Medical equipment work frequency metering method and equipment performance management system

Abstract

The invention provides a medical equipment work frequency metering method and an equipment performance management system. The medical equipment working frequency metering method comprises the following steps: acquiring electrical parameter detection data and auxiliary detection data of medical equipment, and recording result data provided by a system; inputting the electrical parameter detection data to an electrical parameter identification model corresponding to the equipment type to obtain electrical parameter working data; inputting the auxiliary detection data to an auxiliary identification model corresponding to the equipment type to obtain auxiliary working data; comprehensively processing the electrical parameter working data and the auxiliary working data to obtain original data; the actual working times of the medical equipment in the metering period are obtained through mutual verification of the original data and the result data. The auxiliary working data is utilized to assist the electrical parameter working data to obtain the original data as accurate as possible, and the original data and the result data output by the existing recording system of the hospital are mutually verified, so that the accuracy and objectivity of the obtained actual working times are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to a method for measuring the working times of medical devices and a device performance management system. Background Art

[0002] The investment in medical devices, especially the investment and use of large-scale radiological medical devices (mainly including DR devices, gastrointestinal machines, CT devices, MRI devices, DSA devices, mobile DRs, etc.), is directly related to the improvement of the medical quality, economic benefits, and medical expenses of hospitals. Accurately evaluating the performance of medical devices can help hospitals manage and make decisions in the whole life cycle process of large-scale medical devices (especially large-scale radiological medical devices), including budgeting, purchasing, using, maintaining, and scrapping, and provide valuable methods and suggestions for hospitals to save the operating costs of medical devices and improve the efficiency of medical devices.

[0003] Accurately measuring the actual working times of medical devices is a prerequisite for evaluating the performance of medical devices. At present, generally, the result data of medical device examinations are read from existing hospital record systems such as HIS systems, PACS systems, LIS systems, and HRP systems, and the result data are used as the working times of medical devices. In practice, there are situations where patients do not cooperate well and repeat examinations, but the record system only records the result of one examination, patients jump the queue for examinations and the record system does not record, and the record system makes recording mistakes due to network or equipment failures. As a result, the result data cannot truly reflect the working times of medical devices. If the result data are not verified and corrected, it will lead to inaccurate evaluation of the performance of medical devices. Summary of the Invention

[0004] The present invention aims to solve the technical problem in the prior art that it is inaccurate to determine the working times of medical devices based on the result data of medical device examinations read from existing hospital record systems, which in turn leads to inaccurate evaluation of medical devices, and provides a method for measuring the working times of medical devices and a device performance management system.

[0005] To achieve the above object of the present invention, according to the first aspect of the present invention, the present invention provides a method for measuring the working times of a medical device, including: determining the device type of the medical device, obtaining the electrical parameter detection data and auxiliary detection data of the medical device within a measurement period, and recording the result data provided by the system, wherein the auxiliary detection data includes video data and / or audio data of the examination room where the medical device is located; inputting the electrical parameter detection data into a pre-trained electrical parameter recognition model corresponding to the device type to obtain electrical parameter working data; the electrical parameter working data includes the number of electrical parameter recognition operations and the time interval of each electrical parameter recognition operation; inputting the auxiliary detection data into a pre-trained auxiliary recognition model corresponding to the device type to obtain auxiliary working data; the auxiliary working data includes the number of auxiliary recognition operations and the time interval of each auxiliary recognition operation; comprehensively processing the electrical parameter working data and the auxiliary working data to obtain raw data, the raw data includes the number of raw operations and the time interval of each raw operation; using the raw data and the result data to mutually verify to obtain the actual working times of the medical device within the measurement period.

[0006] The above technical solution: Collect the electrical parameter detection data and auxiliary detection data of the medical device, use the pre-trained electrical parameter recognition model corresponding to the device type to identify the electrical parameter detection data to obtain electrical parameter working data, use the pre-trained auxiliary recognition model corresponding to the device type to identify the auxiliary detection data to obtain auxiliary working data. In the comprehensive processing, use the auxiliary working data to assist the electrical parameter working data to obtain as accurate raw data as possible, so that the raw data can reflect the actual working conditions of the medical device, and mutually verify the raw data with the result data output by the hospital's existing record system, which not only makes up for the problem of missing records in the result data, but also can correct the detection deviation of the intelligent model, enabling it to continuously learn and improve the detection accuracy, ensuring the accuracy and objectivity of the obtained actual working times.

[0007] To achieve the above object of the present invention, according to the second aspect of the present invention, the present invention provides a method for evaluating the performance of a medical device, including: obtaining the actual working times of the medical device to be evaluated within each measurement period during the evaluation period according to the method for measuring the working times of the medical device described in the first aspect of the present invention; obtaining the total actual working times of the medical device to be evaluated during the evaluation period; obtaining the performance index value of the medical device to be evaluated based on the total actual working times using a preset performance index calculation method.

[0008] The above technical solution: Can obtain accurate and reliable performance index values of the medical device, and effectively evaluate the performance of the medical device.

