WO2023194499A1 - Method and device for measuring a process in an object - Google Patents

Abstract

The invention relates to a method and to a device for measuring at least one process in an object. An alternating field is irradiated into the object and received once it has passed through the object. At least one feature of the process and/or a temporal representation of the process is then generated time-dependently using a statistical model.

Description

Method and device for measuring a process in an object

The invention relates to a method and a device for measuring at least one process in an object. An alternating field is irradiated into the object and received after it has passed through the object. A statistical model is then used to generate at least one feature of the process and/or a time representation of the process depending on time.

When treating mechanically ventilated patients, there is a technical I see the question in recognizing the patient's independent breathing (spontaneous breathing). During artificial ventilation, air is forced into the lungs of patients at sometimes high pressures. If ventilation is carried out contrary to spontaneous breathing, this can lead to particularly irreversible damage to the lungs and the upper respiratory tract. In particular, the following damages are known:

Barotrauma: damage to the alveoli due to high ventilation pressures,

Volutrauma: overexpansion of the alveoli due to excessive ventilation volumes,

Atelectasis: Collapse of pulmonary alveoli due to ventilation pressures that are too low or insufficient ventilation volume

But even properly synchronized ventilation over a longer period of time can damage the tissue of patients, which is why it is desirable to reduce the duration of ventilation to a minimum. For this purpose, a clear assessment of the patient's vital status must be carried out, which makes the recording of further vital parameters essential. This leads to further technical questions as to how, in addition to respiratory parameters (e.g. expiratory phase, inspiratory phase), cardiologically relevant parameters (e.g. ECG data, QRS complexes, heart rates, heart rate variability, arrhythmias) can also be obtained. The challenge here is to make this possible in a contactless procedure in order to minimize interference with patients and thereby increase acceptance of the measurement procedure.

The collection of vital data such as heart and respiratory rate, respiratory volume, breathing depth, duration and time of inspiration or expiration is useful in a variety of applications. For example, in the control of ventilators, in the detection of sleep apnea or respiratory diseases, in monitoring the stress status of emergency services or in biometrics in vehicles. The sensor system should be as fast, unobstructive and robust as possible against disruptive influences. A technology that is suitable here is the chest monitor (WO 2019/243444 Al). The obtained multidimensional and time-dependent data provide information about underlying dynamic changes in the test system. However, interpretation is prone to artifacts.

When deriving cardiorespiratory data or generally when deriving vital parameters, the majority of procedures use contacting modalities or modalities that are close to the patient's body. This can be very stressful for patients, especially when taking measurements over longer periods of time or in everyday situations in which measuring vital parameters is not the focus (e.g. when driving or sleeping).

The best established method for measuring cardiac activity is certainly the electrocardiogram (ECG). The electrical activity of the heart is recorded using electrodes stuck to the body.

A large number of established procedures combine an ECG with impedance cardiography (ICG) to additionally measure respiratory activity. These methods estimate the cardiac component and subtract it from the impedance signal to obtain the respiratory component from the resulting signal. Since the recording of the respiratory parameters is indirectly derived from the estimation of the cardiac component, this necessarily depends on the quality of the cardiac model. Furthermore, the breathing frequency is mainly determined here. A direct statement about tidal volume and spontaneous breathing is not possible.

Another method for assessing respiratory activity is to measure chest and abdominal distension using respiratory inductive plethymography (RIP). Here, two elastic bands are placed around the patient's thorax and abdomen. Coils are incorporated into the bands, which change their inductance when the bands are stretched. This method can have an uncomfortable and restrictive effect on the patient due to the elastic bands. A comparable method is also conceivable using strain gauges, although there are no further advantages to the method. There are also a number of methods that measure vital parameters without contact. Most methods are based on radar-based measurement. However, these radar-based methods usually only provide information about the breathing rate or heart rate via reflections on the chest.

The object of the present invention is therefore to provide a method for measuring a process in an object that is non-invasive, non-obstructive and scalable.

The object is achieved by the method for measuring at least one process in an object according to claim 1 and the device for measuring at least one process in an object according to claim 21. The respective dependent claims specify advantageous developments of the method according to the invention and the device according to the invention.

According to the invention, a method for measuring at least one process in an object is specified. In principle, the method can be applied to all types of objects, especially those into which an alternating electromagnetic field can penetrate. Preferably the object can be a body of a living being, e.g. B. a body of a human or an animal.

If we are talking about measuring the process in the object, this means deriving the interesting features of the process from the demodulation data. Synonymous with “measurement” “predictions” or “predictions” or “estimates” can be used here.

According to the invention, an alternating electromagnetic field is irradiated into the object. The alternating field is received after it passes through the object. The fact that the alternating field has passed through the object preferably means that it entered the object after being irradiated and exited the object before being received. Preferably, the alternating field has entered the object at least 1 cm, particularly preferably at least 5 cm, particularly preferably at least 10 cm, particularly preferably at least 20 cm. The alternating field is modulated by the process in the object that is to be measured. This can mean in particular that the process in the object leads to a phase shift, frequency shift and/or amplitude change of the received alternating field compared to the irradiated alternating field. Modulation here can advantageously be understood as a process in which useful data is applied to a carrier signal, here the radiated alternating field. Changes in the object that form the process to be measured can be viewed as useful data. In this understanding, these changes are modulated onto the irradiated alternating field as a carrier.

The minimum penetration depth of the alternating field into the object can be dimensioned such that the received alternating field at the given penetration depth still experiences a modulation that exceeds the measurement accuracy of the method. The alternating field should therefore preferably only be viewed as reaching the minimum penetration depth mentioned, i.e. as having penetrated into the object, if the modulation at this penetration depth still leads to a distinguishable difference in the measurement of the at least one process.

The received alternating field is demodulated in a step called a demodulation step to generate multi-dimensional demodulation data. Advantageously, several receiving structures can receive the same or different alternating fields. Multidimensional demodulation data can then be demodulated from the received alternating fields of each reception structure in a separate demodulation step and fed together into a statistical model. For example, asymmetries of the object can be particularly advantageously incorporated into the model (e.g. right and left lung).

Demodulation can then be understood to mean that the useful data is separated from the carrier signal. In this step, the received alternating field can be separated into the irradiated alternating field on the one hand and the changes caused by the process to be measured on the other. In this understanding, the changes in the process can then be viewed as multi-dimensional demodulation data. The fact that the demodulation data is multidimensional can mean that it has at least two independent components, particularly advantageously at least two linearly independent components.

According to the invention, the multidimensional demodulation data and/or data derived from these multidimensional demodulation data are fed to a statistical model. The statistical model then generates, in a step referred to as a generation step, at least one feature of the process and/or a temporal representation of the process from the supplied multidimensional demodulation data and/or from the data derived therefrom in a time-dependent manner.

The fact that generation is time-dependent can mean in particular that it occurs at least at two different points in time. However, time-dependent generation advantageously means regular generation, in particular quasi-continuous generation or, where possible, advantageously also continuous generation. Continuous generation should be viewed as continuous in the sense in which this term is usually understood in the area of digital data processing, i.e. discretized in time, for example with a clock frequency. With analog processing, the term can be understood continuously but also in an analog sense. The data can have a continuous or quasi-continuous course.

According to the invention, a temporal representation of the process can be generated. This representation can therefore, for example, contain at least an electrocardiogram, at least one impedance cardiogram, at least one ballistocardiogram, at least one seismocardiogram, the movement and/or the expansion of the chest and/or the abdomen, at least a blood pressure curve over time, a respiratory volume curve, at least one respiratory pressure curve, at least one respiratory flow as a function of time, at least one course of oxygen saturation over time, temporal data obtained through imaging, and/or a plethysmogram. Temporal processes that are obtained from the image data are, for example, ventilation of the left and/or right lung over time, which is determined from an ultrasound, electrical Roimpedance tomography or computer tomography.

According to the invention, one or more features of a process can also be determined, such as exceeding a threshold value and the like.

In particular, the following features can advantageously be determined:

For example, one or more of the following can be identified as respiratory features:

Tidal volume: The tidal volume refers to the tidal volume that is inhaled into the lungs and exhaled again during normal inhalation and exhalation. Measuring tidal volume can provide information about the efficiency of breathing and is relevant in the treatment of respiratory diseases.

Respiratory cycle: The respiratory cycle measures the total time required for one breath, from the start of inhalation to the end of exhalation. Measurement of the respiratory cycle can be used to monitor respiratory rate and efficiency and is relevant in the treatment of respiratory diseases.

Inspiration time: Inspiration time describes the length of time during which a breath is inhaled. Measuring the inspiratory time can provide information about the respiratory rate and efficiency and is relevant in the treatment of respiratory diseases.

Expiratory time: Expiratory time describes the length of time during which a breath is exhaled. Measuring expiratory time can provide information about respiratory rate and efficiency and is relevant in the treatment of respiratory diseases.

Inhalation to Exhalation Ratio: The ratio of inhalation to exhalation respiration refers to the ratio of the time during which a breath is inhaled to the time during which a breath is exhaled. Measuring the ratio of inhalation to exhalation can provide information about respiratory rate and efficiency and is relevant in the treatment of respiratory diseases.

Peak inspiratory flow: The peak inspiratory flow measures the maximum volume flow during inspiration and provides information about the performance of the airways. Peak inspiratory flow measurement can be used to monitor respiratory function and is relevant in the treatment of respiratory diseases.

Peak expiratory flow: The peak expiratory flow measures the maximum volume flow during exhalation and provides information about the performance of the airways. Peak expiratory flow measurement can be used to monitor respiratory function and is relevant in the treatment of respiratory diseases.

One or more of the following can advantageously be determined as cardiological parameters:

Position type determination: ECG parameters that measure the electrical activity of the heart in different directions to assess the heart rhythm and electrical activity of the heart.

