System and method for detection of an event

Abstract

A system for identification of an event at a toilet visit, said system comprising: a sensor (102) configured to be arranged on a toilet (100), and configured to detect a presence of a person making the toilet visit; a radar (106) configured to be arranged on the toilet, and further configured to: transmit a radar signal to the toilet bowl (108) and detect echoes caused by the radar signal and generate radar data based on the echoes; a processor (110) configured to: receive the radar data, and analyze radar data based on event characteristics to identify the event based on correspondence between the radar data and at least one of the event characteristics, wherein the event characteristics comprises at least event characteristics of a feces event.

Description

[0001] SYSTEM AND METHOD FOR DETECTION OF AN EVENT

[0002] Field of the invention

[0003] The present invention is related to a system and a method for detection of an event at a toilet visit.

[0004] Background art

[0005] Toilets play a crucial role in the sanitation of your household and in keeping human people healthy.

[0006] Humans benefit from going to the toilet to remove waste products from the body. This process helps maintain our health by getting rid of substances that the body doesn’t need or that could be harmful if they build up. Urination and defecation are essential for regulating the body’s balance of water, salts, and other chemicals, and for eliminating toxins. Regularly going to the toilet also helps prevent discomfort and potential health issues like infections or digestive problems.

[0007] Stool withholding can result in painful bowel movements, since more water will be absorbed from the retained stool, which becomes harder and may cause pain once the toilet visit happens. This may result in a vicious cycle of more withholding behavior, which in turn can result in a large, hard lump of stool stuck in the bowel. This is called fecal impaction. Frequent fecal impactions can cause a megarectum, with symptoms of fecal incontinence overflow, decreased rectal sensation and, in the end, an impaired sensation and urge to defecate. If repeated over a long period of time, suppression of the urge to defecate may eventually lead to conditions such as dyssynergy defecation and slow transit constipation.

[0008] Summary of the invention

[0009] To achieve at least one of the above objects and also other objects that will be evident from the following description, a system having the features defined in claim 1, and a method having the features defined in claim 10 are provided according to the present invention. Preferred embodiments will be evident from the dependent claims.More specifically, there is provided according to a first aspect of the present invention a system for identification of an event at a toilet visit. The system comprising:

[0010] a sensor configured to be arranged on a toilet, and configured to detect a presence of a person making the toilet visit;

[0011] a radar configured to be arranged on the toilet, and further configured to:

[0012] transmit a radar signal to the toilet bowl and detect echoes caused by the radar signal and

[0013] generate radar data based on the echoes;

[0014] a processor configured to:

[0015] receive the radar data, and

[0016] analyze radar data based on event characteristics to identify the event based on correspondence between the radar data and at least one of the event characteristics, wherein the event characteristics comprises at least event characteristics of a feces event.

[0017] The provided system has the ability to detect and identify an event of a toilet visit, meaning that the system can identify if a person going to the toilet discharges any excreta in the toilet.

[0018] Sometimes ill people do not remember if they have excreted or not when having been to a toilet visit. This may often be the case with for example people suffering from dementia or other illnesses. A person who does not know if they have excreted or not may be subjected to constipation or further issues that my come from withholding urine and feces. The provided system thus provides improved healthcare as it can identify if excreta has landed in the toilet. The system may provide an alert system if a toilet registers a number of toilet visits above a threshold value wherein the event is null. Thereby any unintentional stool withholding may be detected, and an action may be taken.

[0019] The sensor detects if a person is at the toilet for making the visit. The radar transmits the signal to the toilet bowl and the reflection of the radar is analyzed to identify what event happened in the toilet, and if it was a feces event or another event.The radar transmits the signal into the toilet at in a downwards degree such that the signal is transmitted to the bowl. In a preferred example the radar is pointed at a 45-degree angle downwards into the toilet. It should be noted that the radar may be positioned at different angles downwards into the toilet.

[0020] Thanks to the system using a radar, the system is able to provide accurate identification of events using a simple detector. The system may facilitate mounting of at least the sensor and the radar on a toilet for providing a simple integration of detection of events on the toilet.

[0021] The processor may be any processor that may be able to analyze radar data. The processor may for instance be implemented by a general-purpose processing unit, e.g., a central processing unit (CPU), which may execute instructions of one or more computer programs in order to implement functionality of the processor. The processor may also or alternatively be implemented as firmware arranged e.g., in an embedded system, or as a specifically designed processing unit, such as an Application-Specific Integrated Circuit (ASIC) or a Field-Programmable Gate Array (FPGA). It should be realized that the processor may be implemented as a combination of hardware and software components.