[0009] To achieve the above object of the present invention, according to the third aspect of the present invention, there is provided a medical device performance management system, including: a plurality of electrical parameter sensors for respectively measuring the electrical parameters of different medical devices in operation; a plurality of cameras and / or a plurality of sound pickup devices, with the plurality of cameras respectively installed in the examination rooms where different medical devices are located, and the plurality of sound pickup devices respectively installed in the examination rooms where different medical devices are located; a performance management platform that communicates with the electrical parameter sensors, cameras, and sound pickup devices respectively, and is used for collecting electrical parameter detection data and auxiliary detection data; the performance management platform executes the steps of the medical device working times measurement method described in the first aspect of the present invention, or the performance management platform executes the steps of the medical device performance evaluation method described in the second aspect of the present invention.

[0010] The above technical solution: collect the electrical parameter detection data and auxiliary detection data of medical devices, use the pre-trained electrical parameter recognition model corresponding to the device type to identify the electrical parameter detection data to obtain electrical parameter working data, use the pre-trained auxiliary recognition model corresponding to the device type to identify the auxiliary detection data to obtain auxiliary working data, and use the auxiliary working data to assist the electrical parameter working data in comprehensive processing to obtain as accurate original data as possible, so that the original data can reflect the actual working conditions of medical devices, and mutually verify the original data with the result data output by the hospital's existing record system, which not only makes up for the problem of missing records in the result data, but also can correct the detection deviation of the intelligent model, enabling it to continuously learn and improve the detection accuracy, ensuring the accuracy and objectivity of the obtained actual working times, and further being able to obtain accurate and reliable performance index values of medical devices, and effectively evaluating the performance of medical devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 is a flowchart of the medical device working times measurement method in a preferred embodiment of the present invention;

[0012] Figure 2 is a structural block diagram of the medical device performance management system in a preferred embodiment of the present invention;

[0013] Figure 3 is a deployment diagram of the medical device performance management system in a preferred embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions from beginning to end. The embodiments described below with reference to the drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation of the present invention.

[0015] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention.

[0016] In the description of the present invention, unless otherwise specified and defined, it should be noted that the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it may be a mechanical connection or an electrical connection, or it may be the communication inside two elements. It may be directly connected or indirectly connected through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0017] The present invention discloses a method for measuring the working times of a medical device. In a preferred embodiment, its process is as Figure 1 shown, including:

[0018] Step S1: Determine the device type of the medical device, obtain the electrical parameter detection data and auxiliary detection data of the medical device within the measurement period, and the result data provided by the recording system. The auxiliary detection data includes video data and / or audio data of the examination room where the medical device is located.

[0019] In this embodiment, each medical device in the hospital has a unique ID number. Information such as the ID number, device type, working examination room, examination room number, and examination room location of each medical device is stored in the database. For example, the device types of radiological medical devices include DR, gastrointestinal machine, CT, MRI, DSA, and mobile DR, etc. The recording system mainly refers to the existing PACS, HIS, HRP, and hospital information platforms in the hospital, on which information such as patient information, examination time interval, and examination cost for each examination of all medical devices is recorded. The result data can be extracted from these information, and the result data includes the result working times recorded in the recorded information and the time interval of each result work.

[0020] In this embodiment, the electrical parameter detection data includes current data and / or power data of the main power of the medical device. Preferably, the electrical parameter detection data further includes temperature data at the medical device housing or at key internal components, and it is determined whether the temperature data falls within a preset normal temperature range. If the temperature data falls within the normal temperature range, it is considered that the temperature of the medical device is normal. If the temperature data does not fall within the normal temperature range, the superior system is reported or relevant staff is alerted. Further preferably, the electrical parameter detection data further includes voltage data of the main power of the medical device, and it is determined whether the voltage data falls within a preset normal voltage range. When the voltage data falls within the normal voltage range, it is considered that the medical device has normal power supply. When the voltage data does not fall within the normal voltage range, if the voltage is too low, undervoltage protection is enabled, and if the voltage is too high, overvoltage protection is enabled.

[0021] In this embodiment, for large-scale radiation medical devices, each medical device generally operates in an independent examination room, such as a magnetic resonance examination room, a CT examination room, etc. A camera is installed in the examination room to capture the internal video of the examination room, and information such as the patient's posture and the color change of the status indicator light of the medical device can be captured through the video. A sound pickup device such as a microphone is installed in the examination room. Since the voiceprint characteristics of the medical device are different in the case of failure, normal operation, and standby state, it is possible to determine whether the medical device is operating normally by comparing the voiceprint characteristics extracted from the audio data output from the sound pickup device with the voiceprint characteristics of normal operation.

[0022] Step S2, input the electrical parameter detection data into the pre-trained electrical parameter recognition model corresponding to the device type to obtain electrical parameter working data; the electrical parameter working data includes the number of electrical parameter recognition operations and the time interval of each electrical parameter recognition operation.

[0023] Step S3, input the auxiliary detection data into the pre-trained auxiliary recognition model corresponding to the device type to obtain auxiliary working data; the auxiliary working data includes the number of auxiliary recognition operations and the time interval of each auxiliary recognition operation.

