Wearable sensor system

Abstract

A biomagnetic sensor system comprising one or more sensor units, the, or each, sensor unit including one or more sensors. The biomagnetic sensor system further comprises a processing module. The, or each, sensor is configured to sense a magnetic field. The, or each, sensor unit is configured to be locatable on the body of the subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject. The processing module is configured to process the one or more signals from the one or more sensors to determine (i) a magnetocardiographic (MCG) signal of the subject; and (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

Description

[0001] Wearable Sensor System

[0002] Field of the invention

[0003] The invention relates to systems, devices and methods for biomagnetic sensing, particularly for use in wearable cardiac health monitoring.

[0004] Background to the invention

[0005] Magnetocardiography (MCG) offers a promising alternative to electrocardiography (ECG) as a diagnostic tool in cardiac healthcare.

[0006] Different to ECG, which relies on detecting electrical signals generated by the heart, MCG measures the magnetic fields produced by the heart's electrical activity.

[0007] Widespread adoption of MCG still faces barriers. One significant obstacle associated with existing MCG systems, is that the equipment required is typically bulky and expensive and predominantly found in hospital settings. Furthermore, existing MCG techniques typically require the MCG sensor to be physically separated from the patient in order for a clear MCG signal to be obtained.

[0008] Unlike ECG, which has become increasingly accessible, and is even integrated into small wearable devices like smart watches, the typically high cost of MCG sensing and the specialised equipment required to successfully perform MCG measurements limits its availability to specialised medical facilities. This barrier has impeded its utilisation in community healthcare settings and underserved areas, where affordability and accessibility are crucial factors in providing comprehensive cardiac care. Therefore, while MCG holds immense potential for revolutionising cardiac diagnostics, existing hospital-based techniques are bulky, expensive and often require controlled environments with a physical separation between the sensor and the patient.

[0009] The present invention aims to address at least some of these issues.

[0010] Summary of the invention

[0011] According to a first aspect, the invention provides a wearable biomagnetic sensor system comprising: one or more sensor units; the, or each, sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; and wherein the, or each, sensor unit is configured to be locatable on the body of the subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and wherein the processing module is configured to process the one or more signals from the one or more sensors to determine:

[0012] (i) a magnetocardiographic (MCG) signal of the subject; and

[0013] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0014] According to a second aspect, the invention provides a wearable biomagnetic sensor device comprising: one or more sensor units, the, or each, sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; and wherein the, or each, sensor unit is configured to be locatable on the body of a subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and wherein the processing module is configured to process the one or more signals from the one or more sensors to determine:

[0015] (i) a magnetocardiographic (MCG) signal of the subject; and

[0016] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0017] According to a third aspect, the invention provides a method for cardiac monitoring of a subject using a wearable biomagnetic sensor system, wherein the wearable biomagnetic sensor system comprises: one or more sensor units; the, or each sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; wherein the, or each, sensor unit is configured to be locatable on the body of the subject; the method comprising: locating the, or each, sensor unit on the body of a subject; outputting from the, or each sensor, a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and processing, using the processing module, the one or more signals from the one or more sensors to determine:

[0018] (i) a magnetocardiographic (MCG) signal of the subject; and (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0019] According to a fourth aspect, the invention provides a method for cardiac monitoring of a subject using a wearable biomagnetic sensor device, wherein the wearable biomagnetic sensor device comprises: one or more sensor units; the, or each sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; wherein the, or each, sensor unit is configured to be locatable on the body of the subject; the method comprising: locating the, or each, sensor unit on the body of a subject; outputting from the, or each sensor, a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and processing, using the processing module, the one or more signals from the one or more sensors to determine:

[0020] (i) a magnetocardiographic (MCG) signal of the subject; and

[0021] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0022] According to a fifth aspect, the invention provides a method comprising: receiving, at a processing module, data relating to one or more signals; the, or each, signal being indicative of a sensed magnetic field at or adjacent to a location on the body of a subject; the processing module processing the data to determine:

[0023] (i) a magnetocardiographic (MCG) signal of the subject; and (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0024] According to a sixth aspect, the invention provides a biomagnetic sensor system comprising: one or more sensor units; the, or each, sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; and wherein the, or each, sensor unit is configured to be locatable on the body of the subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and wherein the processing module is configured to process the one or more signals from the one or more sensors to determine:

[0025] (i) a magnetocardiographic (MCG) signal of the subject; and

[0026] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0027] According to a seventh aspect, the invention provides a biomagnetic sensor device comprising: one or more sensor units, the, or each, sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; and wherein the, or each, sensor unit is configured to be locatable on the body of a subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and wherein the processing module is configured to process the one or more signals from the one or more sensors to determine:

[0028] (i) a magnetocardiographic (MCG) signal of the subject; and

[0029] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0030] According to an eighth aspect, the invention provides a method for cardiac monitoring of a subject using a biomagnetic sensor system, wherein the biomagnetic sensor system comprises: one or more sensor units; the, or each sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; wherein the, or each, sensor unit is configured to be locatable on the body of a subject; the method comprising: locating the, or each, sensor unit on the body of a subject; outputting from the, or each sensor, a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and processing, using the processing module, the one or more signals from the one or more sensors to determine:

[0031] (i) a magnetocardiographic (MCG) signal of the subject; and

[0032] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject. According to a ninth aspect, the invention provides a method for cardiac monitoring of a subject using a biomagnetic sensor device, wherein the biomagnetic sensor device comprises: one or more sensor units; the, or each sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; wherein the, or each, sensor unit is configured to be locatable on the body of the subject; the method comprising: locating the, or each, sensor unit on the body of the subject; outputting from the, or each sensor, a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and processing, using the processing module, the one or more signals from the one or more sensors to determine:

[0033] (i) a magnetocardiographic (MCG) signal of the subject; and

[0034] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0035] According to a tenth aspect, the invention provides a method comprising: receiving one or more signals; the, or each, signal being indicative of a sensed magnetic field at or adjacent to a (e.g. respective) location on a body of a subject; inputting data relating to the one or more signals to a trained machine learning model, wherein the trained machine learning model is trained to determine, from data relating to the one or more signals:

[0036] - a magnetocardiographic (MCG) signal of the subject; and

[0037] - a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject; and using the trained machine learning model to determine:

[0038] (i) a magnetocardiographic (MCG) signal of the subject; and

[0039] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0040] According to an eleventh aspect, the invention provides a method comprising: inputting, to a trained machine learning model, data relating to one or more signals indicative of a sensed magnetic field at or adjacent to a (e.g. respective) location on a body of a subject, wherein the trained machine learning model is trained to determine, from data relating to the one or more signals:

[0041] - a magnetocardiographic (MCG) signal of the subject; and

[0042] - a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject; and using the trained machine learning model to determine:

[0043] (i) a magnetocardiographic (MCG) signal of the subject; and

[0044] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0045] The trained machine learning model of the tenth and eleventh aspects may comprise a (e.g. first) artificial neural network. The trained machine learning model of the tenth and eleventh aspects may comprise a first artificial neural network and a second artificial neural network.