ST segment changes: ECG abnormalities that may indicate possible heart disease, particularly damage to the heart muscle.

T-wave changes: Changes in the shape, amplitude, or direction of T waves on the ECG, which may indicate possible heart disease or electrolyte imbalances.

QT interval prolongation: Prolongation of the time interval between the beginning of the QRS complex and the end of the T wave on the ECG, which may indicate possible cardiac arrhythmias. P-wave change: Changes in the shape or amplitude of the P-waves in the ECG, which may indicate possible atrial disease.

QRS complex change: Changes in the shape or duration of the QRS complex on the ECG, which can indicate possible diseases of the heart muscle.

Bundle branch block: Disturbance in electrical conduction in the heart, which can indicate possible heart disease.

Holter, exercise ECG: ECG recordings over a longer period of time or during physical exertion to diagnose heart disease or monitor patients at risk.

Heart Rate: Number of heart beats per minute, an important indicator of heart function.

Heart rate variability: Fluctuations in heart rate that are an important indicator of possible heart disease.

Arrhythmias: Abnormalities in the heart rhythm that can indicate possible heart disease.

Ejection fraction (EF): Ejection fraction is a percentage that indicates how much blood is ejected from the left ventricle with each heartbeat. Decreased EF may indicate heart failure or other cardiac problems.

The temporal representation of the process can also be referred to as a simulation of the process. It is not absolutely necessary that the timing is output as a diagram. It may also be sufficient if it is identified and/or stored in such a way that it is suitable for the intended use.

In an advantageous embodiment of the invention, the electromagnetic see alternating field that is irradiated into the object contains at least two components of different frequencies. In particular, sine oscillations with the corresponding frequencies could be viewed as components. The advantage here is that different frequencies have different sensitivities to disturbance variables. The alternating fields of different frequencies can in turn be demodulated and the data derived from them can be fed into the statistical model. The demodulation data can advantageously be processed separately for the different frequencies. In general, the use of several different frequencies enables the suppression of disturbances by increasing the dimensionality and through different proportionality to the disturbance, and advantageously also improves the signal quality, robustness and reproducibility.

In a particularly advantageous embodiment, the received alternating field can be demodulated in the demodulation step in an in-phase & quadrature method for generating I data and Q data as the multidimensional demodulation data. The I and Q data can all be demodulation data or further demodulation data can be generated in the demodulation step. The I data and the Q data, as well as any other demodulation data, can be fed to the statistical model. Alternatively or additionally, data derived from the I data and the Q data and optionally, if necessary, further demodulation data can also be fed to the statistical model.

The statistical model can advantageously be viewed as a unit that has at least one input and at least one output. Data that is fed to the statistical model can therefore be entered into at least one input of this unit. The statistical model can use this to generate at least one feature of the process or the representation of the process and output this via its at least one output.

If data derived from the multidimensional demodulation data and/or the I data and/or the Q data are used, these can be one or more selected from a group that contains the following data: A derivative from Q to I, a Derivation from I to Q, a derivative of I with time, a derivative of Q with respect to time, an amplitude of the received alternating field, a phase of the received alternating field, an amplitude from the derivative of Q with respect to time and the derivative of I with respect to time, a phase from the derivative of Q to time and the derivative of I with respect to time, derivatives of the amplitude with respect to time, derivatives of the phases with respect to present, a phasor of the derivative of Q with respect to I, second or higher derivatives of the derived data with respect to time and/or to another derived size and/or Q data and/or I data, binarizations of the derived data, binarizations of Q data, binarizations of the I data, moving averages of the generated data, moving averages of the Q data, moving averages of the I data, Q data filtered by frequency filters, I data filtered by frequency filters, generated data filtered by frequency filters, derived data from past points in time, Q data from past points in time and/or I data from past points in time.

One or more of the following data can be used particularly advantageously and have, among others, the following advantages:

A derivative from Q to I can help determine the phase shift between I and Q signals.

A time derivative of I can help identify changes in the amplitude of the signal over time. It correlates directly with flow.

A time derivative of Q can help identify changes in the phase of the signal over time. The value correlates directly with the flow.

An amplitude of the received alternating field can be a direct measurement of the strength of the received signal.

A phase of the received alternating field can help to determine the time shift of the received signal compared to a reference signal.

An amplitude from the derivative of Q with respect to time and the derivative of I by time can help determine the rate of change of the signal over time. It correlates directly with flow.

A phase from the derivative of Q with respect to time and the derivative of I with respect to time can help determine the rate of change of the signal's phase over time.

Derivative of amplitude versus time can help identify changes in the amplitude of the signal over time. It correlates directly with flow.

Derivatives of phases with respect to time can help identify changes in the phase of the signal over time. They correlate directly with flow.

A phasor of the derivative from Q to I can provide an alternative representation of the signal as a vector in phase space.

Second or higher derivatives of the derived data in time and/or to another derived quantity and/or Q data and/or I data may help to capture higher orders of changes in the signal.

Binarizations of the derived data can help translate the signal into simpler, binary formats. This leads to a reduction in complexity.

Binarizations of Q data can help translate the signal into simpler, binary formats. This leads to a reduction in complexity.

Binarizations of the I data can help translate the signal into simpler, binary formats. This leads to a reduction in complexity.

Moving averages of the generated data, moving averages of the Q data, and/or moving averages of the I data can help smooth the signal and reduce noise.

Q data filtered by frequency filters, I data filtered by frequency filters and/or generated data filtered through frequency filters can help reduce unwanted frequency components in the signal.

Derived data from past time points, Q data from past time points, and/or I data from past time points can help track and predict changes in the signal over time.

In an advantageous embodiment of the invention, at least one input feature can be extracted from the demodulation data and/or from data derived therefrom, which is fed to the statistical model. The at least one input feature can advantageously be a linear combination and/or a non-linear combination of the multi-dimensional demodulation data and/or the data derived therefrom. Such input features can also themselves be viewed as data derived from the demodulation data.

Advantageously, the statistical model can contain or be one or more selected from the following group: an artificial neural network, an artificial neural feed-forward network, an artificial neural recurrent network, an artificial convolutional network or a combination of the aforementioned artificial neural networks, a decision tree, a decision forest, a naive Bayes model, a nearest neighbor model, a support vector machine, a linear regression model, a non-linear regression model, an adaptive filter, a Kalman filter, a gamma filter , a Markov model, a model from a linear discriminant analysis, a model from a non-linear discriminant analysis, a model from an independent component analysis, a model from a principal component analysis and/or an ensemble -Model from one or more of the aforementioned models. It should be noted that in the above list, those models named with English names are commonly referred to with this English name.

For example, the following statistical models can be used with the following advantages:

Artificial neural network: Can create complex non-linear relationships model the data and thus provide high prediction accuracy for time series-based forecasts.

Artificial feed-forward neural network: A neural network in which signals only flow in one direction, without feedback. Good for creating simple non-linear relationships such as forecasting simple time series.

Artificial neural recurrent network: A neural network in which the output of the network also serves as an input for later points in time to draw on past states and is therefore suitable for time series prediction.

Artificial convolutional network: A model designed to process signals by extracting local features and merging them. This means that feature extraction is implicit and is therefore well suited for complex time series.

Decision Tree: Allows you to identify important variables and interaction patterns in the data, which is particularly useful when the relationship between the variables is non-linear.

Decision Forest: Similar to a decision tree, but using multiple decision trees, ensemble methods can be used to achieve better prediction accuracy.

Naive Bayes Model: Is particularly good for problems with a large number of variables because it makes a simple assumption about the independence of the variables.

Nearest Neighbor Model: Can be able to recognize the patterns and relationships in the data well, especially when the data is high dimensional.

Support Vector Machine: Can also be useful in cases with high-dimensional data, especially if the data is nonlinear. Linear regression model: Particularly suitable for problems with linear dependence between variables and can be trained quickly.

Non-linear regression model: Can also be useful in cases with non-linear dependence between variables to capture complex patterns and relationships

Adaptive filter: A model capable of adapting to changing environmental conditions and used for signal processing and prediction.

Kalman filter: A stochastic model used to estimate states of a dynamical system and suitable for time series prediction.

Gamma filter: Can reduce noise in the data while providing an effective trade-off between prediction accuracy and response time.

Markov model: A model based on the assumption that the future state depends only on the present state and is suitable for time series prediction.

A model from a linear discriminant analysis: Can be useful for improving the prediction of categories in time series-based data.

A model from a non-linear discriminant analysis: This can provide higher predictive accuracy than linear discriminant analysis models and is particularly useful when the relationships between the variables are non-linear.

An independent component analysis model: Can help identify hidden variables and patterns in the data by transforming the variables to make them statistically independent. A model from a principal components analysis: Can help reduce complex relationships between variables in the data and identify key trends or patterns in the data, which can be particularly beneficial when modeling time series. In addition, PCA can also help reduce the amount of noise in the data and thus improve the accuracy of predictions.

The following, among others, are possible for neural networks (abbreviated NN), with the advantages mentioned in each case. Some of these approaches can also be used in combination and/or consecutively. For example, an encoder-decoder network can be built to be Bayesian and promoted.

Bayesian Neural Networks: An NN based on Bayesian probability to quantify and reduce the uncertainty in the predictions by using a distribution of weights rather than a single weight number. It is a good choice for time series modeling when uncertainty needs to be taken into account in predictions. It is therefore very suitable in a medical context.

Promoted Trained Neural Networks: A group of NNs trained decentrally on multiple devices to address privacy and security concerns when using sensitive data. Suitable for processing large amounts of data collected in distributed environments. Good for later use when there are several decentralized chest monitors that record all the data and train an NN. A good step towards generalization.

Encoder-decoder networks: A type of NNs that encode and decode input data to perform a specific task such as prediction or translation. They are suitable for time series modeling because they are able to capture complex patterns in the data and generate predictions based on these patterns.