[0022] It should further be realized that the processor may be arranged in any relation to the sensor and the radar. For instance, the processor may be arranged in a common housing with the radar and / or the sensor or in a separate housing that may be arranged on the toilet. Thus, the processor may be configured for wired communication or short-distance wireless communication with the sensor and / or the radar. However, the processor may alternatively be arranged at a different location and may be configured to receive information from the sensor and the radar. The processor may comprise a plurality of different entities that may be arranged in different locations for performing different parts of the processing. The processor may be provided anywhere, such as “in the cloud”. The processor may communicate with the sensor and the radar through a computer network and / or telecommunication network, such as through the Internet, enabling the processor to be arranged anywhere in relation to the sensor and the radar.

[0023] The sensor may be configured to trigger the radar for transmitting the radar signal and generating the radar data. Thus, the processor may not needto receive any information from the sensor. Rather, the sensor may only be used for triggering generation of radar data.

[0024] However, in other examples, the processor may also be configured to receive data from the sensor regarding the presence of the person making the toilet visit. This may be combined with the radar data for providing further analysis of the event using the data from the sensor.

[0025] The system may comprise a storing unit configured to store the event characteristics. The processor may be configured to access said storing unit.

[0026] The storing unit may thus contain characteristics that correspond to different events such as feces and / or urine. The event characteristics may contain characteristics for defining different types of feces such as loose or hard etc. The storing unit may be located on a server or any suitable location. In the example, the processor accesses the event characteristics in the storing unit to compare the radar data to the stored event characteristics. The processor the determines the event based on a correspondence between the radar data and the stored event characteristics.

[0027] In some examples, the radar may be configured to be in wireless communication with the processor.

[0028] In some examples the sensor may be configured to be in wireless communication with the processor. Thereby limiting the need for wires connected to the radar and / or sensor. Another benefit of a wireless communication is that the radar can be placed such that it is hidden from visibility as no wires will extend from the radar around the toilet area.

[0029] In some examples the sensor may be a pressure sensor. Thereby for example, the system may detect if a person is sitting on the toilet seat. It should be noted that the sensor may be any type of suitable sensor to detect a human presence. Some examples of sensors are a light sensor, or a heat sensor. In some examples, the sensor may be located in the room of the toilet and not on the toilet itself.

[0030] In some examples the radar may be configured to transmit radar wavelets.

[0031] This may facilitate providing a system with low power consumption. Further, the system may be compact, as the radar being configured totransmit radar wavelets may be very small. Also, this facilitates providing radar data with high precision facilitating correct analysis of the radar data.

[0032] As used herein, the term “radar wavelet” means a signal having a short pulse. The radar is configured to transmit a signal during a short period of time. The transmitted signal may be reflected by an object and the radar is further configured to detect echo(es) in the form of a reflected signal.

[0033] The detected echoes may be dependent on the object causing the transmitted radar signal to be reflected. The amount of energy received back may depend on reflectivity of the object causing reflection, which may in turn be dependent on relative permittivity of the object in relation to a surrounding medium.

[0034] It should be realized that the amount of energy in detected echoes may thus be dependent on the object causing the reflection of the transmitted radar signal. For instance, a detected echo may be different for a feces event comparted to a urine event. The processor may thus be configured to identify the event at least based on the amount of energy received in an echo.

[0035] Using radar wavelets, the possibility of correctly identifying events and differentiating between different events may be facilitated.

[0036] In some examples, the radar is configured to detect amplitude and phase information of the detected echoes. In this respect, the radar may be referred to as a coherent radar.

[0037] The radar may be configured to accurately set a starting phase when a radar wavelet is transmitted. The radar may further be configured to determine a phase of the received signal in relation to the starting phase. This may be used for accurately determining movements and / or speed of the object causing reflection of the radar signal.

[0038] In some examples the event characteristics may be represented by a machine-learned model. The processor may be configured to have access to the machine-learned model for identifying the event.

[0039] This may facilitate analysis of the radar data. The machine-learned model may enable accurate identification of events and differentiating between different events. In addition, the machine-learned model may be asimple model facilitating processing of the radar data without requiring computer-intensive calculations to be performed.