[0046] According to a twelfth aspect, the invention provides a system for cardiac health monitoring of a subject comprising: one or more sensor units; the, or each, sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; and wherein the, or each, sensor unit is configured to be locatable on the body of the subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and wherein the processing module is configured to process the one or more signals from the one or more sensors to determine:

[0047] (i) a magnetocardiographic (MCG) signal of the subject; and

[0048] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0049] The invention extends to software (and to a (non-transitory) computer- readable storage medium bearing the same) comprising instructions that, when executed by a processing system, cause the processing system to: perform any of the methods disclosed herein.

[0050] The invention also extends to a processing system configured to perform any of the methods disclosed herein. Steps disclosed herein may be carried out by hardware (e.g. ASICs or FPGAs or other circuitry) or by software or by a combination of hardware and software. The processing system may comprise one or more processors and a memory storing software for execution by the one or more processors.

[0051] Embodiments of the first to eleventh aspects of the present invention may include one or more features of one or more of the other aspects of the present invention or its embodiments.

[0052] As the, or each, sensor unit is configured to be locatable on the body of the subject, signals indicative of the sensed magnetic field will be influenced by mechanical effects. For example, in embodiments where the, or each, sensor unit is placed on a chest portion of the subject both electrical activity (i.e. described by MCG signals) and mechanical activity (i.e. described by SCG and / or BCG signals) of the heart may be detected by the sensors.

[0053] As will be appreciated by those skilled in the art, seismocardiographic (SCG) signals are caused by the local vibrations of the chest wall in response to the heartbeat and ballistocardiographic (BCG) signals represent ballistic forces generated by the heart.

[0054] Thus, it will be appreciated that, embodiments of the invention may provide an improved approach to biomagnetic sensing by determining information relating to the electrical cardiac activity and mechanical cardiac activity of the subject. While MCG signals describe the electrical activity of the heart by describing the magnetic fields generated by said electrical activity, SCG and BCG signals each relate to the mechanical activity of the heart.

[0055] The Applicant has found that MCG signals, SCG and / or BCG signals of a subject may be determined using signal processing when the, or each, magnetic sensor is locatable on the body of a subject. Therefore, using embodiments of the invention, it is possible to obtain MCG, SCG and BCG information using one type of sensor. Previously, to obtain readings of mechanical activity and electrical activity at the same time, it was necessary to combine ECG sensing with mechanical sensors such as accelerometers, gyroscopes, and piezoelectric sensors. For the first time, using the invention both mechanical and electrical activity may be detected using a single type of sensor (i.e. a magnetic sensor). Previous techniques for measuring MCG have treated this mechanical activity of the body to be a source of undesirable noise. Therefore, existing techniques focus on introducing a physical separation between MCG sensors and the subject being monitored. Such a physical separation between the sensor and the subject means that such sensing systems cannot be used in a portable or wearable manner without clinical supervision as careful positioning of the subject would be required to remove the noise caused by mechanical effects. The applicant has, therefore, found a way to determine MCG signals of a subject without requiring a gap or physical spacing between the subject and the sensor unit(s).

[0056] Unlike ECG, which relies on detecting electrical signals generated by the heart, MCG measures the magnetic fields produced by the heart's electrical activity. This fundamental difference grants MCG several key benefits. Firstly, MCG provides more accurate and detailed information about cardiac function. The magnetic fields captured by MCG are less distorted by surrounding tissues and offer clearer signals, allowing for better spatial resolution and detection of subtle abnormalities. This enhanced precision may enable early detection of cardiac pathologies, leading to timely intervention and improved patient outcomes. For instance, MCG can detect early signs of coronary artery disease (CAD) and heart irregularities, because it is more sensitive to cardiac tangential and vortex currents. Therefore, MCG offers better insights into depolarisation and repolarisation directions than ECG. Furthermore, MCG is non-invasive and radiation-free, posing minimal risk to patients, including those with pacemakers or metallic implants. Its safety profile makes it suitable for repeated use in monitoring and longitudinal studies without concerns about cumulative radiation exposure. Additionally, the ability to use MCG to detect magnetic fields through clothing and tissues allows for greater comfort of the subject and convenience during examinations. MCG is also particularly beneficial for paediatric and geriatric populations who may struggle with electrode placement or require sedation for ECG procedures.

[0057] Diverse applications are envisaged for the invention. For example, embodiments of the invention may be used for cardiac health monitoring systems and methods. Having a cardiac monitoring system that is capable of being worn (i.e. by providing sensor units configured to be locatable on the body of the subject) and that can provide continuous determination of MCG data (i.e. MCG signals) may allow for longer term (e.g. 24 / 7) monitoring compared to existing hospital-based techniques. Traditional non-contact techniques, where the subject and the sensing equipment must be physically spaced apart, can make longer term testing difficult or impossible. Increasing the time spent monitoring a patient can increase the number of cardiac conditions that can be detected including some conditions that would be undetectable with shorter term testing. While embodiments of the invention may be used for everyday cardiac health monitoring, embodiments of the invention may also be used in intensive care and maternal and foetal cardiac monitoring before and during labour.

[0058] In a set of embodiments, the (e.g. each of the) one or more signals representative of a sensed magnetic field are influenced by electrical activity and mechanical activity of the heart of the subject. The sensed magnetic field may be a magnetic field generated by electrical activity of the heart of the subject . Measurements of the sensed magnetic field may be influenced by mechanical activity of the heart of the subject.