Attention-Based Networks: An NN that optimizes the weight of the input variables through an attention function to provide important information identify and focus while suppressing irrelevant information. They are suitable for processing long time series with many variables as it can draw selective attention to certain variables or patterns.

Generative Models (Adversarial Networks (GANs) and Variational Autoencoder (VAE)): NNs capable of generating new data following the statistics of the input data. They are suitable for generating simulated time series that correspond to the properties of real time series. Good for data augmentation and therefore a step towards generalization. The data generated can then be added back to another NN.

Reinforcement Learning Networks: An NN that interacts with an environment and through trial and error learns how to perform a task by receiving positive and negative rewards. Suitable for processing time-dependent data as it is able to make decisions based on history and current state.

LSTMs (Special form of RNNs): An RNN with a special architecture capable of capturing and storing long dependencies in the data. Suitable for time series modeling because it is able to effectively store and process past states. Optionally, for example, an LSTM can also occur multiple times in combination in the encoder-decoder architecture. An LSTM can also optionally follow a CNN (e.g. for feature extraction) consecutively.

Deep Belief Networks: A type of NN that has multiple layers of connected neurons used for feature extraction and output data prediction. Suitable for time series modeling as it is able to capture and exploit complex dependencies between the data.

Echo State Networks (RNN Type): A special type of RNNs that have a random, fixed connection between neurons in the internal layer. They are suitable for time series modeling because it is able to effectively store and process past states, which is useful for forecasting future conditions is crucial.

Gated Recurrent Unit (RNN Type): A special type of RNNs that allows controlling the forgetting of information from previous points in time by using a gating mechanism in the architecture. They are suitable for time series modeling because it is able to retain important information from the past while shedding irrelevant information.

Transformer Networks: An architecture based on the use of multi-head attention mechanisms to model relationships between different parts of the input variables. Are suitable for processing long time series because it is able to focus selective attention on important parts of the input variables and suppress irrelevant information.

One or more of the following models can also be used as an option:

Autoregressive Integrated Moving Average (ARIMA): A statistical model based on the combination of autoregressive and moving average processes and capable of modeling non-linear relationships in a time series. They are suitable, for example, for predicting short-term trends in time series, such as arrhythmias.

Seasonal Autoregressive Integrated Moving Average (SARIMA): An extension of ARIMA that takes into account the seasonal component in a time series and is capable of capturing seasonal patterns in the data series. They are suitable for predicting seasonal trends in time series, such as chronic diseases.

Exponential Smoothing (ETS): A statistical model based on the estimation of trends and seasonal effects, capable of modeling both short-term and long-term trends in a time series. This is suitable for predicting short- and long-term trends in time series. Application, for example, as described above. Vector autoregression (VAR): A statistical model that can model multiple variables in a time series by taking into account the dependencies between the variables. Suitable for modeling time series with multiple variables and predicting relationships between variables. In this case the dependency is Q/l ■

A particularly advantageous embodiment of the invention provides that the at least one feature of the process and/or the temporal representation of the process is generated in real time. “Real time” is understood in particular to mean that the time required for processing in the generation step and advantageously also in the demodulation step is shorter than a predetermined threshold. The threshold is chosen to be suitable for the given application. In particular, it can be chosen so that it is smaller than the smallest time scale of the application in question. Advantageously, rapid generation or processing in the generation step can be achieved by the following measures, which can be advantageous for real-time evaluation, but also for other purposes:

In an advantageous embodiment, a statistical model, for example a deep neural network, with an inference latency of less or equal to 200ms, preferably less or equal to 150ms, preferably less or equal to 100ms, preferably less or equal to 50ms, preferably less or equal to 30ms, preferably less or equal to 20ms, preferably less or equal to 10ms, preferably less or equal to 5ms, preferably less or equal to 1ms can be used. Of course, 0ms can be considered the lower limit, since a minimum inference latency is usually not required.

A deep neural network with a depth of less than or equal to 150, preferably less or equal to 50 layers, preferably less or equal to 20 layers, preferably less or equal to 10 layers, preferably less or equal to 3 layers can advantageously be used. At the same time, however, the network should have sufficient complexity to be able to effect the relevant generation with sufficient precision. The model has preferential wise two or more layers. In general, for fast or real-time processing, the latency of the inference is the more important parameter for characterizing the network.

For rapid generation, in particular for real-time generation in the generation step, one or more of the following optimization techniques can advantageously be used:

Quantization: The accuracy of weights and activations in the network can be reduced, which can significantly reduce memory requirements and inference time while maintaining accuracy.

Pruning can be used in an advantageous embodiment. In doing so, unimportant connections or neurons can be removed from the network, thereby reducing the size of the network, which can significantly reduce memory requirements and inference time while maintaining accuracy.

In an advantageous embodiment, knowledge distillation (transfer learning) can be used. A smaller network can be trained to imitate the behavior of a larger, more complex network. This can significantly reduce memory requirements and inference time while maintaining accuracy.

In an advantageous embodiment, compression algorithms can be used. For example, Huffman coding can be used advantageously. This is a lossless compression technique that encodes commonly used characters, in this case weights of neurons, into a data set with shorter bit strings to reduce the overall size of the network, which can significantly reduce memory requirements and inference time while maintaining accuracy can stay.

In a further advantageous embodiment, Lempel-Ziv compression can be used. This is a lossless compression technique that detects redundant data in a data set, in this case weights of neurons, and replaced with references to previously occurring data to reduce the overall size of the network, which can significantly reduce memory requirements and inference time while maintaining accuracy.

Learning-based compression (see e.g. https : / /ar-xiv . org/pdf / 1802 . 034 94 . pdf) can also be used advantageously. This is a data compression method that uses reinforcement learning to generate a compression policy that determines which parts of the model can be reduced. This can significantly reduce memory requirements and inference time while maintaining accuracy.

In an advantageous embodiment, NAS (Neural Architecture Search) can be used for generation. This involves using automated techniques to search for optimal network architectures that strike a balance between size, i.e. inference latency, and accuracy. In comparison to the other compression techniques mentioned, which attempt to optimize a pre-trained statistical model, NAS directly searches for an optimal architecture that meets the user's needs, in this case the inference latency and accuracy.

The generation step can also be accelerated by parallelizing the model. It is possible, among other things, to run the calculations on multiple processors or GPUs in order to increase the prediction speed.

It is also possible to use several models in parallel in the ensemble. The inference can run in parallel with several smaller models with lower latency and subsequent majority decision or averaging of the inferences.

In addition, the sampling rate and/or the dimension of the values of the device according to the invention or the method according to the invention could be adaptively changed in order to adaptively change the complexity of the input data. this and thereby adapt the inference latency of the statistical model depending on the situation. For example, a device according to the invention (also called a “torax monitor”) can generate 2 signals, two devices can generate 4 signals or 3 devices can generate 6 signals.

In an advantageous embodiment of the invention, the alternating electromagnetic field can be irradiated into the object via at least one transmission structure by generating an alternating voltage by a signal generator, which is applied to the at least one transmission structure. In addition, the alternating field can be received by at least one receiving structure after it has passed through the object.

Advantageously, the alternating voltage and/or the radiated alternating field has a frequency of greater than or equal to 10 MHz, preferably greater than or equal to 30 MHz, preferably greater than or equal to 100 MHz and/or less than or equal to 1,000 MHz, preferably less than or equal to 500 MHz, preferably smaller or equal to 300 MHz.

It is particularly advantageous if the at least one transmitting structure and the at least one receiving structure are arranged on the same side of the object. For example, the transmitting structure and the receiving structure can be arranged in a backrest, at least one siderest and/or at least one armrest of a seat or in a lying surface. In this way, a measurement is possible, for example, when a person is sitting on a seat or lying on a lounger.

Advantageously, the at least one transmitting structure and the at least one receiving structure can be arranged in textiles that are worn on the body and/or close to the body. This could be, for example, jackets or blankets. An arrangement in sports equipment, in medical devices such as a wheelchair, an arrangement in household appliances such as a stove to monitor the water level, an arrangement in industrial plants, for example to monitor the condition of machines and systems, is also possible.

In an advantageous embodiment of the invention, the object can be a human body. In this case, the at least one transmission structure can be used For example, be arranged in the immediate vicinity and/or on a thorax and/or an abdomen of the body. The at least one receiving structure can also be arranged in the immediate vicinity and/or on the thorax and/or the abdomen. The arrangement on the corresponding body part should preferably be considered a constellation in which the transmitting structure or the receiving structure is arranged in sufficient proximity to the corresponding body part so that the measurement is possible with the desired accuracy. Of course, devices can also be provided which hold the transmitting structure or the receiving structure on the corresponding part of the body. Alternatively, the transmitting structure or the receiving structure can also be arranged in a stationary manner and the person can move their body into a corresponding position near the transmitting structure or the receiving structure. This embodiment should also optionally be viewed as an arrangement of the transmitting structure or the receiving structure on the corresponding part of the body.

In an advantageous embodiment, the object to be measured is located in the near field of the at least one transmitting and/or receiving structure. The near field can be understood here as the area around the transmitting and/or receiving structure in which the alternating electromagnetic field has not yet completely separated from the transmitting and/or receiving structure and is not yet radiated into the free space as a plane wave. The alternating electromagnetic field is still coupled to the transmitting and/or receiving structure in the near field. The IEEE definition of the near field can also optionally be used as a basis. Here the outer limit is defined as the distance X/(2n) from the antenna surface, where X is the wavelength in free space (see IEEE Standard for Definitions of Terms for Antennas, IEEE Std 145-2013). Consequently, the distance between the transmitting and/or receiving structure and the object to be measured is advantageously less than or equal to X/(2n).