[0040] In some examples the event characteristics may comprise characteristics of a urine event.

[0041] Thereby the system may be able to detect a urine event in the toilet. In some examples, the sensor may be configured to be attached to a toilet seat.

[0042] In a second aspect of the inventive concept, there is provided a method for identification of an event at a toilet visit. The method comprises: detecting a presence of a person making the toilet visit;

[0043] transmit, by a radar, a radar signal to the toilet bowl and detecting echoes caused by the radar signal and

[0044] generating radar data based on the echoes;

[0045] receiving, by a processor, the radar data, and

[0046] analyzing radar data based on a set of event characteristics to identify the event based on correspondence between the radar data and at least one of the event characteristics, wherein the set of event characteristics comprises at least event characteristics of a feces event.

[0047] In some examples, the method may comprise a step of determining the volume of the event.

[0048] In some examples, the method may comprise a step of determining a volume of excreta in the toilet. According to a third aspect of the inventive concept there is provided a toilet seat comprising a sensor configured to detect a presence of a person making a toilet visit; a radar configured to: transmit a radar signal to the toilet bowl and detect echoes caused by the radar signal and generate radar data based on the echoes.

[0049] According to a fourth aspect, a system and a method are provided for detecting a volume event in a toilet. The system may determine a volume of a feces and / or a urine event that is execrated into the toilet. The system for detecting a volume may comprise all features mentioned in relation to the first aspect.In some examples the system may determine a volume that has entered the toilet and store at least one parameter in addition to the volume measurement in the storing unit.

[0050] In some examples, the system may detect a volume of a mass present in the toilet bowl.

[0051] A benefit or detecting and determining volume is that such a system may be able to prevent a flooding if the volume is so high that it can cause further problems with the toilet systems. Another benefit is that determining a volume can provide improved patient safety in a care facility. Instead of just detecting what event happens in the toilet, determining the volume may provide health care personnel with further information that may aid in treatment of the patient. Another benefit is prevention of diseases when identifying and determining volume of excreta in the toilet.

[0052] To better support health monitoring and preventive care, the provided system measures the volume of urine and faces. Tracking output volumes provides valuable insights into hydration status, kidney function, bowel health, and early signs of complications. It allows caregivers and healthcare providers to detect trends or abnormalities that might otherwise go unnoticed and thereby enable timely interventions.

[0053] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated.

[0054] Brief description of the drawings

[0055] The above, as well as additional objects, features and advantages of the present invention, will be better understood through the following illustrative and non-limiting detailed description of preferred embodiments of the present invention, with reference to the appended drawings, where the same reference numerals will be used for similar elements, wherein:

[0056] Fig. 1a shows a system according to an example of the inventive concept.Fig. 1b shows a system according to another example of the inventive concept.

[0057] Fig. 2 is a flowchart of a method for identification of an event at a toilet visit.

[0058] Fig. 3 shows another example of the inventive concept.

[0059] Fig. 4 shows a cross sectional view of a radar located in a toilet.

[0060] Fig. 5 Shows a network of toilets.

[0061] Description of embodiments

[0062] The present invention will now be described more fully hereinafter with reference to the accompanying drawings, in which currently preferred embodiments of the invention are shown. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided for thoroughness and completeness, and fully convey the scope of the invention to the skilled person.

[0063] Fig. 1a shows a system according to the present invention. The system is installed on a toilet 100. A sensor 102 is located at an underside of the toilet seat 104. The sensor 102 may be a pressure sensor 102. The pressure sensor can detect a change in pressure onto the toilet seat 104. When a change in pressure occurs, it can be detected that a person is sitting on the toilet seat 104. The sensor may be any type of sensor that can detect a presence of a person in the vicinity of the toilet to make a toilet visit.

[0064] A person making a toilet visit is a person excreting into the toilet, which can be for example feces and / or urine.

[0065] A radar 106 is attached to the toilet. In the shown figure the radar 106 is located at a back portion of the toilet bowl 108.

[0066] The radar 106 is pointed downwards towards the toilet bowl 108. The radar 106 is configured to transmit a radar signal into the toilet bowl 108. The radar signal will be reflected by objects in the toilet bowl 108 such that reflections will be caused by feces and / or urine. The reflection of the transmitted radar signal will form an echo traveling back to the radar 106 andso as to be detected by the radar 106 as a detected echo. Based on the detected radar echoes, radar data is generated.