[0059] In a set of embodiments, the sensed magnetic field at or adjacent to a (e.g. respective) location on the body of the subject is a sensed magnetic field adjacent to biological tissue of the subject. The sensed magnetic field may be a sensed biomagnetic field.

[0060] Each signal indicative of a sensed magnetic field may be indicative of a sensed magnetic field at or adjacent to a location on the body of the subject corresponding to the location of the corresponding sensor or sensor unit.

[0061] The term “biomagnetic” is used to refer to magnetic fields that are generated through electric activity in biological tissue, e.g. heart tissue.

[0062] It will be appreciated that the sensed magnetic field may comprise a background component (i.e. due to ambient magnetic noise), which may be removed using known methods, such as those disclosed in WO 2024 / 062068 A1 . For example, a gradiometer unit may output a background signal indicative of ambient magnetic noise; and an active noise cancellation unit may remove the background signal from one or more (e.g. each) of the signals to generate a background-subtracted biomagnetic signal.

[0063] The biomagnetic sensor system and / or biomagnetic sensor device may be wearable (i.e. capable of being worn by a subject).

[0064] It should be appreciated by the skilled person that a wearable biomagnetic sensor system may involve at least one wearable component and does not require every feature of the biomagnetic sensor system to be wearable. In a set of embodiments, the biomagnetic sensor system may comprise one or more wearable sensor units and the signal processing module may be provided by a separate non-wearable apparatus. The biomagnetic sensor system and / or the biomagnetic sensor device may be portable (i.e. capable of being carried). In a set of embodiments, the biomagnetic sensor system and / or device may comprise one or more portable sensor units and the processing module may be provided by a separate non-portable apparatus.

[0065] The entire biomagnetic sensor system and / or device may be portable. The entire biomagnetic sensor system and / or device may be wearable.

[0066] However, in a set of embodiments, the, or each, sensor unit is a portable and / or wearable sensor unit.

[0067] The, or each, sensor unit may be configured to in use contact an external surface of the body of the subject. The, or each, sensor may be positioned within the, or each, sensor unit so that the, or each, sensor in use contacts an external surface of the body of the subject.

[0068] The, or each, sensor unit may be attachable (e.g. in use attached) to the body (e.g. to a body portion) of the subject. Therefore, the biomagnetic sensor device and / or system may be attachable to the body (e.g. to a body portion) of the subject.

[0069] The, or each, sensor unit; biomagnetic sensor device and / or system may be attachable (e.g. in use attached) to (e.g. a torso (e.g. chest) portion of) the body of the subject, e.g. via one or more securing elements (e.g. straps or other suitable means). Therefore, the biomagnetic sensor device and / or system may comprise one or more securing elements (e.g. one or more straps) arranged to attach the, or each, sensor unit to (e.g. a torso (e.g. chest) portion of) the body of the subject. Each securing element may be a strap. The, or each, sensor unit may be arranged in layers and may comprise a plurality of layers; wherein at least one of the layers is a sensor layer. There may also be a motherboard layer and / or a battery layer. The sensor layer may comprise a substrate and one or more sensors. For example, the sensor layer may comprise a single sensor or an array of sensors. The plurality of layers may be arranged so that in use the sensor layer is the layer most proximal to the subject.

[0070] The, or each, sensor unit may be arranged within a housing. Each sensor unit preferably has a respective housing. However, the plurality of sensor units may be arranged within a common housing. The housing may be a plastic housing. The housing may comprise an opening (e.g. an aperture or window) proximal to the one or more sensors in the, or each, sensor unit. The opening may be in use positioned between the, or each, sensor of each sensor unit and the body of the subject. In a set of embodiments, the system is arranged so that in use the, or each, opening is positioned between the, or each, sensor of each sensor unit and the body of the subject.

[0071] The, or each, housing may have a lid portion and a base portion. The, or each, sensor of the sensor unit may be adjacent to the base portion of the, or each, housing. The lid portion and the base portion may be connectable. The base portion of the, or each, housing may be arranged to in use contact the body of the subject. The base portion of the, or each, housing may be arranged to be in use placed (e.g. held or pressed) against the body of the subject. The base portion of the, or each, housing may comprise a contact surface for contacting the body of the subject. The opening of the, or each, housing may be an opening in the contact surface of the base portion of the, or each, housing. The housing of the, or each, sensor unit may comprise one or more attachment portions for attaching a securing element (e.g. a strap) thereto. The housing may have one or more side walls. The one or more attachment portions (e.g. each attachment portion) may be located on one or more side walls (e.g. each side wall) of the housing, e.g. positioned proximal to the base portion. Each attachment portion may comprise an aperture sized and shaped for a securing element to pass therethrough.

[0072] The, or each, attachment portion may allow the sensor unit to be worn by the subject. The attachment portions may be used to attach the, or each, sensor unit to one or more further sensor units - e.g. to build a multichannel biomagnetic sensor device and / or system, wherein each sensor or sensor unit outputs a respective signal on a respective measurement channel.

[0073] Preferably, the, or each, sensor unit is a sensor unit configured to sense weak biomagnetic fields. In a set of embodiments, the, or each, sensor unit is a sensor unit configured to sense magnetic fields having a magnetic field strength below 100 pT, e.g. below 10 pT.

[0074] The, or each, sensor may be an optically pumped magnetometer (OPM).

[0075] The, or each, sensor may be a superconducting quantum interference device (SQUID).

[0076] The, or each, sensor may be a nitrogen-vacancy (NV) diamond sensor.

[0077] The, or each sensor unit may be a magnetoresistance (MR) sensor unit (i.e. each sensor being a magnetoresistance (MR) sensor). In a set of embodiments, the, or each, sensor is a tunnel magnetoresistance, TMR, sensor and the, or each, sensor unit is a TMR sensor unit.

[0078] The applicant has found that TMR sensors are well-suited for wearable systems and devices because of the small sensor footprint (e.g. the sensor package may be as small as 1 .5 mm by 1 .5 mm) that can be achieved without compromising on sensitivity.

[0079] When TMR sensors are used, cardiac monitoring may be performed in ambient environments without a magnetically shielded room and sensor heating or cooling requirements. The ability to incorporate magnetocardiography into a wearable device and / or system which can be comfortably worn under clothing, like a t-shirt, opens possibilities for continuous, non-invasive monitoring of cardiac activity in various settings, including everyday life and clinical environments. This could lead to earlier detection of cardiac abnormalities and more personalized healthcare solutions.