Particularly preferably, the at least one transmitting structure and/or the at least one receiving structure is arranged laterally, dorsally or ventrally to the abdomen and/or laterally, dorsally or ventrally to the thorax of the body.

In an advantageous development of the invention, the alternating electromagnetic field can be irradiated via at least two transmit-receive pairs are received. Each of the send-receive pairs can contain a send structure and a receive structure. The alternating field, which is emitted by a transmitter of one of the transmit-receive pairs and received by the receiver of this pair, can be viewed as a single alternating field and which can be viewed by the transmitter and receiver of other transmit-receive pairs as separate alternating fields. Each transmit-receive pair can therefore carry out the process of irradiation and reception with its own alternating field. On the other hand, since the entirety of all alternating fields generated in this way can be viewed as a common alternating field with a specific spatial distribution, this can also be formulated in such a way that the alternating electromagnetic field is irradiated into the object via the at least two transmit-receive pairs and is received after it passed through the object. This configuration allows suppression of disturbance variables or an improvement in the estimation of the determination of the characteristic of the process below the temporal representation of the process by increasing the dimensionality of the demodulation data or the model. The signal quality, robustness and reproducibility can also be improved in this way. It is also possible to use an unequal number of receiving and sending structures. The use of more receiving structures than transmitting structures is particularly advantageous. For example, the at least one transmitting structure can radiate the alternating field into the object and the at least two receiving structures can receive the alternating field after it has passed through the object. Advantageously, the at least two receiving structures can be arranged at a different position around the object or on the object.

In an advantageous embodiment of the invention, the method can be carried out in such a way that it continuously monitors the process and continuously generates at least one feature of the process and/or a temporal representation of the process. Any regular repetition of the process should be considered continuous here, i.e. execution of the process at discrete times, at quasi-continuous times or continuously in the sense of digital data processing.

The characteristic determined in this way or the representation of the process determined in this way can can be compared with previous values of the feature or previous representations of the process. In particular, an event can be triggered if the last determined feature deviates from an earlier value of the feature by more than a predetermined threshold value and/or if a characteristic of the representation of the process deviates significantly from a previous embodiment of this characteristic of the process. For example, an alarm can be issued as an event. Possible applications for such a procedure include, for example, those for monitoring vital functions in which a change can mean a critical condition.

For this purpose, the method is preferably carried out regularly and a value of the at least one process is regularly compared with at least one previous value of the feature. An alert event can then be triggered if the value of the feature deviates from the previous value of the feature by more than a predetermined threshold.

In a particularly advantageous embodiment, parameters of the statistical model and/or the statistical model can be adjusted based on a difference between the value of the generated feature and the corresponding previous value of the feature if the value of the generated feature is less than the threshold value from the previous value differs. In this way, the model can adapt to the current state of the person, for example, so that, for example, a cue event is only triggered if the deviation between the last value of the feature and the corresponding previous values is greater than the threshold value.

In an advantageous embodiment of the invention, the at least one feature of the process can be a trigger event and/or exceeding or falling below a predetermined threshold value. If the process is, for example, a respiratory volume curve or respiratory flow curve, and the method is intended to support the breathing process, for example, the trigger event can be viewed as, for example, exceeding a certain respiratory volume or respiratory flow when inhaling. This trigger event can then, for example, trigger an inhalation support process. Advantageously, the at least one feature of the process and/or the temporal representation of the process can describe the process continuously, quasi-continuously or at a plurality of, preferably equidistant, times. In particular, the at least one feature and/or the time representation can describe the process at regularly recurring points in time.

In many embodiments of the invention, it can be advantageous if the statistical model is trained before the actual measurement of the process begins. The training would therefore preferably take place before the alternating electromagnetic field is irradiated. The statistical model can be trained in a training process in which an alternating electromagnetic training field is irradiated into at least one training object. Advantageously, the training alternating field can share one or more properties with the alternating electromagnetic field that is later irradiated for measurement or, particularly preferably, can be identical. The training object can advantageously be an object that shares one or more properties of interest with the object in which the process is to be measured. For example, the training object can be a living being of the same species as the object to be measured. For an individualized procedure, the training object can also be the object itself in which the process is to be measured. If the object is a living being, for example a human or an animal, an object can be used as the training object that is representative, for example, of a cohort into which the living being to be measured later falls.

As described above, a subsidized approach is advantageous here. There are, among others, the following options for creating a trained model from several individual models:

Model fusion: A method of merging multiple models into a single model to achieve better predictions. This makes sense because different models have different strengths and weaknesses, and combining these models can result in a more powerful model. Transfer learning: A method in which a pre-trained model is used as a starting point for training a new model to reduce the amount of training data required. This makes sense because it can reduce the effort required to train models and improve the accuracy of models.

Ensemble methods: A method in which multiple models are grouped together and decisions are made by majority vote to achieve better predictions. This makes sense because combining multiple models can result in better prediction accuracy than any single model.

Knowledge Distillation: A method in which a pre-trained model is used as a teacher model to train a smaller model as a student model. This can be useful because it can help a smaller model perform similarly to the larger model but require fewer computing resources.

For the training process, the alternating electromagnetic training field is irradiated into at least one training object and the alternating training field is then received after it has passed through the at least one training object. The received training alternating field can then be demodulated to produce multi-dimensional demodulation data. The multidimensional demodulation data and/or training data derived from them can then be fed to the statistical model. The statistical model can then be used to generate at least one training feature of the process in a time-dependent manner from the multidimensional demodulation data supplied and/or the training data derived therefrom. The at least one training feature can then be compared, depending on time, with a measured feature of the process. The measured feature can be generated, for example, by a direct measurement with a suitable measuring device or sensor. The statistical model and/or parameters of the statistical model can then be adjusted based on a result of this comparison. The adaptation can preferably be directed in such a way that the difference between the training feature and the measured characteristic is reduced.

In an optional embodiment, it may also be possible to use the following instead of data generated by the chest monitor itself:

Generative models (Adversarial Networks (GANs) and Variational Autoencoder (VAE)) are neural networks capable of generating new data that follow the statistics of the input data. These can be suitable for generating simulated time series that correspond to the properties of real time series. This is good for data augmentation and therefore a step towards generalization. The data generated can then be added back to another NN.

Data augmentation is a technique to increase the amount and variety of training data by creating artificial variations:

As already mentioned above, the following networks can also be used for data augmentation:

Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU): LSTM and GRU are special types of recurrent neural networks (RNNs) that are particularly suitable for modeling sequential data such as time series. You can recognize patterns in the data and generate new data accordingly.

Further variants from stochastics that can be used optionally are:

Markov Chain Monte Carlo (MCMC) Method: MCMC is a stochastic method for generating data from an unknown distribution by using a Markov chain that converges to the target distribution. MCMC methods such as Metropolis-Hastings and Gibbs Sampling can be used to produce time series that follow a specific distribution. Manual or automatic data augmentation:

Phase Shift (Rotation): o The time series is cyclically shifted along the time axis to create new variations in the data. This can help train the model for different phase shifts within the time series data.

Smooth or Dynamic Shift: o Here data can be multiplied by a factor to change the amplitude of the signals. This can help train the model for different amplitude ratios.

Add noise: o Random noise (e.g. Gaussian noise) can be added to the time series data to increase the robustness of the model to noise in the input data.

Time Warping: o This allows the speed at which the time series runs to be changed by stretching or compressing data points along the time axis.

Mirroring: o Mirroring the time series along the time or amplitude axis to produce variations in the order or direction of the time series data.

Frequency domain manipulation: o Transformations in the frequency domain, e.g. B. by using the Fourier transform. The frequency spectrum can also be manipulated by boosting, weakening or shifting certain frequency bands. This can help train the model for different frequency characteristics.

Random Permutations: o There can be random permutations of time series sections be executed. This can help train the model for different orders of events within the time series.

Regarding the alternating training field, the training object, the received alternating training field, the demodulation data and the training data derived therefrom, what has been said about the alternating electromagnetic field used for the measurement, object, demodulation data, derived data and features applies analogously. In fact, it is advantageous if the training process is carried out identically to the measurement process, since this achieves the best possible match between the result of the measurement and the values of the feature or process that would be generated by direct measurement with a suitable measuring device.

Advantageously, the training process can be carried out for a plurality of training objects that are representative of the object. For example, the training process can be carried out for a large number of living beings in a cohort, which also includes the living being in which the process is to be measured. A large number of training objects makes the model more robust and a generalized statistical model can be created. However, it is also advantageously possible to use the object on which the measurement is to be carried out itself as the training object. The training process is then carried out on this object.

The method according to the invention is particularly suitable for controlling a ventilation process. For this purpose, a breathing process can be measured as the process to be measured. The onset of a spontaneous breathing process, for example, can then be determined as the at least one feature of the process. The ventilation process can then be controlled so that no ventilation takes place against spontaneous breathing.

According to the invention, a device for measuring at least one process in an object is also specified. What was said about the process applies analogously to the object. The device has at least one transmission structure with which an alternating electromagnetic field can be irradiated into the object by generating an alternating voltage by a signal generator with which the at least one transmission structure is applied. The alternating voltage can be sinusoidal or contain a superposition of several frequencies. The device also has at least one receiving structure with which the alternating field can be received after it has passed through the object. What was said above regarding penetration depth applies analogously.

The device according to the invention also has a demodulation unit with which the received alternating field can be demodulated to generate multi-dimensional demodulation data. In addition, the device has a modeling unit to which the demodulation data and/or data derived therefrom can be supplied, and with which at least one feature of a process and/or process is time-dependently determined by a statistical model from the multidimensional demodulation data supplied and/or the data derived therefrom. or a temporal representation or replica of the process can be generated.

Preferably, the device can be used to carry out a method as described above.