[0024] In this embodiment, Step S2 and Step S3 can be executed synchronously or sequentially. A model library is preset, and the model library stores the pre-trained electrical parameter recognition models and auxiliary detection models corresponding to different device types (the auxiliary detection models include visual recognition models and / or voiceprint recognition models).

[0025] In this embodiment, the specific form of the electrical parameter detection data is a time series of current or power. The electrical parameter recognition model adopts an existing LSTM-RNN network structure, where RNN represents a recurrent neural network and LSTM represents a long short-term memory network, and the output is a sequence of judgment results. The electrical parameter recognition model compares the electrical parameter value at each moment with the electrical parameter value of the medical device during normal inspection work to obtain the judgment result at each moment. For example, when the electrical parameter is the current of the medical device, the electrical parameter detection data is represented as a sequence: I 0 I 1 …I N1 , I 0 、I 1 、I N1 represent the current values at the 0th moment, the 1st moment, and the N 1 th moment respectively. Input the electrical parameter detection data into the electrical parameter recognition model to obtain a sequence of judgment results corresponding to the time of the electrical parameter detection data. N Y…Y, where N indicates that the medical device is not performing normal inspection work, and Y indicates that the medical device is performing normal inspection work. Traverse the sequence of judgment results in turn. The change from the first N to Y represents the start of the first electrical parameter recognition work, record the start time of the first electrical parameter recognition work, and the change from the first Y to N after the change from the first N to Y represents the end of the first electrical parameter recognition work, record the end time of the first electrical parameter recognition work, obtain the time interval of the first electrical parameter recognition work, traverse the sequence of judgment results in turn until the end, and organize to obtain the electrical parameter work data.

[0026] Step S4, comprehensively process the electrical parameter work data and the auxiliary work data to obtain the original data, where the original data includes the original number of work times and the time interval of each original work.

[0027] In this embodiment, to improve the processing speed, the original data can be obtained by sequentially obtaining the intersection of the time intervals of the electrical parameter work data and the auxiliary work data. The number of intersections obtained is used as the original number of work times, and the time interval of each intersection is used as the time interval of the original work.

[0028] Step S5, use the original data and the result data to mutually verify to obtain the actual number of work times of the medical device within the metering period.

[0029] In a preferred embodiment, for an examination room where a camera can be installed, a camera can be installed in the examination room. When the auxiliary detection data consists of video data of the examination room where the medical device is located, the auxiliary work data consists of visual work data. In step S3, input the auxiliary detection data into a pre-trained auxiliary recognition model corresponding to the device type to obtain the auxiliary work data, including: input the video data into a pre-trained visual recognition model corresponding to the device type to obtain the visual work data; the visual work data includes the number of visual recognition work times and the time interval of each visual recognition work.

[0030] In this embodiment, a trained visual recognition model corresponding to the device type is stored in the model library. The visual recognition model includes a human pose recognition model. The video data is converted into an image sequence, and the pictures in the image sequence are sequentially input into the human pose recognition model. The human pose recognition model outputs the inspection and judgment results corresponding to each picture. When the pose of the person in the picture meets the inspection pose corresponding to the device type, a judgment result of YES is output. When the pose of the person in the picture does not meet the inspection pose corresponding to the device type, a judgment result of NO is output. Therefore, a sequence of pose judgment results corresponding to the time of the image sequence is finally obtained. Referring to the above process of obtaining the electrical parameter working data, the jumps from NO to YES and from YES to NO in the sequence of pose judgment results are sequentially identified, and the pose recognition data is obtained after sorting. The pose recognition data includes the number of pose recognition operations and the time interval corresponding to each pose recognition operation. The human pose recognition model can adopt an existing convolutional neural network (CNN) structure.

[0031] In this embodiment, since the pose of the patient during the examination may not be very standard, and the examination may be aborted due to non-cooperation during the examination, etc., it is not accurate enough to judge the working state of the medical device only relying on the patient's pose. Therefore, for medical devices equipped with working status indicators, such as magnetic resonance devices configured with working status indicators, the indicator shows different colors during normal examination work, standby, and preparation stages, and the video images collected also include the image of the indicator. Therefore, to improve the accuracy of visual recognition, the visual recognition model also includes an indicator status recognition model, and the image sequence is input into the indicator status recognition model to obtain status recognition data. The pictures in the image sequence are sequentially input into the indicator status recognition model. The indicator status recognition model outputs the inspection and judgment results corresponding to each picture. When the color of the indicator area in the picture is the indication color corresponding to the normal examination work of the device type, a judgment result of YES is output. When the color of the indicator area in the picture is not the indication color corresponding to the normal examination work of the device type, a judgment result of NO is output. Therefore, a sequence of color judgment results corresponding to the time of the image sequence is finally obtained. Referring to the above process of obtaining the electrical parameter working data, the jumps from NO to YES and from YES to NO in the sequence of color judgment results are sequentially identified, and the status recognition data is obtained after sorting. The status recognition data includes the number of indicator status recognition operations and the time interval corresponding to each indicator status recognition operation. The indicator status recognition model can adopt an existing convolutional neural network (CNN) structure.