[0080] The, or each, TMR sensor unit may comprise a plurality of TMR sensors, each of which may have an array of magnetic tunnelling junctions fabricated on a substrate. The magnetic tunnelling junctions may be fabricated using known nanoscale techniques so that the array has a small footprint. The, or each, TMR sensor unit may comprise a TMR sensor comprising four TMR sensor elements in a Wheatstone bridge arrangement. The, or each, TMR sensor unit may comprise one or more of such Wheatstone bridge arrangements.

[0081] The, or each, sensor or sensor unit may (e.g. be configured to) output a signal on a respective measurement channel. The signal on each measurement channel may be a signal derived from the signal indicative of a sensed magnetic field (e.g. it may be a processed (e.g. background subtracted) version of the signal output from the one or more sensors). For example, a signal output from a sensor unit on a measurement channel may be a signal derived from the one or more signals indicative of a sensed magnetic field output from the one or more sensors therein. The system and / or device may use a single measurement channel or a plurality of measurement channels (e.g. the system and / or device may be a multi-channel system). The system and / or device may use at least three (e.g. from four to six) measurement channels. In a set of embodiments, the system and / or device comprises at least three sensor units and at least three measurement channels. The method may, therefore, comprise outputting signals (e.g. derived from the one or more signals indicative of a sensed magnetic field) from at least three sensor units on at least three corresponding measurement channels.

[0082] In a set of embodiments, the one or more sensor units is a plurality of sensor units, each sensor unit being configured to output a signal on a respective measurement channel derived from the signals indicative of a sensed magnetic field output from the one or more sensors therein.

[0083] In a set of embodiments, the biomagnetic sensor system and / or device is a multi-channel biomagnetic sensor system and / or device. The biomagnetic sensor system and / or device may be configured to output a plurality of (e.g. biomagnetic) signals to a respective plurality of measurement channels. Each of the sensor units may be arranged to detect a magnetic field adjacent to biological tissue and output a biomagnetic signal on a respective measurement channel. In embodiments comprising a plurality of measurement channels, signals on the plurality of measurement channels may be selectively coupled (e.g. using a multiplexer) to readout circuitry and subsequently processed by the processing module to determine: (i) the MCG signal of the subject; and (ii) the SCG signal of the subject and / or the BCG signal of the subject. Data representative of the one or more signals may be provided to the trained machine learning model (i.e. the first artificial neural network).

[0084] In a set of embodiments, there are at least three measurement channels. Each measurement channel may be provided by an output of a sensor or a sensor unit. Each sensor unit may comprise a plurality of sensors in an array.

[0085] The biomagnetic sensor system and / or device may comprise a plurality of sensors. The biomagnetic sensor system and / or device may comprise a plurality of sensor units. The, or each, sensor unit may comprise a respective plurality of sensors.

[0086] At least one (e.g. each) of the one or more sensor units may be (e.g. configured to be) locatable on a torso (e.g. chest) portion of the body of the subject, such that the, or each, sensor in said sensor unit in use outputs a signal indicative of a sensed magnetic field at or adjacent to the torso (e.g. chest) portion on the body of the subject. The method may comprise locating at least one (e.g. each) of the one or more sensor units on a torso (e.g. chest) portion of the body of the subject, such that the, or each, sensor in said sensor unit in use outputs a signal indicative of a sensed magnetic field at or adjacent to the torso (e.g. chest) portion on the body of the subject. The, or each, sensor unit may be configured to be attachable to a torso portion of the body of the subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to the torso portion on the body of the subject. The method may comprise attaching at least one (e.g. each) of the one or more sensor units on a torso (e.g. chest) portion of the body of the subject, such that the, or each, sensor in said sensor unit in use outputs a signal indicative of a sensed magnetic field at or adjacent to the torso (e.g. chest) portion on the body of the subject.

[0087] In a set of embodiments, the biomagnetic sensor system and / or device comprises a plurality of sensor units, wherein the plurality of sensor units is configured so that they are each locatable at different positions on the body of the subject. Each sensor unit of the plurality of sensor units may be configured to be locatable on a torso (e.g. chest) portion of the body of the subject. Therefore, the plurality of sensor units may be arranged so that they are in use contacting the torso (e.g. chest) portion of the body of the subject.

[0088] The biomagnetic sensor device and / or system may comprise a plurality of sensor units, e.g. each sensor unit being arranged to detect a magnetic field adjacent to biological tissue and output a biomagnetic signal on a respective channel. The plurality of sensor units may be arranged to contact respectively different locations on the body of the subject. The method may comprise placing each of the one or more sensor units at respectively different locations on the body of the subject (e.g. and attaching the one or more sensor units thereto). For example, each sensor unit may be arranged with respect to each other to in use contact a different area of the chest portion of the body. This may allow measurements to be taken at a plurality of locations on the chest portion of the body.

[0089] The processing module may receive a plurality of signals each signal being indicative of a respective sensed magnetic field, e.g. at a corresponding plurality of locations on the body of the subject. The plurality of signals may be acquired by a respective plurality of sensors or sensor units.

[0090] Having a multi-channel system and / or device may allow for the simultaneous recording of magnetic signals from multiple locations on the body of the subject, providing a more comprehensive view of cardiac activity. This may help to accurately capture the complex spatial and temporal dynamics of the magnetic field generated by the heart. When placed at multiple locations on the torso (e.g. chest), the multichannel system and / or device therefore may capture data from different regions of the torso (e.g. chest) simultaneously, enhancing the resolution and reliability of the measurements.

[0091] Where there is a plurality of sensor units, the sensor units may be connected in sequence (e.g. daisy-chained). Each of the biomagnetic sensor units may connect to a common processing unit (e.g. the processing module may be shared by the plurality of sensor units). Each of the sensor units may output a respective signal indicative of a respective sensed magnetic field to the common processing unit. The common processing unit may be where the MCG, SCG and / or BCG signals are determined. The plurality of sensor units may be arranged azimuthally around the common processing unit (e.g. in a star configuration). In a set of embodiments, the one or more sensor units is a plurality of sensor units and each of the biomagnetic sensor units are connected to a common processing unit; and each sensor unit of the plurality of sensor units is configured to output a respective signal indicative of a respective sensed magnetic field to the common processing unit.