The device can advantageously have an I/Q demodulator, into which, on the one hand, the alternating voltage is fed, with which the transmission structure is applied, and into which, on the other hand, a reception signal generated by the reception structure from the received alternating field can be introduced. The I/Q demodulator can then be set up to carry out I/Q demodulation of the received signal.

In an advantageous embodiment, the at least one transmitting structure and the at least one receiving structure can be arranged in a contact surface, preferably a backrest or a lying surface, next to one another and/or one below the other and/or diagonally to one another. For this purpose, the transmitting structure and/or the receiving structure can be designed, for example, as flat coils.

Advantageously, the device according to the invention or a device for carrying out the method according to the invention can have one or more transmitting and/or receiving antennas as transmitting or receiving structures. The number of I/Q signal pairs can then be determined from the number of Transmission signals emitted by one or more antennas (e.g. separated into several frequency bands or through different signal modulation) multiplied by the number of receiving antennas result. The different transmission signals can be sent in the same way on all transmission antennas or, for example, each transmission antenna can send its own transmission signal that is different from the other transmission antennas. Any combinations of these are also possible.

Advantageously, the device can have N transmitting antennas and IVI receiving antennas, where N is a natural number greater than or equal to 1 and IVI is a natural number greater than or equal to 2, the N transmitting antennas being supplied with alternating voltages L of different frequencies, where L is a natural number greater or is equal to 1. Preferably, L can also be less than or equal to N. If L is equal to N, each transmitting antenna can be supplied with an alternating voltage of different frequency. If L is smaller than N, at least one of the transmitting antennas can be supplied with alternating voltages of several different frequencies.

In an advantageous variant, no transmitting antenna is subjected to several of the alternating voltages of different frequencies, so each transmitting antenna is subjected to an alternating voltage of exactly one frequency.

In a further advantageous variant, each of the receiving antennas can be set up to receive exactly one of the L frequencies or to receive several of the L frequencies. If each receiving antenna is set up to receive exactly one frequency, the number IVI of the receiving antennas is preferably equal to the number L of different frequencies of the radiated alternating voltages. If one or more of the receiving antennas are set up to receive multiple frequencies, the number of receiving antennas can be smaller than L.

Optionally, the number L of transmission signals of different transmission frequencies multiplied by the number IVI of reception antennas can result in the maximum number of I/Q signal pairs for the evaluation/algorithm. But there can also be fewer I/Q signals if, for example, not all receiving antennas are available all transmission frequencies are sensitive. A very broadband receiving antenna can also optionally be used, which does not have a specific frequency selectivity.

The following combinations of antennas of the transmitting and receiving structure are particularly advantageous:

One transmitting antenna and three receiving antennas. This results in three l/Q signal pairs.

Two transmitting antennas, both of which are preferably supplied with the same transmitting signal, i.e. the same alternating voltage, and five receiving antennas. This results in five I/Q data pairs.

Two transmitting antennas (preferably with different transmitting signals on, for example, different transmitting frequencies) and three receiving antennas. This results in six I/Q signal pairs.

The advantages of these multiple transmit/receive antennas lie in data redundancy if the signal quality is better or worse due to different positioning of the antennas for respective antenna pairs (the best transmit signal/receive signal combination can be selected by the algorithm). Alternatively, the algorithm can use all or any combination of the I/Q signal pairs to more specifically calculate out interference effects (e.g. movement artifacts) or to separate the actual useful information (breathing) from the interference effects and thus filter them out. In an advantageous embodiment, the selection of the I/Q signal pairs used can also be preceded by a signal quality estimate upstream of the evaluation algorithm, on the basis of which the most suitable signal pairs are selected for further processing.

In particular, it can be advantageous if the at least one transmission structure and/or the at least one reception structure has or is a broadband antenna. In this case, a bandwidth of the antenna can preferably be greater than or equal to one eighth of the carrier frequency, preferably greater than or equal to a quarter of the carrier frequency, preferably greater than or equal to half the carrier frequency and/or the bandwidth can be less than or equal to the carrier frequency.

In an advantageous embodiment, the device according to the invention can have an adaptation device with which a transmission frequency band or a transmission frequency of the transmission structure and/or a reception frequency band or a reception frequency of the reception structure can be adapted. Such an adaptation device can advantageously have or be at least one capacitance diode. In this way, the at least one transmission structure and/or at least one reception structure can be adapted so that it is best suited for the intended measurement.

The transmitting structure and/or the receiving structure can each be connected via a cable to the signal generator, the demodulation unit or another component of the device. Advantageously, these cables can be forcibly balanced, for example by means of a termination resistor together with an impedance converter and/or by means of at least one current-compensated choke and/or at least one ferrite.

Advantageously, the at least one transmitting structure and/or the at least one receiving structure can be supplied with the alternating voltage at a first connection and can be supplied with an alternating voltage shifted by 180° compared to the alternating voltage at a second connection. For this purpose, the device can preferably have at least one balun, which generates two alternating voltages from the first alternating voltage, which have a phase offset of 180° to one another. The input alternating voltage can be that generated by the signal generator. As already described for the method, it can be advantageous if the measuring device has a plurality of transmission structures, i.e. two, three, four or more than four transmission structures and/or a plurality of reception structures, i.e. two, three, four or more than four reception structures having. This allows a more precise measurement result to be achieved.

The signal generator can advantageously be set up to superimpose a plurality of sinusoidal alternating voltages as the alternating voltage. to generate voltages that can advantageously also have time-varying frequencies.

The method and device presented here enable a timely derivation of dynamic states of the target system. With the present algorithm, the data of a non-invasive, non-obstructive and scalable (simultaneous or serial use of several sensor sub-elements at a local distance) and surface and volume-sensitive measurement technology (alternating field penetrates at least partially into the target structure) based on the time-dependent complex folding at least two electromagnetic fields (sent and received) are processed at runtime, interpreted and converted, for example, into cardiothoracic vital data or other physidynamic data of a target structure. For this purpose, the system can at least initially be trained to map the changes in the data of the measurement system resulting from the physiodynamic changes in the target structure or a generic model of the target structure with reference system data (e.g. cardiothoracic vital data). The following may be possible, for example:

Detection of spontaneous breathing in mechanically ventilated patients

Determination of the expiratory and inspiratory phases and follow-up of patients

Alarm if specific vital parameters are interrupted. Contactless recording of ECG curves

Determination of heart rate variability and/or respiratory volumes, volume flows and/or changes

Predicting complications during surgery by measuring vital signs and monitoring data streams, analyzing cardiovascular data to predict cardiovascular disease,

Tracking patients' progress in rehabilitation.

The invention can also be used for one or more of the following applications:

Fitness tracking: monitoring physical activities and analysis of fitness or training progress.

In sports medicine, for example for performance analysis and training planning.

In health monitoring, for example for monitoring chronic diseases, e.g. long-term ECG.

For respiratory gating during radiotherapy of patients.

In stress management, for example to monitor stress levels in people, but also, for example, in mice.

In telemedicine, for example, for remote monitoring of elderly or chronically ill patients.

The invention will be explained below by way of example using a few figures. The same reference numbers indicate the same or corresponding features. The features described in the examples can also be combined between the examples and implemented independently of the respective example.

It shows:

1 shows a schematic representation of a device for carrying out a method according to the invention,

2 is a schematic representation of an exemplary sequence of a method according to the invention,

3 shows exemplary curves of I data and Q data as well as of derivatives dQ/dl during a breathing process,

4 exemplary demodulation data and data generated from them,

5 shows an example of preprocessing of I and Q data, 6 shows an exemplary sequence of training a statistical model,

7 shows an exemplary configuration for obtaining data for training in FIG. 6,

8 shows an exemplary model used for classification,

9 shows an exemplary model used for regression,

10 shows an example of using a neural network as a statistical model for estimating a respiratory flow,

11 shows the use of an FIR filter as a statistical model for estimating respiratory flow,

12 combinations of an electromagnetic transmission measurement to form a soft sensor system,

13 shows an exemplary connection of a transmitting antenna with a tunable matching network,

14 shows an example of using buffer circuits with termination to bridge a cable without mismatch, and

Fig. 15 shows a detailed example of a method according to the invention.

1 shows schematically a device according to the invention for measuring a process in an object 5, with which the method according to the invention for measuring a process in an object can also be carried out. The device has a transmission structure 4 with which an alternating electromagnetic field can be irradiated into the object 5 by generating an alternating voltage by a signal generator 2, here a frequency generator 2, with which the transmission structure 4 is applied. The device also has a receiving structure 6 with which the alternating field can be received after it has passed through the object 5. The device according to the invention also has a demodulation unit 8, with which the received alternating field can be demodulated to generate multi-dimensional demodulation data. In addition, the device has a control unit 1, which functions in particular as a modeling unit. The demodulation data and/or data generated from these are supplied to this modeling unit 1. With the modeling unit 1, at least one feature of a process and/or a temporal representation of the process can then be generated in a time-dependent manner using a statistical model from the supplied multidimensional demodulation data and/or the data generated from them.

In the example shown in FIG. 1, the device also has an optional transmission amplifier 3, to which the signal generated by the signal generator is supplied and which amplifies it and then outputs it to the transmission structure 4. In addition, in the example shown, the device has an optional reception amplifier 7, into which the signal received by the reception structure is introduced and which amplifies it and supplies the amplified signal to the demodulation device 8.

In the device shown, the alternating voltage generated by the frequency generator 2 can also be fed to the demodulation unit 8, which can thereby carry out I/Q demodulation of the received signal.

The control unit 1 can take over the configuration of the entire system. For example, you can set the transmission frequency and transmission power. The signal or frequency generator 2 can generate an alternating voltage with the configured frequency and optionally power. It also has two synchronous outputs. The first output is connected to the optional transmit amplifier 3 or connected directly to the transmit structure. The second output serves as a reference path for the I/Q demodulator 8, which acts as a receiver. The alternating voltage present at the transmitting structure 4 is emitted by it as an electromagnetic wave or electromagnetic field. The wave or field is transmitted through the object 5, for example a patient, and after transmission is converted back into an alternating voltage by the receiving structure 6. During transmission, the electromagnetic wave or the electromagnetic field modulated by processes inside the patient, such as cardio-respiratory activity.