[0067] A processor 110 is in the figure arranged on the toilet 100 and is adapted to receive the generated radar data. The processor is configured to analyze the received radar data. Upon analyzing the radar data against event characteristics, the processor is configured to generate an output. The output may be related to the analysis. The output may be an identification of a toilet event.

[0068] A battery 112 may optionally be arranged on the toilet to provide power to the system.

[0069] In Fig. 1b a similar system to that of Fig. 1a is shown. The shown system is a wireless system. Hence the radar 106 is in wireless communication with the processor 110. The sensor 102 is in wireless communication with the processor 110.

[0070] The processor 110 (not shown) may be located in another room from the toilet 110. In some examples, the processor 110 may be located “in the cloud”. The processor 110 may be on a computer. In some examples the processor 110 is on a server.

[0071] Referring nor to Fig. 2 a flowchart is shown of a method of identifying an event at a toilet visit.

[0072] The event is an event where a person discharges excreta into the toilet.

[0073] The method comprises the following steps:

[0074] (S10) detecting a presence of a person making the toilet visit;

[0075] (S12) transmitting, by a radar, a radar signal to the toilet bowl and detecting echoes caused by the radar signal and

[0076] (S14) generating radar data based on the echoes;

[0077] (S16) receiving, by a processor, the radar data, and

[0078] (S18) analyzing radar data based on a set of event characteristics to identify the event based on correspondence between the radar data and at least one of the event characteristics, wherein theset of event characteristics comprises at least event characteristics of a feces event.

[0079] In step S10, a presence of a person at the toilet is detected. The sensor 102 detects if a person is present to make the toilet visit. Thereby the system may be indicated to if it is excreta or something else that may be dropped into the toilet.

[0080] Step S10 may be performed by the pressure sensor 102 for detecting if a person is sitting on a toilet. Other sensors may be IR sensor, or a heat sensor. The concept includes a sensor 102 to detect a person in the vicinity of the toilet 100.

[0081] In step S12, the radar 106 transmits a radar signal into the toilet bowl 108. When transmitting the radar signal, radar echoes are created by reflection of the radar signal.

[0082] Optionally the radar 106 may transmit radar wavelets. In some examples the radar 106 transmits radar wavelets at a predetermined time interval.

[0083] The radar 106 may be configured to transmit radar wavelets with a carrier frequency corresponding to a millimeter wavelength. This implies that the carrier frequency may for instance be in a range of 10 - 300 GHz, such as in range of 30- 100 GHz. According to an example, a carrier frequency of approximately 60 GHz may be used. This may allow detecting a distance to the object with a resolution in millimeter range.

[0084] The radar 106 may further be configured to sample echoes at a number of locations in relation to the radar 106. The radar 106 may for instance be configured to sample echoes corresponding to discrete measuring points in relation to the radar 106. The measuring points may be defined in relation to integer number of wavelengths in relation to the carrier frequency, but it should be realized that the measuring points may be defined in another manner.

[0085] Using a 60 GHz carrier frequency, a distance between the measuring points may be set to correspond to a size of two wavelengths. Since the radar signal travels back and forth to the object, the distance between themeasuring points may thus be 5 mm. In an example, the radar 106 may be configured to detect echoes from a range of 40 - 300 mm from the radar 106. It should be realized that echoes in a range of 10 - 500 mm may be used instead and that the range may be set in relation to a design of the toilet bowl 108 and a position of the radar 106 in relation to the toilet bowl 108.

[0086] The use of measuring points being separated by a few mm, such as in a range of 1 - 10 mm, such as being separated by 5 mm, ensures that objects moving into the toilet bowl 108 will be detected. However, by having relatively few measuring points, such as by the distance between measuring points not being smaller than 1 mm, the amount of radar data generated is limited such that computer resources for generating, transmitting and processing of the radar data may be limited.

[0087] The radar 106 may be configured to acquire sequences of frames, wherein each frame aggregates echo information from objects across a specified number of measuring points. A frame may be constructed by performing multiple sweeps, during which data may be collected from the measuring points for different transmitted wavelets. A number of sweeps per frame may for instance be in a range of 1 - 64 and may be based on a predicted velocity of the objects. The pulse repetition frequency, i.e. , a rate at which wavelets are transmitted, may be in order of MHz, such as in a range of 1 - 20 MHz, such as approximately 20 MHz.

