or transmit the formed data to PC or other carriers through a certain way, such as wired communication or wireless communication, to determine the severity of dyskinesias. The judgment result can assist in the diagnosis of the disease, such as evaluating the effectiveness of the treatment method, adjusting the patient's medication according to the judgment result, adjusting the stimulation program of the implanted medical device in real time, or automatically selecting the preset stimulation program in the implanted medical device, so that The patient's dyskinesias were treated in a timely manner.

It should be understood that although this description is described according to implementation modes, not each implementation mode only contains an independent technical solution, and this description in the description is only for clarity, and those skilled in the art should take the description as a whole, and each The technical solutions in the embodiments can also be properly combined to form other embodiments that can be understood by those skilled in the art.

The series of detailed descriptions listed above are only specific descriptions for feasible implementations of the present invention, and they are not intended to limit the protection scope of the present invention. Any equivalent implementation or implementation that does not depart from the technical spirit of the present invention All changes should be included within the protection scope of the present invention.

Claims (6)

1. a kind of unusual fluctuation disease assessment equipment, including external sensing device, which is characterized in that the equipment is configured as executing following Step:

Acquire external sensing device data；

High-pass filtering and low-pass filtering remove noise；The cutoff frequency of the low-pass filtering is 3.5-4Hz to remove tranquillization shake Quiver ingredient;The cutoff frequency of the high-pass filter is 0.05Hz-1Hz to remove linear drift;

Continuous acquisition time window is the sensing data of S；

The time window is intercepted as the sensing data of S with time slip-window S1, and S1 < S extracts movement by force from interception section The characteristic of degree, frequency and times of exercise, alternatively, integrating to extract movement again by force after obtaining speed or angle to interception segment data The characteristic of degree, frequency and times of exercise；

It is to be associated with the characteristic and unusual fluctuation disease severity referring to assessment models are established with unusual fluctuation disease marking scales；

According to the assessment models, the characteristic of extraction is corresponded into unusual fluctuation disease menace level；

Wherein, the external sensing device includes uniaxial or multiaxis accelerometer and/or gyroscope;

The frequency characterization data of the movement includes wire length feature, half wave characteristic, or carries out quick Fu to the data intercept section In leaf transformation obtain its spectrum information, extract the corresponding basic frequency of maximum spectrum or basic frequency bandwidth Wp;

Times of exercise characteristic in the time slip-window S1 is frequency energy and totality in the basic frequency bandwidth Wp The ratio of energy;Or, the step of times of exercise characteristic extracted in time slip-window S1 includes being divided into S1 point The subdata section of S2 point calculates the mean value of each subdata section, then seeks the ratio or difference of S2 subdata section mean value and S1 mean value Value.

2. unusual fluctuation disease assessment equipment according to claim 1, which is characterized in that described to correspond to the characteristic of extraction The step of unusual fluctuation disease menace level includes analyzing characteristic according to the assessment models with classifier.

3. unusual fluctuation disease assessment equipment according to claim 1, which is characterized in that described to correspond to the characteristic of extraction The step of unusual fluctuation disease menace level includes unsupervised clustering being carried out to characteristic, or carry out to characteristic changeable Amount returns and is arranged fixed threshold or dynamic threshold, or fixed threshold or dynamic threshold is arranged to single feature data, then adopts Logic judgment is carried out with door or door, NOT gate.

4. unusual fluctuation disease assessment equipment according to claim 1, which is characterized in that the time window S be 30 seconds -10 points Clock.

5. unusual fluctuation disease assessment equipment according to claim 1, which is characterized in that the exercise intensity characteristic includes equal Root, Hilbert transform envelope, cumlative energy, side and wire length feature or area features.

6. unusual fluctuation disease assessment equipment according to claim 1, which is characterized in that it further include et al. Ke device, it is described to set It is standby to be configured as can also carry out another step: according to unusual fluctuation disease severity evaluation result, stimulation parameter adjustment signal being transmitted To et al. Ke device.