The signal received by the reception structure 6 is forwarded to the optional reception amplifier 7 or directly to the I/Q demodulator 8. This can mix the received signal into the baseband with the help of the reference signal from the frequency generator 2. The I/Q demodulator can generate a complex output signal with an in-phase channel (I for short) and a quadrature channel (Q for short). These signals correspond to the modulation of the real and imaginary parts of the wave transmitted through the object 5. I and Q are then digitized by a data sink 1, which is here part of the control unit 1.

In other embodiments of the invention, it may also be possible to increase the number of transmitting structures 4 and receiving structures 6. This can be done symmetrically or asymmetrically. The same number of transmitting structures 4 as receiving structures 6 can be provided, or different numbers can be provided. The use of several receiving structures 6 per transmitting structure 4 can be advantageous. It can be particularly advantageous to attach the transmitting and receiving structures 4, 6 in a coplanar arrangement, e.g. B. be in a backrest or a mattress.

Since both the transmitting structures 4 and the receiving structures 6 can be attached very close to the body when used on people, the transmitting structures 4 and the receiving structures 6 can experience a detuning of their resonance frequency. In order to counteract this detuning and still achieve good transmission and reception performance, the following configurations of the transmission and reception structures can be advantageous:

A very broadband antenna or generally an antenna with a high level of robustness against detuning can advantageously be used.

A tunable matching network can advantageously be used in front of the actual transmission and reception structure. A possible embodiment can use capacitance diodes. Fig. 13 shows an example of such a tunable matching network for transmission. As soon as the object is in the When the near field of the transmitting and receiving structure comes, the detuning of the transmitting or receiving structure can be compensated for using control voltages. As a metric for the quality of the compensation, e.g. B. the size of the received signal can be used.

The circuit shown in Fig. 13 is a matching network. This allows the impedance of the antenna to be transformed to a different impedance on port TX. The ideal case here is power matching in which the impedance of the circuit parts coming from TX is equal to the impedance of the matching network with antenna. This allows mismatches in the antenna to be compensated for, which can lead to power losses and reflections in the signal generator or IQ demodulator.

The matching network shown in Fig. 13 corresponds to an n-circuit. With this circuit topology, any antenna impedance can theoretically be mapped to any impedance on the TX. The basic impedance adjustment is carried out by the inductance L_antenna and the two capacitances Cpl and Cp2. However, as soon as the antenna impedance changes, for example due to an object in the antenna's near field, the matching network consisting of L_antenna, Cpl and Cp2 is no longer correct. For this reason, additional capacitance diodes Dl, D2 and D3 are introduced. By reducing the control voltages Vtunel and Vtune2, the capacitance of the varactor diodes Dl and D2 can be increased and the effective parallel capacitances of the matching network can be increased or can also be increased by increasing the control voltages. The capacitance of D3 can be varied using the control voltage Vtune3, thereby allowing the inductance of L_antenna to be varied for a specific working frequency.

The high-frequency part of the circuit is separated from the control voltages by the inductors L_dc; these should preferably be chosen to be significantly larger than L_antenna in order not to have a significant influence on the matching network. In order to prevent the control voltages from having an impact on other circuit parts, they are separated from them by the capacitances C_dc. The capacitances C_dc must be chosen to be much larger than the capacitances Cpl and Cp2 in order not to have a significant effect. to have access to the matching network. An adaptable matching network is advantageously used in both transmitting and receiving structures.

The detuning of the transmit/receive structure can lead to a mismatch between the transmit/receive gain or the frequency generator/IQ demodulator. With long cable lengths between the transmit/receive structure and the actual signal source or sink, standing waves can arise due to the mismatch. This is especially the case when using cables. These standing waves can create artifacts, but also unwanted sensitive areas along the cable. In order to counteract these artifacts, the matching network described above and/or the following embodiments may be advantageous.

Forced symmetrization can be carried out between the leading and returning conductors of the cable to at least one transmitting structure and/or to at least one receiving structure, e.g. B. via current-compensated chokes, by means of a termination resistor together with an impedance converter, by means of at least one current-compensated choke and/or by means of ferrites.

It is also possible to use differential transmission and reception structures and/or the use of baluns.

It is also possible to terminate the cables with the wave impedance of the cable (typically 'Q) and to use buffer circuits to pick up the voltage at the termination with high impedance and pass it on to the next segment in the signal chain with low impedance. This is shown as an example in FIG. 14.

In Fig. 14, matching resistors are shown together with impedance converters as a buffer circuit. There is a termination resistor Rt2 or Rt4 to ground at the non-inverting inputs of the operational amplifiers Ul and U2. Since the inputs of the operational amplifiers usually have a much higher impedance than the termination resistors, only the termination resistor is relevant when considering the total impedance of the termination resistor and operational amplifier input. That's how he can Termination resistor must be chosen so that it corresponds exactly to the wave impedance of the previous circuit part in order to prevent mismatches caused by impedance jumps. The termination resistor Rt2 in Fig. 14 has the same impedance as the coaxial cable in front of it with 50 ohm characteristic impedance. The series resistors Rtl and Rt3 can also be used to adjust the output resistances of the operational amplifiers Ul and U2 so that they correspond to the wave impedance of the downstream circuit parts. Since the impedance of the operational amplifier outputs is significantly lower impedance than the resistors Rtl and Rt3, the total resistance from the operational amplifier output and series resistance is formed primarily by the series resistance. The series resistor Rt3 in Fig. 14 has the same impedance as the subsequent coaxial cable with 50 ohm characteristic impedance. The values of the termination resistors and the series resistors can differ from each other, for example to connect two circuit parts with different impedances without mismatch.

For the differential transmitting and receiving structure described, the at least one transmitting structure 4 and/or the at least one receiving structure 6 can be supplied with a first alternating voltage at a first connection and can be supplied with a second alternating voltage shifted by 180° compared to the first alternating voltage at a second connection become. When using a balun, it can generate two alternating voltages from an input alternating voltage, which have a phase offset of 180° to one another.

Fig. 2 shows an exemplary process of a method according to the invention for measuring at least one process in an object. The process can be carried out, for example, in the device shown in FIG. An alternating field is received by the receiving structure 6, which in the example shown is e.g. B. can be an RF signal. This is fed to the I/Q demodulation 21, which in the example shown generates I data and Q data per antenna. These are fed to an optional pre-processing step 22, which z. B. may contain filtering and/or transformation. In the preprocessing step 22, data generated from the multidimensional demodulation data, here the I and Q data per antenna, can be generated, which can also be referred to as guided variables can be referred to. The data generated by the preprocessing can now be fed to a statistical model 24. In addition or alternatively, it is possible to first carry out a feature extraction in the generated variables, with which features of the multidimensional demodulation data and/or the data generated therefrom are determined. These features can then be supplied to model 24. Such a feature generation is particularly useful if the multidimensional demodulation data is unsuitable or poorly suited for direct use in the model 24.

The model generates from the supplied multi-dimensional modulation data, here I and Q, and/or from the data generated from these, depending on time, at least one feature of the process and/or a temporal representation of the process, e.g. B. as predicted values, for example a respiratory flow curve. This result can optionally be fed to a curve interpretation 25. In this way, for example, a real-time carrier, volumes and/or a comparison to a reference can be generated.

Fig. 3 shows an example of the course of I and Q in the left part of the image and of the derivatives of dQ/dl in the right part of the image. The course shown is characteristic of a breath. In the left part of the image, the lower left section of the curve corresponds to inhalation and the upper right section corresponds to exhalation. So during a breathing cycle, Q and I move counterclockwise along the curve shown.

In the right part of the picture, the gradient dQ/dl is shown as arrows on the curve shown in the left part of the picture. The change in direction of the gradient can be seen when changing from inspiration to expiration. Using the gradient, it can be recognized in real time, for example, whether there is an inhalation or an exhalation.

Fig. 4 shows examples of preprocessing of data that can be obtained in the demodulation step. This preprocessing can take place, for example, in step 22 in FIG. 2. In the partial images of Fig. 4, a 3 x 3 pair plot of the data in the raw state is shown on the left. In the pair plots, the non- Diagonal plots show the dependence of the raw data on other raw data, for example to show the correlation. Histograms of the raw data are shown in the diagonal plots. In the right panel, the pair plots of the corresponding data are shown at corresponding positions after preprocessing. The preprocessing can e.g. B. include filtering and/or a transformation.

The preprocessing can also e.g. B. include an inspection of data, including the investigation of statistical relationships between the input variables, e.g. B. via a Generalized Scatterplot Matrix (GSPLOM), as shown in Fig. 4. The preprocessing can then be carried out e.g. B. have smoothing, exemption from the mean and/or scaling.

Such preprocessing can be used both when measuring the process and when training the statistical model.

5 shows an example of a preprocessing step 22. Here, for example, I and Q are supplied to the preprocessing step 22 as multidimensional demodulation data. This can then generate one or more of the variables shown on the right in FIG. 5 from this data. The sizes shown on the right are just examples and some of these sizes, all of these sizes and also other sizes can be created.

6 shows an example of a process with which the statistical model 24 can be trained in a training process. In the training process, the statistical model 24 is trained based on data. Model 24 can then be applied to new data during operation. A distinction can be made in particular between two supervised learning tasks, namely classification and regression.