[0092] For embodiments of the biomagnetic sensor system and / or device comprising a plurality of sensor units, each of the sensor units may comprise a respective battery or a single battery may supply power to the plurality of sensor units. For example, the common processing unit may comprise a battery for supplying power to the plurality of sensor units.

[0093] Some processing steps, e.g. digitisation of the one or more signals, make take place inside each sensor unit and some may take place centrally in the common processing unit.

[0094] The system and / or device may comprise a power management module. The power management module may be configured to efficiently regulate power consumption and ensure prolonged operation of the device and system, by optimising power usage across components (e.g., sensors; signal conditioning modules; signal digitisation modules, digital signal processing modules etc.) of the biomagnetic sensor system and / or biomagnetic sensor device.

[0095] The power management module may employ techniques such as power gating, voltage scaling, dynamic voltage and frequency scaling (DVFS), and duty cycling to adaptively control power consumption based on the operational requirements and constraints of the system and / or device.

[0096] In a set of embodiments, the system comprises a display arranged to output information relating to the MCG, SCG and / or BCG signals. For example, the display may output a visual representation of the MCG, SCG and / or BCG signals - e.g. to a user of the system.

[0097] The term ‘subject’ as used herein includes any human or non-human animal subject, including any human or non-human mammal, bird, fish, reptile, amphibian etc. However, in preferred embodiments the subject is a human.

[0098] Where reference is made to the body of the subject it should be understood that either direct contact or indirect contact may be made with the body. In a set of embodiments, the, or each, sensor unit is configured so that the, or each, sensor is configured to, in use, contact the body of a subject either: (i) directly (e.g. by contacting a skin surface of the body of the user); or (ii) indirectly by contacting the body of the user via a layer of clothing. The layer of clothing may be up to 1 cm, e.g. up to 1 mm, in thickness.

[0099] Preferably, there is no physical spacing between the, or each, sensor and the body of the subject (e.g. preferably, there is no air gap).

[0100] In a preferred set of embodiments, the, or each, sensor unit is configured to be locatable on the body of the subject, such that the, or each, sensor in use is no further than 1 cm, e.g. 1 mm, from an external surface (e.g. the skin) of the body of the subject. Ensuring that the, or each, sensor is close to the body of the subject allows the mechanical activity of the body (e.g. the heart) to be detectable via the one or more signals indicative of the sensed magnetic field which improves the determination of SCG and BCG signals. As mentioned above, the processing module processes the one or more signals from the one or more sensors to determine:

[0101] (i) a magnetocardiographic (MCG) signal of the subject; and

[0102] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0103] The processing module may, therefore, only determine:

[0104] (i) a magnetocardiographic (MCG) signal of the subject; and

[0105] (ii) a seismocardiographic (SCG) signal of the subject.

[0106] Alternatively, the processing module may only determine:

[0107] (i) a magnetocardiographic (MCG) signal of the subject; and

[0108] (ii) a ballistocardiographic (BCG) signal of the subject.

[0109] In a set of embodiments, the processing module is configured to process the one or more signals from the one or more sensors to determine:

[0110] (i) a magnetocardiographic (MCG) signal of the subject;

[0111] (ii) a seismocardiographic (SCG) signal of the subject; and

[0112] (iii) a ballistocardiographic (BCG) signal of the subject.

[0113] The processing module may be a signal processing module.

[0114] The MCG signal, SCG signal and / or BCG signal may be estimated signals.

[0115] In a set of embodiments, a first artificial neural network is used to generate (e.g. determine or estimate):

[0116] (i) data relating to the MCG signal of the subject; and (ii) data relating to the SCG signal of the subject and / or data relating to the BCG signal of the subject.

[0117] The processing module may process the data generated by the first artificial neural network to determine:

[0118] (i) the magnetocardiographic (MCG) signal of the subject; and

[0119] (ii) the seismocardiographic (SCG) signal of the subject and / or the ballistocardiographic (BCG) signal of the subject.

[0120] A second artificial neural network may be used to classify the data output from the first artificial neural network (e.g. as MCG, SCG or BCG data).

[0121] The first artificial neural network may be continuously trained by a training module by using sensor data from the one or more sensors as training data. Said training data may be acquired by the training module when the device and or system is in operation (e.g. during start up of the device and / or system).

[0122] The processing step may, therefore, comprise an estimation step (e.g. using the first artificial neural network) and a classification step (e.g. using the second artificial neural network). The classification step may prevent inaccurate attribution of the source of the estimated signal data.

[0123] The second artificial neural network may be trained to classify input data as MCG, SCG and / or BCG signal data. The second artificial neural network may be trained using MCG, SCG and BCG data (e.g. signal traces) as training data. The second artificial neural network may be a linear classifier, e.g. using linear discriminant analysis. The second artificial neural network may be a temporal convolutional network (TCN). The processing step may comprise performing (e.g. the first artificial neural network may be trained to perform) a method of constrained independent component analysis on the one or more signals to estimate (e.g. determine):

[0124] (i) the magnetocardiographic (MCG) signal of the subject; and

[0125] (ii) the seismocardiographic (SCG) signal of the subject and / or the ballistocardiographic (BCG) signal of the subject.

[0126] The method of constrained independent component analysis may be a method of constrained FastICA.

[0127] The method of constrained independent component analysis may comprise using an auxiliary measure (e.g. JauX1) for each of the MCG signal, SCG signal and / or BCG signal in a corresponding compound cost function (e.g. J = Jmain+ a JauX1) and minimising said compound cost function to recover a scaled and permuted estimation of each of the MCG signal, SCG signal and / or BCG signal.

[0128] In a set of embodiments, the method of constrained independent component analysis comprises: using a first auxiliary measure and a corresponding first compound cost function = Jmain+ a JauX1) for determining the MCG signal; using a second auxiliary measure and a corresponding second compound cost function (e.g. J2= Jmain+ a JauX2) for determining the SCG signal; and / or using a third auxiliary measure and a corresponding third compound cost function (e.g. / 3= Jmain+ a JauX3) for determining the BCG signal.