[0032] In this embodiment, to improve the accuracy of visual working data, pose recognition data and status recognition data are fused to obtain visual working data. Specifically, on the time axis, the union of the time intervals in the pose recognition data and the status recognition data is obtained in sequence. The multiple discrete time intervals obtained after the union processing are the time intervals of different visual recognition operations respectively, and the number of discrete time intervals corresponds to the number of visual recognition operations.

[0033] In this embodiment, preferably, to ensure the accuracy of the original data, considering the actual situation, the confidence level of the electrical parameter working data is set higher than that of the visual working data, and the following determination rules are set to accurately obtain the original data. Specifically, step S4 includes:

[0034] Step S411: Determine the first part of the original data, which consists of the electrical parameter working data. Set the determination label of the first part of the original data as the first determination. The number of original operations in the first part of the original data is equal to the number of electrical parameter recognition operations in the electrical parameter working data. The time interval of the original operation is the time interval of the electrical parameter recognition operation. The determination label of each original operation in the first part of the original data is the first determination.

[0035] Step S412: Determine the second part of the original data, which consists of part of the data in the auxiliary working data that meets the first preset condition. Set the determination label of the second part of the original data as the second determination; wherein, the first preset condition is that the time interval of the i-th auxiliary recognition operation in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter working data, and i is the index of the number of auxiliary recognition operations in the second part of the original data, and i is a positive integer.

[0036] In this embodiment, the auxiliary working data consists of the visual working data. Therefore, in step S412, the second part of the original data consists of part of the data in the visual working data that meets the first preset condition. Set the determination label of the second part of the original data as the second determination; wherein, the first preset condition is that the time interval of the i-th visual recognition operation in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter working data, and i is the index of the number of visual recognition operations in the second part of the original data.

[0037] In this embodiment, a lower confidence level is set for the visual working data. When only visual recognition of the medical device working is detected in a certain time interval, that is, the time interval of this visual recognition operation has no intersection with all the time intervals corresponding to the electrical parameter working data, reaching the first preset condition, at this time, this visual recognition is regarded as an original operation, and the time interval of this visual recognition is used as the time interval of this original operation, and the determination label of this original operation is set as the second determination, and then it is verified in combination with the result data.

[0038] Step S413, combine the first part of the original data and the second part of the original data to obtain the original data.

[0039] In a preferred embodiment, for the examination room of medical devices where it is inconvenient to install a camera, such as the examination room of MRI (Magnetic Resonance Imaging). At this time, the auxiliary detection data consists of the audio data of the examination room where the medical device is located, and the auxiliary work data consists of voiceprint work data. In step S3, input the auxiliary detection data into the pre-trained auxiliary recognition model corresponding to the device type to obtain the auxiliary work data, including: input the audio data into the pre-trained voiceprint recognition model corresponding to the device type to obtain the voiceprint work data; the voiceprint work data includes the number of voiceprint recognition operations and the time interval of each voiceprint recognition operation.

[0040] In this embodiment, the voiceprint recognition model adopts an existing Convolutional Neural Network (CNN) structure. The audio data is frame-processed to obtain a continuous sequence of audio frames, and the audio frames in the audio frame sequence are sequentially input into the voiceprint recognition model. The voiceprint recognition model outputs the voiceprint judgment result corresponding to each audio frame. When the voiceprint feature in the audio frame matches the voiceprint feature of the normal inspection work corresponding to the set device type, a voiceprint judgment result of YES is output. When the voiceprint feature in the audio frame does not match the voiceprint feature of the normal inspection work corresponding to the set device type, a voiceprint judgment result of NO is output. Therefore, a sequence of voiceprint judgment results corresponding to the time of the audio frame sequence is finally obtained. Referring to the above process of obtaining the electrical parameter work data, the jumps from NO to YES and from YES to NO in the audio frame sequence are sequentially identified, and after sorting, the voiceprint work data is obtained. The voiceprint work data includes the number of voiceprint recognition operations and the time interval corresponding to each voiceprint recognition operation.

[0041] In this embodiment, preferably, to ensure the accuracy of the original data, combined with the actual situation, the confidence level of the electrical parameter work data is set higher than that of the voiceprint work data, and the following determination rules are set to achieve the accurate acquisition of the original data. Specifically, step S4 includes:

[0042] Step S421, determine the first part of the original data. The first part of the original data consists of the electrical parameter work data, and set the determination label of the first part of the original data as a first determination. The original number of operations in the first part of the original data is equal to the number of electrical parameter recognition operations in the electrical parameter work data, the time interval of the original work is the time interval of the electrical parameter recognition work, and the determination label of each original work in the first part of the original data is a first determination.

[0043] Step S422: Determine the second part of the original data. The second part of the original data consists of the part of the auxiliary work data that meets the first preset condition, and set the determination label of the second part of the original data as secondary determination. Among them, the first preset condition is that the time interval of the i-th auxiliary recognition work in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter work data, where i is the index of the auxiliary recognition work times in the second part of the original data and i is a positive integer.

[0044] In this embodiment, the auxiliary work data consists of voiceprint work data. Therefore, in step S422, the second part of the original data consists of the part of the voiceprint work data that meets the first preset condition, and set the determination label of the second part of the original data as secondary determination. Among them, the first preset condition is that the time interval of the i-th voiceprint recognition work in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter work data, where i is the index of the voiceprint recognition work times in the second part of the original data.