During the training process, the model 24 receives, in addition to the demodulation data generated from a received alternating field, additional measured values from at least one reference sensor (ground truth). The at least one reference sensor can measure the quantity that is to be measured or predicted using the statistical model. By comparing the measured values of the Reference sensors with the values of the feature of the process predicted by the model, and/or the temporal representation of the process, an error value can be calculated, for example by means of an optimization algorithm, in order to train the model 24 accordingly. The training can e.g. B. can be carried out in a generalized manner with an entire representative population or it can take place individually, for example on a patient-specific basis. During patient-specific training, the reference sensor system can initially be connected to the patient in order to train the mathematical model individually. After training, the reference sensors can be removed.

As shown in Fig. 6, the model 24 for training is supplied with the demodulation data or data generated from these that are to be used to measure the process in the object during operation. In FIG. 6, these are, for example, I, Q, an amplitude, a phase and gradients of the received alternating field. The model 24 generates an estimated value from these, which is fed to the quality metric 62. The quality metric 62 compares the estimated value with the value measured by the reference sensors and produces an error value. This is fed to an optimization algorithm 61, which then trains the model 24 so that the agreement between the measured value and the value estimated by the model 24 is improved.

7 shows an example of an arrangement for teaching or training the model 24 in FIG. 6. In the example shown in FIG. 1 shown device 72, via the transmitting structure 4 with an alternating field, which is received by the receiving structure 6 after it has passed through the patient 71. A feature of an affirmative process and/or a time representation of the process can be generated from the received alternating field, as described in FIG. 2. In the example shown, the process to be measured is a cardiorespiratory activity of the patient. An electrocardiogram 73 and a flow meter 74 for directly measuring a respiratory flow are provided as reference sensors. The sensors 73 and 74 generate the measured values that are fed to the quality metric 62. This compares the values of the process features of interest with those predicted by the model and generates the said error value.

During operation, the model 24 no longer receives reference data, but rather estimates the characteristics of the process and/or the temporal data of the process based on the various input data. The system can act as a soft sensor here, since the target variable to be determined is not measured directly, but is estimated using correlated variables and a trained model.

Two types of estimation can be distinguished, both of which are advantageous. In one case, a classification of the demodulation data and/or the data generated from them can take place. In this case, variables with a discrete range of values, mostly binary, sometimes also ternary, can be distinguished. For example, the data could be classified as inhalation, exhalation, or optionally a pause in breathing. It can e.g. B. based on the current demodulation data, a distinction can be made as to whether inhalation, exhalation or a breathing pause is currently taking place. For example, it can also be determined whether an R wave occurred in the previous time slice.

8 shows such a classification again schematically. The said demodulation data and/or data generated from these serve as input data for the model 24. The model classifies the specific constellation of this data into classification results such as inhalation, exhalation, presence of an R wave, etc.

The second type of estimation is regression. In regression, the model generates quasi-time-continuous target variables with a continuous or quasi-continuous range of values. Such a target size can e.g. B. the current respiratory flow or an excitation curve of an ECG.

Fig. 9 shows schematically the model as a regression. The demodulation data or data generated from these are again entered and the model uses them to generate, for example, the respiratory flow, an ECG or a similar curve.

10 shows an example of an embodiment of the model 24 as a trained neural network, in this case a multi-layer perceptron based on the Input variables I and Q and the derived features from I and Q estimate the respiratory flow. The left partial image shows an example of a course of I and Q over time, which is supplied as a model to the neural network 24, and the right partial image shows a comparison of a predicted flow with an actual flow, as determined using a reference sensor. It can be seen that the trained neural network 24 predicts the respiratory flow very well. The neural network has an input layer, four hidden layers with 200, 50, 60 and another 60 neurons and an output layer.

11 shows an example of respiratory flow estimation in which the statistical model 24 is a trained adaptive FIR filter. The input variables here are the I data and the Q data, as shown in the left part of the image. The result is again the respiratory flow, as shown in the right part of the image in comparison to the measured respiratory flow.

Fig. 12 shows another example of a soft sensor system. The method shown in FIG. 12 begins with an electromagnetic transmission measurement, for example using the arrangement shown in FIG. 1. The data is in turn fed to preprocessing 22, which is also optional here. The data generated in this way is then fed to the model 24, which was previously trained with a reference system 124. Optionally, the data can be fed to a post-processing step 122 and ultimately output as an estimated value in the output step 123. This soft sensor system in turn does not measure the output variable directly, but rather estimates the output variable via the mathematical model 24 using several input variables correlated with the output variable.

Fig. 15 shows a more detailed example of a method according to the invention. In a step 151, flow data is imported, which is subjected to linear interpolation in a step 152. The data generated by the linear interpolation is subjected to an exemplary Butterworth bandpass filter in a step 153, the lower limit of which in the example shown is 0.11 Hz and the upper limit of which is 1.1 Hz. The data filtered in this way is then subjected to a Yeo-Johnson performance transformation 154 which generates the reference flow as ground truth 165. This reference flow is used as reference sensor data for training the statistical model 24, which here is a multi-layer perceptron regressor.

The reference data determined in steps 151, 152, 153, 154 and 165 is used for training only. In operation as a process for measuring the process, these steps are omitted.

During operation and for estimation during training, demodulation data, for example Q and I, are generated in a step 155. These are subjected to linear interpolation in a step 156. In a step 157, for example, an amplitude is calculated from the interpolated data. The result of step 157 is fed here to a Butterworth bandpass filter 158 and the filtered signal is in turn fed to a Yeo-Johnson power transformation 159. Numerous characteristics of the course to be measured can now be determined from the data transformed in the Yeo-Johnson performance transformation 159. For example, the data generated in 159 can be fed to a shift operator 164, whereby shifted data can be determined, for example Q, I, amplitude, dQ/dt, dl/dt, dAmplitude/dt, binary dQ/dt, binary dl/dt and/or or binary dAmplitude/dt, as shown in 167. The data generated by transformation 159 can also be used directly as features Q, I, amplitude as in 171.

In addition, gradients can be generated in step 160, which in turn can be subjected to a Yeo-Johnson power transformation 161 and can thereby generate the time-dependent change in Q, I and the amplitude. This is shown in 170. The data generated by transformation 161 may also be subjected to gradient binarization 162, thereby producing the binary dQ/dt, dl/dt, dAmplitude/dt shown in 169.

The data generated in step 162 may also be fed to a moving average method along with the data generated in transformation 161. This allows the data shown in Figure 168 to be generated. The features 166 can then be fed to the statistical model 24. This creates the process as a regression.

Due to the non-contact nature, the invention is also suitable for patients who have easily irritable, burned or inflamed skin.

It is also suitable for very small patients who do not have much space to attach sensors, e.g. premature babies or newborns.

With the invention, due to the non-contact nature, the probability of poor measurement results due to user errors is low. With ECGs, for example, the electrodes can be placed incorrectly or poorly. When measuring respiration with flow or pressure sensors, leaks can occur, which can distort both the flow and pressure measurements. Leaks in face masks often occur among people with beards.

Only a low level of patient compliance is required for the use of the invention. The electromagnetic transmission measurement is imperceptible to patients and can, for example, be easily integrated into the back of a chair or mattress so that they can adopt a comfortable position. Tightness that may occur with procedures such as RIP wristbands or ECG chest straps is completely eliminated. Since the invention can also measure through thick textiles, patients do not have to undress.

Due to the lack of contact, the risk of contamination and wear on the transmitting and receiving structures is also significantly reduced.

By implementing a soft sensor, the invention can virtually emulate conventional sensors such as ECG or flowmeters. This allows the invention to be seamlessly integrated into existing sensor chains in which a conventional sensor is normally used.

With the present algorithm, the data can be used in a non-invasive, non-obstructive and scalable (simultaneous or serial use of multiple more sensor sub-elements at a local distance) and surface and volume sensitive measurement technology (fields penetrate at least partially into the target structure) based on the time-dependent complex folding of at least two electromagnetic fields (sent and received) processed at runtime, interpreted and converted into cardiothoracic vital data or other physiodynamics - technical data of a target structure are converted. For this purpose, the system can at least initially be trained to map the changes in the data of the measurement system resulting from the physiodynamic changes in the target structure or a generic model of the target structure with reference system data (e.g. cardiothoracic vital data).

The invention is particularly suitable for the following applications:

Sensor in artificial ventilation, monitoring for sudden infant death, monitoring the progress of cardiac and respiratory diseases (comparable to a long-term ECG), tracking the current state of a person (awake, sleeping, pathological), contactless patient monitoring in intensive care units, contactless recording of an ECG, contactless Tracking of pacemakers, contactless determination of the breathing rate, installation in wearables, mattresses, car seats -> stress display etc., sleep evaluation, measuring training parameters e.g. VOZmax, monitoring of sleep apnea, use for biofeedback, input data for artifact or failure compensation Vital data sensors, signal generators for dead man's switches, determination of performance data, use of the "procedure" for evaluating the data from the chest monitor, measurement methods beyond the derivation of pure cardiothoracic vital data (e.g. technical time-varying conditions such as fill levels, proximity, compression of bulk material or porous structures or biological such as the tension of muscle tissue, peristalsis, movement in general of humans as well as animals).

Claims

Patent claims

1. Method for measuring at least one process in an object, wherein an alternating electromagnetic field is irradiated into the object, the alternating field is received after it has passed through the object and has been modulated by the process in the object, in a demodulation step the received alternating field is used Generation of multi-dimensional demodulation data is demodulated, the multi-dimensional demodulation data and / or data generated from them are fed to at least one statistical model, and in a generation step by the at least one statistical model from the supplied multi-dimensional demodulation data and / or the data generated from them, time-dependent at least one feature of the process and/or a temporal representation of the process is generated.

2. Method according to the preceding claim, wherein the alternating electromagnetic field contains at least two components of different frequencies.

3. The method according to any one of the preceding claims, wherein the received alternating field is demodulated in the demodulation step in an ln phase & quadrature method for generating I data, I, and Q data, Q, as the multidimensional demodulation data, and the I data and the Q data and/or data generated from these are fed to the statistical model.