[0129] The method may comprise: minimising the first compound cost function to determine the MCG signal; minimising the second compound cost function to determine the SCG signal; and / or minimising the third compound cost function to determine the BCG signal.

[0130] Minimising each compound cost function may recover a scaled and permuted estimation of each of the MCG signal, the SCG signal and / or the BCG signal. Each scaled and permuted estimation of each of the MCG signal, the SCG signal and / or the BCG signal may be classified by the second artificial neural network to determine the MCG signal of the subject, the SCG signal of the subject and / or the BCG signal of the subject.

[0131] Each auxiliary measure may be determined based on one or more (known) spatial characteristics and / or temporal characteristics of each signal type (e.g. the MCG signal type, SCG signal type and / or BCG signal type). The spatial characteristics and / or temporal characteristics may be signatures of each target source - e.g. MCG, SCG and / or BCG each being respective target sources. The spatial characteristics may be based on the spatial distribution of the target source with respect to the body of the subject; while temporal characteristics may be based on particular temporal features of signal traces of the target sources.

[0132] Each auxiliary measure may be determined using a method of factor analysis (e.g. parallel factor analysis (PARAFAC)). Factor analysis may be used to determine the spatial and / or temporal characteristics of MCG signals, SCG signals and / or BCG signals. Another method for determining each auxiliary measure may involve using change point detection on the sensed signals. Such methods may involve use of the Teager-Kaiser Energy Operator tuned to separating MCG, SCG and BCG signals.

[0133] In a set of embodiments, the processing module comprises a digital signal processing module. The digital signal processing module may be configured to perform the methods described herein to determine:

[0134] (i) a magnetocardiographic (MCG) signal of the subject; and

[0135] (ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

[0136] Using a digital signal processing module (e.g. a digital signal processor) instead of other hardware architectures may be preferred for its comparable low cost and low power consumption.

[0137] Features of any aspect or embodiment described herein may, wherever appropriate, be applied to any other aspect or embodiment described herein. Where reference is made to different embodiments or sets of embodiments, it should be understood that these are not necessarily distinct but may overlap.

[0138] Brief description of the drawings

[0139] Embodiments of the invention will not be described, by way of example, with reference to the drawings, in which:

[0140] Fig. 1 is a schematic of a system in accordance with embodiments of the present invention;

[0141] Fig. 2 is an exploded view of a portable biomagnetic sensor device in accordance with embodiments of the invention;

[0142] Fig. 3 is an illustration showing the portable biomagnetic sensor device of Fig. 2, in use, attached to a subject;

[0143] Fig. 4 is a schematic diagram showing prior art methods of obtaining ECG and SCG information;

[0144] Fig. 5 is a schematic diagram showing how MCG; SCG and BCG signals can be obtained in accordance with embodiments of the invention;

[0145] Fig. 6 is a schematic diagram showing how MCG, SCG and BCG signals can be obtained using a machine learning model in accordance with embodiments of the invention; and

[0146] Fig. 7 shows preliminary experimental results for determining MCG, SCG and BCG traces from simulated sensed data.

[0147] Description of preferred embodiments

[0148] Fig. 1 shows a biomagnetic sensor system 1 for cardiac monitoring of a subject. The biomagnetic sensor system 1 achieves wearable magnetocardiography using sensitive magnetic sensors 4 and processes the sensed signals to obtain magnetocardiographic (MCG), seismocardiographic (SCG) and ballistocardiographic (BCG) signals.

[0149] The biomagnetic sensor system 1 may be implemented in a biomagnetic sensor device including one or more sensor units 2 locatable on the body of a (e.g. a human or animal) subject and has a processing module 20 for processing the sensed signals from the magnetic sensors 4 in the sensor units 2.

[0150] Each sensor unit 2 includes one or more magnetic sensors 4, e.g. tunnel magnetoresistance (TMR) sensors 4. For example, each sensor unit 2 may include one TMR sensor 4 or an array of TMR sensors 4. Each sensor unit 2 may provide one or more measurement channels depending on its magnetic sensor 4 configuration.

[0151] Although the invention requires only one type of sensor, other types of auxiliary sensors 13 may also provide additional sensor data via corresponding outputs 5 to be processed by the processing module 20 of the system 1 . The sensor unit 2 also has a signal conditioning module 6, a signal digitisation module 8, a digital signal processing (DSP) module 16, and an optional auxiliary digital signal processing module 18. The system also has a power management unit (PMU) 10, a memory 12, and a connectivity controller 14.

[0152] A display output 22 allows information (e.g. MCG, SCG and BCG traces) obtained by the system 1 to be displayed visually, for example, on a screen of the biomagnetic sensor device itself or a separate device (e.g. a personal computer or smart phone). A digital health & care platform output 24 allows the information obtained by the system to be stored and / or used by a digital health & care platform for the benefit of the subject or a clinician.

[0153] Turning to Fig. 2, an exploded view of part of a biomagnetic sensor device embodying the invention is shown. In this example, the portable biomagnetic sensor device has a single sensor unit 2. However, it is to be appreciated that the device may comprise further sensor units 2. The sensor unit 2 has a layered internal structure. The sensor unit 2 includes a sensor layer 4a, a motherboard layer 32 and lithium-ion battery layer 30 arranged in that order. The sensor layer 4a may be connected to the motherboard by pin connectors 34. The sensor layer 4a may comprise one or more TMR sensor(s) 4- e.g. an array of TMR sensors.

[0154] Each sensor unit 2 is contained in a plastic housing 38. The housing is substantially cuboidal and has a lid portion 38a and a base portion 38b, which may be connectable together via a friction fit or may click into place. The base portion 38b of the housing is arranged to be placed (e.g. pressed) against the body of the subject. The base portion 38b has a window or aperture 36 proximal to the sensor layer 4a to help allow biomagnetic signals to be sensed by the sensor layer 4a.

[0155] The base portion 38b has a contact surface (not shown) for contacting the body portion of the subject. The window or aperture 36 may be an opening in the contact surface of the base portion 38b. Extending away from the contact surface and toward the lid portion 38a are one or more side walls. Each side wall of the base portion 38b has an attachment portion 40 for attaching a strap thereto. The attachment portions 40 may be used to attach the sensor unit 2 to one or more straps (shown in Figs. 3a and 3b) so that the sensor unit 2 can be worn by the subject. The attachment portions 40 may additionally or alternatively be used to attach the sensor unit 2 to one or more further sensor units 2 to build a multi-channel biomagnetic sensor device.