[0045] In this embodiment, a lower confidence level is set for the voiceprint work data. When only the voiceprint recognizes the operation of the medical device in a certain time interval, that is, the time interval of this voiceprint recognition work has no intersection with all the time intervals corresponding to the electrical parameter work data, meeting the first preset condition. At this time, this voiceprint recognition is regarded as an original work, and the time interval of this voiceprint recognition is used as the time interval of this original work, and set the determination label of this original work as secondary determination, and subsequent verification needs to be combined with the result data.

[0046] Step S423: Combine the first part of the original data and the second part of the original data to obtain the original data.

[0047] In a preferred embodiment, for the examination room that can install a camera, in order to improve the accuracy of the number of original works in the obtained original data, a sound pickup device can also be installed in the examination room to use audio data to increase the accuracy of the original data. At this time, the auxiliary detection data includes the video data and audio data of the examination room where the medical device is located, the auxiliary work data includes visual work data and voiceprint work data, and the auxiliary recognition model includes a visual recognition model and a voiceprint recognition model. In step S3, input the auxiliary detection data into the pre-trained auxiliary recognition model corresponding to the device type to obtain the auxiliary work data, including:

[0048] Step S31: Input the video data into the pre-trained visual recognition model corresponding to the device type to obtain the visual work data. The visual work data includes the number of visual recognition works and the time interval of each visual recognition work.

[0049] Step S32: Input the audio data into the pre-trained voiceprint recognition model corresponding to the device type to obtain voiceprint working data. The voiceprint working data includes the number of voiceprint recognition operations and the time intervals for each voiceprint recognition operation.

[0050] In this embodiment, preferably, to ensure the accuracy of the original data, considering the actual situation, the confidence level of the electrical parameter working data is set higher than that of the visual working data, and the confidence level of the visual working data is higher than that of the voiceprint working data. The following determination rules are set to accurately obtain the original data. Specifically, step S4 includes:

[0051] Step S431: Determine the first part of the original data. The first part of the original data consists of the electrical parameter working data and the part of the visual working data that meets the second preset condition. Set the determination label of the first part of the original data as a first determination. Among them, the second preset condition is that the time interval of the j-th visual recognition operation in the first part of the original data has no intersection with all the time intervals corresponding to the electrical parameter working data, and the time interval of the j-th visual recognition operation has an intersection with the time interval of the voiceprint recognition operation. j is the index of the number of visual recognition operations in the first part of the original data, and j is a positive integer.

[0052] In step S431, the first part of the original data includes two parts. The first part is the electrical parameter working data. In this part, the number of electrical parameter recognition operations is used as the number of original operations included in this part of the original data, and the time interval of the electrical parameter working data is used as the time interval of this part of the original data. The second part is the part of the visual working data that meets the second preset condition. This part is in the time interval when the electrical parameter recognition model fails to recognize the inspection work. If both the visual recognition model and the voiceprint recognition model recognize that the medical device work is performing a normal inspection work, then this inspection work is considered as an original work, and the determination label of this original work is a first determination. At the same time, the time interval recognized by the visual recognition model is used as the time interval of this original work. The visual recognition confidence level is higher than the voiceprint recognition level because audio signals are more likely to be mixed with interference signals.

[0053] Step S432: Determine the second part of the original data. The second part of the original data consists of the part of the visual working data that meets the third preset condition and the part of the voiceprint working data that meets the fourth preset condition. Set the determination label of the second part of the original data as secondary determination. Among them, the third preset condition is that the time interval of the k-th visual recognition work in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter working data, and the time interval of the k-th visual recognition work has no intersection with the time interval of the voiceprint recognition work. j is the index of the number of visual recognition works in the second part of the original data, and j is a positive integer. The fourth preset condition is that the time interval of the p-th voiceprint recognition work in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter working data, and the time interval of the p-th voiceprint recognition work has no intersection with the time interval of the visual recognition work. p is the index of the number of voiceprint recognition works in the second part of the original data, and p is a positive integer.

[0054] In this embodiment, step S432 means that in a certain time region on the time axis, if only the visual recognition model recognizes that the medical device performs a normal inspection work, or only the voiceprint recognition model recognizes that the medical device performs a normal inspection work, then this time region is used as the time interval of an original work, and the determination label of this original work is secondary determination. It is also necessary to verify the determination of this time interval in combination with the result data.

[0055] Step S433: Combine the first part of the original data and the second part of the original data to obtain the original data.

[0056] In a preferred embodiment, in step S5, the actual number of times the medical device works within the measurement period is obtained by mutually verifying the original data and the result data, including:

[0057] Step S51: The result data includes the result number of times of work and the time interval of each result work.

[0058] Step S52: If the determination label of the n-th original work in the original data is primary determination, then the n-th original work is regarded as an actual work.

[0059] Step S53: If the determination label of the n-th original work in the original data is secondary determination, and the time interval of the n-th original work has an intersection with the time interval of the m-th result work, then the n-th original work is regarded as an actual work.