[0088] A frame rate may be set to a few frames per second, such as 5 - 20 frames per second, such as approximately 10 frames per second. The number of frames per second may be controlled to ensure that all objects moving through the toilet bowl 108 are detected, while ensuring a relatively low power consumption by not having too high frame rate.

[0089] In step S14, radar data is generated from the radar echoes created in the toilet bowl. The radar data may thus contain data with regards to masses dropping into the toilet at different times, and distances. The radar data may be provided as a time series in form of amplitude and / or phase information in a plurality of measuring points based on echoes of the transmitted radar signal. The radar data may represent location and velocity of the objectreflecting the radar signal and may also represent a texture of the object based on an amplitude of the radar signal.

[0090] In step S16 a processor is receiving the radar data. The processor 110 is in communication with the radar 106. The processor may be located on a server at a distance from the toilet 100. In some examples, the processor may be located at the site of the toilet.

[0091] In step S18, the processor is adapted to analyze the radar data. When analyzing the radar data, the radar data is analyzed based on event characteristics. The event characteristics may be characteristics associated with different excreta. Based on a correspondence between the radar data and the event characteristics the processor may be able to identify the toilet visit event. For example, the analysis may show that the excreta were feces based on the characteristics of the radar data. In some examples, the event may be determined to be a urine event.

[0092] The analysis may comprise analyzing the amount of excreta that has been passed through the toilet bowl 108. Meaning that a volume of urine and / or feces can be detected and determined through the analysis.

[0093] The system may determine the event in terms of ml, or cl or any other suitable unit.

[0094] The system may be adapted to determine a measurement of a volume unit of a mass being present in the toilet bowl 108.

[0095] Optionally the processor may be configured to access a set of predefined event characteristics for the analysis of the radar data.

[0096] Optionally, the method may comprise step S20. In step S20, the processor determines if a number of toilet visits have exceeded a predefined threshold value wherein the event has been identified as null or “no feces”. Upon determining that the number of toilet event identified as null are above the threshold value, an alarm is sent to a display interface.

[0097] An alarm may also be sent out upon detecting that the volume is exceeding a predetermined threshold. The predetermined threshold may be related to an indication of disease or risk factor for a patient levels of excreta.

[0098] The step may in other examples send out an alarm if the system detects that the same toilet is frequently used for a specific type of eventwhich could be connected to a disease. It should be noted that the skilled person would be able to set the alarm parameters as suitable depending on which health parameter is sought to be detected identified.

[0099] It should also be realized that event information generated by the system may be further provided to generate alarms or indications in other ways. For instance, the processor may be configured to identify a duration since a last feces event associated with a patient. If the duration exceeds a threshold, an alarm may be generated.

[0100] Turning to figure 3, another example of the system is shown. In the Figure, the toilet seat comprises a sensor that is visible from the top of the toilet seat, the sensor may be a light sensor for detecting a presence above the toilet seat.

[0101] Reference is now made to Fig. 4. A cross-sectional side view of a portion of a toilet is shown and a cross sectional side view of an example of aa radar.

[0102] The radar 106 is attached to the back top portion of the toilet bowl. The radar 106 is sized and shaped so that it fits under the toilet seat, and so that it is not in contact with the toilet lid 46 when the toilet lid 46 is closed.

[0103] The radar 106 comprises a radar chip 44 and a radar lens 42. The radar lens facing in a downwards direction into the toilet bowl 108.

[0104] The radar lens 42 may be directed downwards in a 45-degree angle A. In some examples the angle A may be between 30 - 60 degrees. In some examples the angle A may be between 30 - 40 degrees. In some examples the angle A may be between 40 - 50 degrees. In some examples the angle A may be between 50 - 60 degrees.

[0105] The radar set up is dependent on the dimensions of the toilet. The radar is further set up such that there are enough measuring points and high enough frame rate to capture all objects moving through the toilet. Preferably energy and power consumption is kept at a minimum while meeting the other requirements.

[0106] In one example the radar may be a radar for transmitting radar wavelets, such as a radar produced by Acconeer AB, headquartered in Malmo, Sweden.Fig. 5 shows a network of a plurality of toilets. Each toilet 100 has a radar 106 and a sensor 102 attached thereto. The group of toilets are connected to a server having the processor 110. The processor 110, which is represented in the cloud in the figure, is adapted to send a data signal to a display. The display may be a wearable device. The display may be a portable device such as a laptop, a mobile phone or a tablet.