[0156] As shown in Figs. 3a and 3b, the portable biomagnetic sensor device may have one or more sensor units 2.

[0157] Fig. 3a shows an example of a first variant of the biomagnetic sensor device in use attached to a subject 42a. This variant comprises one sensor unit 2 attached to the subject via a strap 44a secured to the chest of the subject 42a.

[0158] In the case of a multi-channel system, multiple sensors 4 can be placed into one sensor unit 2 or distributed over the body in respectively different locations for targeted coverage.

[0159] Therefore, the biomagnetic sensor system 1 may comprise a plurality of sensor units 2. The plurality of sensor units 2 may be daisy-chained together (i.e. connected in sequence in a ring) or connected to a common processing unit 80.

[0160] Fig. 3b shows an example of a second variant of the portable biomagnetic sensor device in-use attached to a subject 42b. This variant comprises a common processing unit 80 and eight sensor units 82 attached to the subject 42b via straps 44b secured to the chest of the subject 42b. As shown in Fig. 3b, additional shoulder straps 44b may be used to compensate for the increased size and weight of the biomagnetic sensor device. Fig. 4 shows how cardiac monitoring was previously performed in accordance with prior art methods, to track both electric cardiac activity and mechanical cardiac activity. Clinicians traditionally have used electrocardiographic sensors (ECG) to obtain an ECG signal in conjunction with accelerometers to obtain an SCG signal. This monitoring would require bulky clinical equipment which is costly and not accessible outside of clinical settings.

[0161] Embodiments of the invention allow for measurement of MCG, SCG and BCG signals, concurrently, using a single type of magnetic sensor and processing the sensed magnetic signal(s) to approximate the MCG, SCG and BCG signals.

[0162] As shown in Fig. 5, a TMR sensor 4 or TMR sensor unit obtains one or more sensed signals which are provided via output 3 to the processing module 20. Each sensed signal is a mix of MCG, SCG and BCG sources. The processing module 20 processes the signals to unmix the sensed signals into their constituent sources. Output from the processing module 20 are approximated MCG signals 7, SCG signals 9 and BCG signals 11 .

[0163] Fig. 6 is a schematic diagram showing how MCG, SCG and BCG signals can be obtained using a trained machine learning model 48 (e.g. using one or more artificial neural networks) in accordance with embodiments of the invention. As can be seen in Fig. 6 the system may use artificial intelligence (Al), i.e. a trained machine learning model 48 (e.g. a first artificial neural network), to approximate the MCG, SCG and BCG signals. The machine learning model 48 may be implemented in hardware by the DSP module 16 of the processing module 20 and may be continuously trained by a training module 17 which uses sensor data 3 from the magnetic sensor(s) 4 as training data 46. How the signals are separated into MCG, SCG and BCG will be explained in more detail below. A signal classifier 50 (e.g. a second artificial neural network) may receive the output of the trained machine learning model 48 (e.g. the first artificial neural network) before providing as output an MCG signal 7, an SCG signal 9 and a BCG signal 11 .

[0164] Fig. 7 shows preliminary results of MCG, SCG and BCG traces being extracted from simulated signals using the method described herein.

[0165] The operation of the system 1 will now be described with reference to the figures.

[0166] The system 1 comprising one or more sensor units 2 may be put into contact with the body of the subject 42a, 42b by being worn by the subject as shown in Figs. 3a and 3b. The sensor units 2 may be attached to the subject via one or more straps 44a, 44b.

[0167] On start-up of the system 1 , data acquired by the TMR sensor(s) 4 may optionally be provided via sensor output(s) 3 to the training module 17 and used as training data 46 to continuously train the model 48.

[0168] During operation of the system 1 , MCG, SCG and BCG signals are identified and separated from the sensed signal in real-time. The MCG, SCG and BCG signals may be provided for further clinical analysis to a user of the system 1 (e.g. the subject or a clinician for the subject).

[0169] Although not shown in the figures, the signal conditioning module 6 includes high-pass, low-pass, and notch filters to pre-process the signals from the sensors 4, 13 before they are digitised by the signal digitisation module 8. The digitised data is then transmitted to the DSP module 16 for further processing.

[0170] The DSP module 16 may employ various signal processing methods including filtering, denoising, single or multi-channel processing methods, blind source separation, independent component analysis, empirical mode decomposition, blind source extraction, blind identification, as well as machine learning methodologies such as deep learning. These techniques may be used to approximate an MCG signal 7, an SCG signal 9 and a BCG signal 11 from the sensed signal 3. One example of how this separation of MCG signals 7, SCG signals 9 and BCG signals 11 is achieved will be explained in more detail further below.

[0171] Optionally, an auxiliary DSP module 18 - e.g. provided in either a computer, a mobile application, or a remote server - may assist the DSP module 16 via connection 26, to enhance the processing capabilities and efficiency of the system 1. The processing module may be a distributed processing module comprising a primary processing module 20 and the auxiliary DSP module 18.

[0172] Data output from the system 1 can be transmitted either through wired or wireless means to a mobile phone application or a digital health and care platform for remote monitoring. This may allow for real-time display and analysis of the physiological MCG, SCG and / or BCG parameters, enabling remote monitoring of the condition of a subject (e.g. patient). This also allows a clinician to intervene promptly when necessary.

[0173] The power management unit 10 efficiently regulates power consumption and ensures prolonged operation of the device and system, by optimising power usage across the components of the system 1 , including: for example - the sensors 4,13; the signal conditioning module 6; the signal digitisation module 8, the DSP module 16; and the optional auxiliary DSP module 18.

[0174] The power management unit 10 may employ techniques such as power gating, voltage scaling, dynamic voltage and frequency scaling (DVFS), and duty cycling to adaptively control power consumption based on the operational requirements and constraints of the system 1. This may ensure that the system 1 operates with minimal power consumption while maintaining optimal performance.

[0175] By efficiently managing power consumption, the power management unit 10 may enhance the energy efficiency of the system 1 , extend the battery life, and reduce the need for frequent recharging or replacement of power sources. This may be particularly beneficial for portable or wearable applications where low power consumption is important.