[0060] If the determination label of the n-th original work in the original data is secondary determination, and the time interval of the n-th original work has no intersection with the time interval of the m-th result work, then the n-th original work is regarded as a misrecognition. Among them, n is the index of the number of original works, m is the index of the number of result works, and both n and m are positive integers.

[0061] In this embodiment, similarly, the confidence level set for the original work with a determination label of primary determination is higher than the result data. If the determination label is a primary determination, it is considered that this original work is an actual work. This can make up for the situation where the result data of the recording system is missing records, such as when the patient does not cooperate well and repeats the examination, but the recording system only records the result of one examination, the patient cuts in line for the examination and the recording system does not record it, and the recording system makes recording errors due to network or equipment failures. For a certain original work with a determination label of secondary determination, it needs to be verified in combination with the result data. For a certain original work with a determination label of secondary determination, it is considered an actual work only when there is an inspection record in the result data corresponding to the time interval of this original work. This can verify the inspection results of the intelligent model, feedback to the intelligent model for learning, and improve its detection accuracy.

[0062] In a preferred embodiment, to improve the detection accuracy of the electrical parameter identification model and the auxiliary identification model, the following is executed:

[0063] Step A: Extract the electrical parameter test data corresponding to the time interval of the original data as the training samples of the electrical parameter identification model corresponding to the device type, and set labels for the training samples according to the verification result of the original data. Specifically, the electrical parameter detection data corresponding to the time interval in the electrical parameter work data is used as the positive training samples of the electrical parameter identification model corresponding to the device type; the electrical parameter detection data corresponding to the time interval of the original work with misidentification and a determination label of secondary determination is used as the negative training samples of the electrical parameter identification model corresponding to the device type. The electrical parameter detection data corresponding to the time interval of the original work data composed of visual work data that satisfies the second preset condition, although the determination label is a primary determination, is used as the negative training samples of the electrical parameter identification model corresponding to the device type.

[0064] Step B: Extract the auxiliary test data corresponding to the time interval of the original data as the training samples of the auxiliary identification model corresponding to the device type, and set labels for the training samples according to the verification result of the original data.

[0065] Specifically, the video data corresponding to the time interval of the visual work data that satisfies the second preset condition with a determination label of primary determination is used as the positive samples of the visual identification model corresponding to the device type, and the audio data corresponding to the above time interval is used as the positive samples of the voiceprint identification model corresponding to the device type.

[0066] Although the determination label is a secondary determination, it is recognized as actual work. If the original work data for this time is obtained from visual work data that meets the third preset condition, then the video data corresponding to the time interval of this original work is used as a positive sample for the visual recognition model corresponding to the device type, and the audio data corresponding to the above time interval is used as a negative sample for the voiceprint recognition model corresponding to the device type. Although the determination label is a secondary determination, it is recognized as actual work. If the original work data for this time is obtained from voiceprint work data that meets the fourth preset condition, then the audio data corresponding to the time interval of this original work is used as a positive sample for the voiceprint recognition model corresponding to the device type, and the video data corresponding to the above time interval is used as a negative sample for the visual recognition model corresponding to the device type.

[0067] For the original work where the determination label is a secondary determination and is recognized as a misrecognition, the video data corresponding to the time interval of this original work is used as a negative sample for the visual recognition model corresponding to the device type, and the audio data corresponding to the time interval of this original work is used as a negative sample for the voiceprint recognition model corresponding to the device type.

[0068] In this embodiment, by continuously feeding back the obtained electrical parameter detection data, video data, and audio data as samples for the corresponding intelligent models to learn, the inspection accuracy of the models is improved.

[0069] The present invention also discloses a medical device performance evaluation method. In a preferred embodiment, it includes:

[0070] Step A, obtain the actual number of working times of the medical device to be evaluated in each measurement cycle within the evaluation time period according to the above-mentioned medical device working times measurement method;

[0071] Step B, obtain the total actual number of working times of the medical device to be evaluated within the evaluation time period;

[0072] Step C, obtain the performance index value of the medical device to be evaluated based on the total actual number of working times using a preset performance index calculation method.

[0073] In this embodiment, the evaluation time period can be one year, one month, or one week, and can be custom-set. The measurement cycle can be set according to hours, days, or weeks. The evaluation time period is successively divided into multiple measurement cycles, and the total actual number of working times within the evaluation time period is obtained by accumulating the actual number of working times in all measurement cycles. The performance index is preferably but not limited to the payback period and / or the return on investment (%).

[0074] In this embodiment, the shorter the payback period is, the better the operating efficiency of the medical device is. Among them, the preset performance index calculation method for the payback period is: Payback period = Original investment amount of the device / Annual income of the device. The original investment of the device generally refers to the purchase amount of the medical device. The annual income of the device means that the evaluation period is one year, and the total actual working times in one year are obtained according to steps A and B. The examination fees of each actual patient work are obtained from the data in the recording system. For the newly added actual working times exceeding the result data, an examination fee can be preset according to different device types, so that the annual income of the device is obtained by combining the total actual working times and the examination fees of each actual work.