[0107] Hence, the above-described system for identifying a toilet event may be connected into a network of toilets, and display device. For example, the system may be connected to a hospital care system, or any other people care facility for detecting irregularities in patients’ toilet visits. Thereby the system may be applied to improve patient safety during hospital care.

[0108] In some examples, the event characteristics used by the processor 110 to analyze the radar data are represented by a machine-learned model. In such examples, the processor 110 is configured to access the machine-learned model for identifying the event.

[0109] In the following, an example for generating the machine-learned model will be described.

[0110] Training of the machine-learned model may comprise acquiring training data. The training data may be acquired using a radar of a corresponding type that is used in the system for identification of the event. Thus, the same type of hardware is used for acquiring training data and later used for generating radar data for analysis.

[0111] The training data may thus be acquired using a radar as described above. The training data may include a time series in form of amplitude and / or phase information in a plurality of measuring points based on echoes of the transmitted radar signal.

[0112] The training data may be acquired during toilet visits performed by persons involved in a training project. The training data may alternatively be acquired by performing measurements on objects simulating events at toilet visits. In this regard, measurements may be performed by pouring water into the toilet bowl for representing a urine event and by dropping porridge into the toilet bowl for representing a feces event.The training data may further be annotated for providing ground-truth information of the training data. For instance, a person involved in the training project may indicate information relating to the toilet visit, such as indicating whether a feces event and / or a urine event is included in the training data related to the toilet visit. In addition, information of texture of feces, spanning from loose diarrhea to hard pellets, may be provided, estimation of quantity of feces may be provided and / or information of sex of the person may be provided.

[0113] The training data including measurements from a radar with associated ground-truth information may be provided for training a neural network to provide analysis of radar data. The machine-learned model formed by training the neural network may be configured to provide classification of a toilet visit. Thus, the processor may be configured to analyze radar data using the machine-learned model for classifying the toilet visit into one or more classes. The classes may be set based on the ground-truth information.

[0114] The ground-truth information may also include information relating to information in the training data that does not relate to excreta of a person. For instance, ground-truth information may be provided for indicating toilet paper being dropped into the toilet bowl. Thus, the machine-learned model may be trained to differentiate relevant events from events that are not of interest, such as events relating to toilet paper.

[0115] According to an embodiment, the processor may be configured to use the machine-learned model for classifying the toilet visit as comprising a feces event or not comprising a feces event. Thus, the processor may be configured to indicate whether or not a feces event occurred during the toilet visit.

[0116] According to another embodiment, the processor may be configured to use the machine-learned model for classifying the toilet visit as comprising one or more of a feces event or a urine event. Thus, the processor may be configured to indicate whether or not a feces event occurred during the toilet visit. Further, the processor may be configured to indicate whether or not a urine event occurred during the toilet visit.The processor may also be configured to use a part of the training data for verification. Thus, the machine-learned model may be trained using a first part of training data and, once trained, the machine-learned model may be verified using a second part of the training data.

[0117] According to an embodiment, a signal acquired by the radar may be pre-processed before being provided to the neural network for training the machine-learned model. The pre-processing may include performing a Fourier transform on the signal acquired by the radar. This implies that a time series of the signal acquired by the radar may be converted to frequency domain for forming the training data. This may facilitate identifying of events in data acquired by the radar. For instance, a fast Fourier transform (FFT) may be calculated on the signal acquired by the radar.

[0118] The pre-processing may further include subtracting a mean value from the signal acquired by the radar before performing Fourier transform. This may imply that the signal is feature scaled or normalized facilitating comparison of signals from different sessions.

[0119] The pre-processing may also include performing a windowing function on the signal acquired by the radar before performing Fourier transform. For instance, a Hann function may be applied. This may imply that spectral leakage is reduced. Spectral leakage may otherwise be introduced by analysis of a time duration of the signal.

[0120] According to an embodiment, the signal acquired by the radar may further, after being converted to frequency domain via Fourier transform, be pre-processed to determine a maximum magnitude at discrete frequencies. These frequencies, determined by factors such as sampling parameters and the number of sweeps per frame, correspond to different Doppler shifts of the signal caused by movement of the object(s). The frequencies thus correspond to different velocities of the object(s). The maximum magnitudes at each discrete frequency may be associated with a time window forming a set of information for the time window. A plurality of sets of information may be extracted for respective time windows during a toilet visit. Each set may be separately annotated in the ground-truth information. Alternatively, all of the time windows are associated with a common ground-truth information.According to an embodiment, a set of information comprising maximum magnitudes for discrete frequencies may be combined with groundtruth information for forming training data. Thus, the set of information associated with ground-truth information may be provided as input to a neural network for training the machine-learned model.