4. The method according to any one of the preceding claims, wherein the data generated from the multidimensional demodulation data and/or the I data, I, and the Q data, Q, contains at least one from the group comprising: a derivative from Q to I, a derivative of I with respect to time, a Derivation of Q with respect to time, an amplitude of the received alternating field, a phase of the received alternating field, an amplitude from the derivative of Q with respect to time and the derivative of I with respect to time, a phase from the derivative of Q with respect to time and the Derivative of I with respect to time, derivatives of amplitude with respect to time, derivatives of phase with respect to time, a phasor of the derivative of Q with respect to I, the first derivative of the generated data with respect to time, another generated quantity, the Q data and /or the I data, the second or higher derivative of the generated data with respect to time, another generated quantity, the Q data and/or the I data, the first or a higher integral of the generated data, the Q data and/or the I data according to time, another generated quantity, the Q data and/or the I data, binarizations of the generated data, binarizations of Q data, binarizations of I data, moving averages of the generated data, moving averages of Q data, moving averages of I data, Q data filtered by frequency filters, I data filtered by frequency filters, generated data filtered by frequency filters, generated data from past points in time, Q data from past points in time and /or from I-data from past points in time.

5. The method according to any one of the preceding claims, wherein at least one input feature is extracted from the demodulation data and / or from data generated by them and is fed to the statistical model, the at least one input feature preferably being a linear combination and / or a non-linear combination the multidimensional demodulation data and/or the data generated from it.

6. Method according to one of the preceding claims, wherein the object is a body, preferably a human or animal body.

7. The method according to any one of the preceding claims, wherein the process is a physiological process in a human or animal body. 8. The method according to any one of the preceding claims, wherein the statistical model is an artificial neural network, an artificial neural feedfoward network, an artificial neural recurrent network, an artificial convolutional network or a combination of the aforementioned artificial neural networks, a decision tree, a decision forest, a naive -Bayes model, a nearest neighbor model, a support vector machine, a linear regression model, a non-linear regression model, an adaptive filter, a Kalman filter, a gamma filter, a Markov model, a model from a linear discriminant analysis , a model from a non-linear discriminant analysis, a model from an independent component analysis, a model from a principal component analysis and / or an ensemble model from one or more of the aforementioned models.

9. The method according to any one of the preceding claims, wherein the alternating electromagnetic field is irradiated into the object via at least one transmission structure, in that an alternating voltage is generated by a signal generator, with which the at least one transmission structure is applied, and the alternating field is received by at least one reception structure after it has traversed the object.

10. The method according to the preceding claim, wherein the at least one transmitting structure and the at least one receiving structure are arranged on the same side of the object.

11. The method according to one of the two preceding claims, wherein the object is a human body and the at least one transmitting structure in the immediate vicinity, preferably at a distance of equal to or less than 20 cm, preferably at a distance of equal or less than 10 cm, and / or is arranged on a thorax and/or an abdomen of the body, and the at least one receiving structure is in the immediate vicinity, preferably at a distance of equal to or less than 20 cm, preferably at a distance of equal to or less than 10 cm, and/or on the thorax and/or the abdomen is arranged, wherein preferably the at least one transmitting structure and/or the at least one receiving structure is arranged laterally, dorsally or ventrally to the abdomen and/or laterally, dorsally or ventrally to the thorax.

12. The method according to any one of the preceding claims, wherein the alternating electromagnetic field is irradiated into the object via at least two transmit-receive pairs, each containing a transmitting structure and a receiving structure, and is received after it has passed through the object, advantageously at least one the at least one transmission structures can also be part of several of the transmission-reception pairs, and / or at least one of the at least one reception structures can also be part of several transmission-reception pairs, with particular advantage of generating separate demodulation data for each of the reception structures in a demodulation step, wherein the entirety of the demodulation data of all receiving structures can then be understood as the multi-dimensional demodulation data according to the preceding claims.

13. The method according to any one of the preceding claims, wherein the method is carried out regularly and repeatedly, a value of the at least one feature of the process being regularly repeatedly compared with at least one previous value of the feature, and an indication event is triggered when the value of the feature increases deviates more than a predetermined threshold value from the previous value of the feature, preferably adjusting parameters of the statistical model and / or the statistical model based on a difference between the value of the generated feature and the corresponding previous value of the feature when the value of the generated characteristic deviates from the previous value by less than the threshold value.

14. The method according to any one of the preceding claims, wherein the sequence includes at least one electrocardiogram, at least one impedance cardiogram, at least one ballistocardiogram, at least one seismocardiogram, a movement and/or an expansion of a chest and/or an abdomen, at least one blood pressure over time, at least one respiratory volume curve, at least one respiratory pressure curve, at least one respiratory flow, at least one course of oxygen saturation over time, temporal data obtained through imaging, and/or is a plethysmogram. Method according to one of the preceding claims, wherein the at least one feature of the process is a trigger event or an exceeding or falling below a predetermined threshold value. Method according to one of the preceding claims, wherein the at least one feature of the process and/or the temporal representation of the process describes the process continuously, quasi-continuously or at a plurality of, preferably equidistant, times. Method according to one of the preceding claims, wherein the statistical model is trained before irradiation of the alternating electromagnetic field in a training process by irradiating an alternating electromagnetic training field into at least one training object, receiving the alternating training field after it has passed through the training object and through a sequence was modulated in the training object, the received alternating training field is demodulated to generate multi-dimensional demodulation data, the multi-dimensional demodulation data and/or training data generated from them are supplied to the statistical model, by the statistical model from the supplied multi-dimensional demodulation data and / or at least one training feature of the process is generated in a time-dependent manner from the data generated from these, the at least one training feature is compared in a time-dependent manner with a measured feature of the process and parameters of the statistical model and / or the statistical model based on a result of this comparison is adjusted. 18. The method according to the preceding claim, wherein the training process is carried out for a plurality of training objects that are representative of the object or wherein the training process is carried out for the object as the training object.

19. Method for controlling a ventilation process, wherein in a method according to one of the preceding claims a breathing process is measured as the process and the onset of a spontaneous breathing process is determined as the at least one feature of the process, the ventilation process being controlled in such a way that no ventilation against the Spontaneous breathing takes place.

20. The method according to any one of the preceding claims, wherein the statistical model is a neural network with an inference latency of less or equal to 200ms, preferably less or equal to 150ms, preferably less or equal to 100ms, preferably less or equal to 50ms, preferably less or equal to 30ms, preferably is less or equal to 20ms, preferably less or equal to 10ms, preferably less or equal to 5ms, preferably less or equal to 1ms.

21. Device for measuring at least one process in an object, having at least one transmitting structure with which an alternating electromagnetic field can be irradiated into the object by generating an alternating voltage by a signal generator with which the at least one transmitting structure is applied, at least one receiving structure, with which the alternating field can be received after it has passed through the object, further comprising a demodulation unit with which the received alternating field can be demodulated to generate multi-dimensional demodulation data, further comprising a modeling unit to which the demodulation data and / or data generated from them can be fed, and with the multidimensional data supplied by a statistical model Demodulation data and / or the data generated from them, at least one feature of the process and / or a time representation of the process can be generated depending on time. Device according to the preceding claim, wherein the device has an I/Q demodulator into which the alternating voltage and a received signal generated by the receiving structure from the received alternating field can be introduced, and which is set up to carry out an I/Q demodulation of the received signal . Device according to one of the two preceding claims, wherein the at least one transmitting structure and the at least one receiving structure in a contact surface, preferably in a backrest, at least one siderest, at least one armrest, and / or a lying surface, next to each other and / or among each other and / or diagonally are arranged to each other. Device according to one of claims 21 to 23, wherein the at least one transmitting structure and / or the at least one receiving structure has a broadband antenna, preferably a bandwidth of the antenna greater than or equal to an eighth of the carrier frequency, preferably greater than or equal to a quarter of the carrier frequency, preferably greater or is equal to half of the carrier frequency and/or the bandwidth is less than or equal to the carrier frequency. Device according to one of claims 21 to 24, comprising an adaptation device with which a transmission frequency band of the transmission structure and / or a reception frequency band of the reception structure can be adapted, the adaptation device advantageously having or being at least one varactor diode. Device according to one of claims 21 to 25, wherein at least one cable to which the transmitting structure and/or the receiving structure is connected is forcibly balanced, preferably by means of a termination resistor together with an impedance converter and/or at least one current-compensated choke and/or ferrites.

27. Device according to one of claims 21 to 26, wherein the at least one transmitting structure and / or the at least one receiving structure are supplied with the alternating voltage as a first alternating voltage at a first connection and with an alternating voltage 180° relative to the first alternating voltage at a second connection shifted second alternating voltage are applied, the device preferably having at least one balun which generates two alternating voltages from an input alternating voltage, which have a phase offset of 180 ° to one another.

28. Device according to one of claims 21 to 27, wherein the device has two, three, four or more than four of the transmitting structures and / or has two, three, four or more than four of the receiving structures.

29. Device according to one of claims 21 to 28, wherein the device has N transmitting antennas and IVI receiving antennas, where N is a natural number greater than or equal to 1 and IVI is a natural number greater than or equal to 2, the N transmitting antennas having alternating voltages L different Frequencies are applied, where L is a natural number greater than or equal to 1 and preferably less than or equal to N, preferably no transmitting antenna being applied to several of the alternating voltages of different frequencies and / or each of the receiving antennas for receiving exactly one of the L frequencies or for receiving several of the L frequencies is set up.

30. Device according to one of claims 21 to 29, wherein the signal generator is set up to generate the alternating voltage with different and / or time-varying frequencies.

31. Device according to one of claims 21 to 30, wherein a method according to one of claims 1 to 20 can be carried out with the device.