[0075] In this embodiment, the larger the return on investment is, the better the device benefit is. The preset performance index calculation method for the return on investment (%) is:

[0076] Return on investment (%) = (Annual income of the device / Total investment amount of the device) * 100%

[0077] Among them, the total investment amount of the device refers to the actual expenditures of various device purchases that make up the basic investment completion amount.

[0078] The present invention also discloses a medical device performance management system. In a preferred embodiment, as Figure 2 shown, it includes: a plurality of electrical parameter sensors that respectively measure the electrical parameters of different medical devices in operation; a plurality of cameras and / or a plurality of sound pickup devices, with the plurality of cameras respectively installed in the examination rooms where different medical devices are located, and the plurality of sound pickup devices respectively installed in the examination rooms where different medical devices are located; a performance management platform that communicates with the electrical parameter sensors, cameras, and sound pickup devices respectively, and is used for collecting electrical parameter detection data and auxiliary detection data; the performance management platform executes the steps of the above-mentioned medical device working times measurement method, or the performance management platform executes the steps of the above-mentioned medical device performance evaluation method.

[0079] In this embodiment, the electrical parameter sensors are installed on the main power of the power distribution cabinet of the medical device or the power distribution cabinet in the power distribution room of the power plant, and specifically can be three-phase intelligent power meters, as well as sensing devices such as current transformers and temperature sensors for measuring three-phase voltage / current / active power / power factor / temperature. The cameras are installed in the detection rooms where the medical devices are located, and are used to capture images of patients and the device operation status indicators. The sound pickup device is preferably a microphone, and specifically, it is installed in the detection rooms where the medical devices are located.

[0080] In this embodiment, further preferably, to improve the system operation speed and reduce the calculation amount of the performance management platform, an edge processor is equipped for the sound pickup device, and a voiceprint recognition model corresponding to the device type of the medical device is deployed on the edge processor, and the voiceprint work data acquisition is completed in the edge processor.

[0081] The hardware of the performance management platform can be a server, or a cluster composed of a server and multiple client computers. A database and a model library are set up inside the performance management platform. Information such as the IDs of all medical devices in the hospital, the examination rooms where they work, and the device types is stored in the database. The model library stores the electrical parameter identification models and auxiliary identification models corresponding to different device types.

[0082] Figure 3 A hardware deployment solution of the present system is shown. The performance management platform is connected to the hospital intranet through a firewall. The hospital intranet is linked to various recording systems, and electrical parameter sensors, as well as cameras and / or sound pickup devices, are deployed inside each examination room. The electrical parameter detection data, video data, and audio data (or voiceprint working data) are transmitted through the intranet to the performance management platform for processing.

[0083] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0084] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and purposes of the present invention. The scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for measuring the number of times a medical device works, characterized in that: include: Determine the device type of the medical device, obtain electrical parameter detection data and auxiliary detection data of the medical device during the metering cycle, and the result data provided by the recording system, wherein the auxiliary detection data includes video data and / or audio data of the examination room where the medical device is located; Input the electrical parameter detection data into a pre-trained electrical parameter identification model corresponding to the device type to obtain electrical parameter working data; the electrical parameter working data includes the number of electrical parameter identification work times and the time interval of each electrical parameter identification work; Input the auxiliary detection data into a pre-trained auxiliary recognition model corresponding to the device type to obtain auxiliary work data; the auxiliary work data includes the number of auxiliary recognition work times and the time interval of each auxiliary recognition work; Comprehensively process the electrical parameter working data and the auxiliary working data to obtain original data, wherein the original data includes the number of original working times and the time interval of each original working; The original data and result data are used to verify each other and obtain the actual working times of the medical equipment within the measurement cycle.

2. The method for measuring the number of times a medical device works as claimed in claim 1, characterized in that: When the auxiliary detection data consists of video data of the examination room where the medical device is located, the auxiliary working data consists of visual working data; The inputting the auxiliary detection data into a pre-trained auxiliary recognition model corresponding to the device type to obtain auxiliary working data includes: Input the video data into a pre-trained visual recognition model corresponding to the device type to obtain visual work data; the visual work data includes the number of visual recognition work times and the time interval of each visual recognition work; When the auxiliary detection data consists of audio data of the examination room where the medical device is located, the auxiliary working data consists of voiceprint working data; The inputting the auxiliary detection data into a pre-trained auxiliary recognition model corresponding to the device type to obtain auxiliary working data includes: The audio data is input into a pre-trained voiceprint recognition model corresponding to the device type to obtain voiceprint working data; the voiceprint working data includes the number of voiceprint recognition work times and the time interval of each voiceprint recognition work.

3. The method for measuring the number of times a medical device works as claimed in claim 1, characterized in that: The auxiliary detection data includes video data and audio data of the examination room where the medical equipment is located, and the auxiliary recognition model includes a visual recognition model and a voiceprint recognition model; The inputting the auxiliary detection data into a pre-trained auxiliary recognition model corresponding to the device type to obtain auxiliary working data includes: Input the video data into a pre-trained visual recognition model corresponding to the device type to obtain visual work data; the visual work data includes the number of visual recognition work times and the time interval of each visual recognition work; Input the audio data into a pre-trained voiceprint recognition model corresponding to the device type to obtain voiceprint working data; the voiceprint working data includes the number of voiceprint recognition work times and the time interval of each voiceprint recognition work; The auxiliary work data includes visual work data and voiceprint work data.