[0121] The neural network may comprise a plurality of layers for processing input data. For instance, the neural network may be formed as a convolutional neural network (CNN). The CNN may comprise two convolutional layers. The stacking of multiple convolutional layers helps to achieve automatic feature extraction for classification, where downstream layers capture more complex or differentiating features.

[0122] However, it should be realized that the neural network may be formed in a different manner. For instance, a recurrent neural network (RNN) may be used for analyzing time series data for inferring sequential / time-variant information, since an RNN incorporates contextual information from past inputs.

[0123] The CNN may for instance comprise three convolutional layers. The network may further comprise a pooling layer followed by a dense layer for reducing spatial dimensions of the input, controlling overfitting, and extracting dominant features.

[0124] The CNN may further comprise batch normalization layers and dropout layers between the CNN layers. This may further reduce overfitting of the machine-learned model.

[0125] The machine-learned model may be trained as set forth above to provide a high accuracy of identifying events at a toilet visit. For instance, the machine-learned model may be trained for binary classification of a feces event, i.e. , to differentiate between a feces event and a not feces event, with a 95% accuracy.

[0126] The machine-learned model trained as described above may further be deployed in the system for identification of an event at a toilet visit, as described above.

[0127] The processor may thus be configured to analyze radar data using the machine-learned model. The processor may be configured to initially pre-process the radar data in a corresponding manner as described above for generating suitable data to be input to the machine-learned model for identifying the event.Thus, the processor may be configured to pre-process the radar data. According to an embodiment, the processor may be configured to perform a Fourier transform on the radar data, such as performing a FFT on the radar data.

[0128] According to an embodiment, the processor may be configured to determine a maximum magnitude at discrete frequencies in a range of frequencies. Thus, a set of information may be formed. The set of information may be input to the machine-learned model and the machine-learned model may output a classification of the radar data, such as indicating whether a feces event and / or a urine event occurred during the toilet visit.

[0129] It will be appreciated that the present invention is not limited to the embodiments shown. Several modifications and variations are thus conceivable within the scope of the invention which thus is exclusively defined by the appended claims.

Claims

CLAIMS1. A system for identification of an event at a toilet visit, said system comprising:a sensor configured to be arranged on a toilet, and configured to detect a presence of a person making the toilet visit;a radar configured to be arranged on the toilet, and further configured to:transmit a radar signal to the toilet bowl and detect echoes caused by the radar signal andgenerate radar data based on the echoes;a processor configured to:receive the radar data, andanalyze radar data based on event characteristics to identify the event based on correspondence between the radar data and at least one of the event characteristics, wherein the event characteristics comprises at least event characteristics of a feces event.

2. The system of claim 1 , further comprising a storing unit configured to store the event characteristics, and wherein the processor is configured to access said storing unit.

3. The system of claim 1 or 2, wherein the radar is configured to be in wireless communication with the processor.

4. The system of any one of the preceding claims wherein the sensor is configured to be in wireless communication with the processor.

5. The system of any one of the preceding claims, wherein the sensor is a pressure sensor.

6. The system of any one of the preceding claims, wherein the radar is configured to transmit radar wavelets.

7. The system of any one of the preceding claims, wherein the event characteristics are represented by a machine-learned model and wherein the processor is configured to have access to the machine-learned model for identifying the event.

8. The system of any one of the preceding claims, wherein the event characteristics comprises characteristics of a urine event.

9. The system of any one of the preceding claims, wherein the event characteristics comprises characteristics for determining a volume of the event.

10. A method for identification of an event at a toilet visit, said method comprising:detecting a presence of a person making the toilet visit;transmit, by a radar, a radar signal to the toilet bowl and detecting echoes caused by the radar signal andgenerating radar data based on the echoes;receiving, by a processor, the radar data, andanalyzing radar data based on a set of event characteristics to identify the event based on correspondence between the radar data and at least one of the event characteristics, wherein the set of event characteristics comprises at least event characteristics of a feces event.

Images