4. The method for measuring the number of times a medical device works as claimed in claim 2 or 3, characterized in that: The visual recognition model corresponding to the device type includes a human posture recognition model and an indicator light state recognition model; Convert the video data into an image sequence, input the image sequence into a human posture recognition model to obtain posture recognition data, and input the image sequence into an indicator light state recognition model to obtain state recognition data; The posture recognition data and state recognition data are integrated to obtain visual working data.

5. The method for measuring the number of times a medical device works as claimed in claim 2, characterized in that: The comprehensive processing of electrical parameter working data and auxiliary working data to obtain original data includes: Determine a first portion of original data, the first portion of original data consisting of electrical parameter working data, and set a determination tag of the first portion of original data to be a single determination; Determine the second part of the original data, where the second part of the original data is composed of part of the data in the auxiliary work data that meets the first preset condition, and set the determination label of the second part of the original data to be a secondary determination; wherein the first preset condition is that the time interval of the i-th auxiliary identification work in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter work data, and i is the index of the number of auxiliary identification work in the second part of the original data, and i is a positive integer; The first portion of original data and the second portion of original data are combined to obtain original data.

6. The method for measuring the number of times a medical device works as claimed in claim 3, characterized in that: The comprehensive processing of electrical parameter working data and auxiliary working data to obtain original data includes: Determine a first part of original data, the first part of original data is composed of electrical parameter working data and part of data in visual working data that meets a second preset condition, and set a determination label of the first part of original data to be a single determination; wherein the second preset condition is: the time interval of the jth visual recognition work in the first part of original data has no intersection with all time intervals corresponding to the electrical parameter working data, and the time interval of the jth visual recognition work has an intersection with the time interval of the voiceprint recognition work, j is an index of the number of visual recognition work times in the first part of original data, and j is a positive integer; Determine the second part of the original data, the second part of the original data is composed of part of the data in the visual working data that meets the third preset condition and part of the data in the voiceprint working data that meets the fourth preset condition, and set the judgment label of the second part of the original data to secondary judgment; wherein, the third preset condition is: the time interval of the kth visual recognition work in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter working data, and the time interval of the kth visual recognition work has no intersection with the time interval of the voiceprint recognition work, j is the index of the number of visual recognition work in the second part of the original data, and j is a positive integer; the fourth preset condition is: the time interval of the pth voiceprint recognition work in the second part of the original data has no intersection with all the time intervals corresponding to the electrical parameter working data, and the time interval of the pth voiceprint recognition work has no intersection with the time interval of the visual recognition work, p is the index of the number of voiceprint recognition work in the second part of the original data, and p is a positive integer. The first portion of original data and the second portion of original data are combined to obtain original data.

7. The method for measuring the number of times a medical device works as claimed in claim 5 or 6, characterized in that: The method of using the original data and the result data to verify each other and obtain the actual working times of the medical device within the measurement cycle includes: The result data includes the number of result operations and the time interval of each result operation; If the judgment label of the nth original work in the original data is a judgment, the nth original work is regarded as an actual work; If the judgment label of the nth original work in the original data is a secondary judgment, and the time interval of the nth original work and the time interval of the mth result work have an intersection, then the nth original work is regarded as an actual work; If the judgment label of the nth original work in the original data is a secondary judgment, and the time interval of the nth original work does not intersect with the time interval of the mth result work, then the nth original work is regarded as a misidentification; Among them, n is the original work number index, m is the result work number index, and both n and m are positive integers.

8. The method for measuring the number of times a medical device works as claimed in claim 7, characterized in that: Extracting electrical parameter test data corresponding to the time interval of the original data as training samples of the electrical parameter recognition model corresponding to the device type, and setting labels for the training samples according to the verification results of the original data; Auxiliary test data corresponding to the time interval of the original data is extracted as training samples of the auxiliary recognition model corresponding to the device type, and labels are set for the training samples according to the verification results of the original data.

9. A medical equipment performance evaluation method, characterized in that: include: Obtain the actual working times of the medical device to be evaluated in each measuring cycle during the evaluation time period according to the medical device working times measuring method according to any one of claims 1 to 8; Obtain the total number of actual operations of the medical device to be evaluated during the evaluation period; Based on the total number of actual work times, the performance indicator value of the medical equipment to be evaluated is obtained using a preset performance indicator calculation method.

10. A medical equipment performance management system, characterized in that: include: Multiple electrical parameter sensors to measure the electrical parameters of different medical devices; Multiple cameras and / or multiple sound pickup devices, the multiple cameras are respectively installed in the examination rooms where different medical devices are located, and the multiple sound pickup devices are respectively installed in the examination rooms where different medical devices are located; The performance management platform communicates with the electrical parameter sensor, the camera and the sound pickup device respectively to collect electrical parameter detection data and auxiliary detection data; The performance management platform executes the steps of the medical equipment working times measurement method described in any one of claims 1-8, or the performance management platform executes the steps of the medical equipment performance evaluation method described in claim 9.

Images