[0176] Turning to Fig. 6, data from one or more magnetic sensors 4, e.g. TMR sensors, is provided to the trained model 48. For multi-channel systems, the plurality of signals output from the respective plurality of magnetic sensors 4 may be selectively coupled (e.g. using a multiplexer) to readout circuitry and data representative of the plurality of signals may be provided to the trained model 48.

[0177] As shown in Fig. 6, the one or more magnetic sensors 4 provide one or more signals which are pre-processed in the processing module 20 before being provided as input data to the trained machine learning model 48 (e.g. a first artificial neural network). The trained machine learning model 48 is trained to generate an MCG signal; an SCG signal and a BCG signal from the input data derived from the outputs of the magnetic sensor(s) 4. For quality control purposes, the data output from the trained machine learning model 48 relating to approximated MCG signals, SCG signals and BCG signals are provided to a signal classifier 50 (e.g. a second artificial neural network). The signal classifier 50 may provide as output an MCG signal 7, an SCG signal 9 and a BCG signal 11 .

[0178] The trained model 48 may be trained using machine learning techniques for constrained source separation. In this example, a machine learning technique of ‘independent component analysis’, known as FastICA, is used. Independent component analysis is a method of recovering a version of original source signals by multiplying data from mixed statistically independent signals by an unmixing matrix. FastICA is a particularly efficient version which is preferable for use in low power applications.

[0179] A plurality of sources (i.e. in this case MCG, SCG and BCG signals) may be mixed in one or more different ways. The system 1 records these one or more mixed signals by measuring the sensed magnetic field at one or more locations on the body via the magnetic sensor(s) 4. The one or more sensed signals are input to the processing module 20. Conditioned and digitised signals from the signal digitisation module 8 are output to the DSP module 16. In the DSP module 16, source separation software (e.g. FastICA) is used to separate the one or more signals into scaled and permuted versions of the original source signals - i.e. the target sources: MCG, SCG and BCG.

[0180] FastICA uses a measure called negative entropy (i.e. negentropy) to obtain scaled and permuted versions of the original source signals. This measure of negative entropy may be known as Jmain. Typically, FastICA minimises Jmainto achieve separation of source signals.

[0181] In this case, different to blind source separation, some prior knowledge is incorporated into the separation process to extract only the signals of interest. For this, auxiliary cost functions are used. This approach has been applied to other types of signals and may be termed constrained ICA, or semi-blind source extraction. However, the applicant proposes using this method for the first time in relation to MCG, SCG and BCG signals.

[0182] A set of auxiliary measures {JauX1, JauX2..., employed, each measure being respectively tuned to each one of the target sources (e.g. MCG, SCG and BCG signals).

[0183] A set of corresponding compound cost functions are also constructed - e.g. A = Jmain + a JauX1for extracting the first target source, J2= Jmain+ a JaUx2for extracting the second target source and so on (wherein a is a constant). The auxiliary measures provide the constraints relevant to each target source and are determined based on the spatial or temporal characteristics (i.e. signatures) of each target source. For example, the spatial characteristics of a target source may be based on the spatial distribution of the target source with respect to the body of the subject; while temporal characteristics may be based on particular temporal features of signal traces of the target sources.

[0184] These constraints may be determined using factor analysis (e.g. PARAFAC). Factor analysis may be used to determine spatial or temporal or spatio-temporal characteristics (i.e. signatures) of each signal of interest (i.e. target source). Another method for determining constraints may involve using change point detection on the sensed (i.e. mixed) signals. Such methods may involve use of the Teager-Kaiser Energy Operator tuned to separating MCG, SCG and BCG.

[0185] In other words, the method involves “constraining” the fastICA (or equivalent method) to “latch” to the signals of interest (i.e. MCG, SCG or BCG), one at a time, by minimising the compound cost functions.

[0186] Data representative of the (estimated) MCG, SCG and BCG signals of the subject are output from the trained model 48.

[0187] The estimated MCG, SCG and BCG signal data output from the trained model 48 is classified for quality control purposes by the signal classifier 50. This prevents inaccurate attribution of the source of the estimated signals. For example, if the trained model 48 extracts MCG instead of SCG in a particular step, the signal classifier 50 can identify this, preventing the MCG signal being labelled as an SCG signal. The signal classifier 50 may help to address potential inaccuracies arising from the fastICA’s output being the “scaled” and “permuted” version of the target sources.

[0188] The signal classifier 50 may be provided by an artificial neural network trained to classify MCG, SCG and BCG signal data. The signal classifier 50 may be trained using MCG, SCG and BCG traces as training data. The signal classifier 50 may be a linear classifier, e.g. using linear discriminant analysis. The signal classifier 50 may be a temporal convolutional network (TCN). Turning to Fig. 7, preliminary results of MCG, SCG and BCG traces being extracted from a simulated sensed signal using the above method are shown.

[0189] Preliminary results indicate that the method can successfully separate MCG, SCG and BCG traces from mixed signals. In this example, three simulated sensed signals 60 are shown. Each simulated signal represents a signal which would be output on a measurement channel from a respective sensor unit placed on the body of the subject, each sensor unit being in contact with the body of the subject. Each simulated signal represents a different mixture of MCG, SCG and BCG. The simulated signal traces 60 were obtained by mixing TMR sensor data with electrocardiographic (ECG) data obtained using traditional ECG techniques. Therefore, the upper-most traces 60 are examples of mixed sensed signals to be processed by the method in accordance with embodiments of the invention.

[0190] The estimated MCG signal trace 62, SCG signal trace 64 and BCG signal trace 66 derived from the method described above is shown below. The lowermost three traces 68 show the estimated signals overlaid on the original ECG data 70. A close match between the MCG signal output from the method and the original ECG signal 70 indicates that the electrical activity of the subject is accurately extracted when separating the sensed signal into MCG, SCG and BCG as set out herein.

[0191] It will be appreciated by those skilled in the art that the invention has been illustrated by describing one or more specific embodiments thereof, but is not limited to these embodiments; many variations and modifications are possible, within the scope of the accompanying claims.

Claims

Claims1. A biomagnetic sensor system comprising: one or more sensor units; the, or each, sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; and wherein the, or each, sensor unit is configured to be locatable on the body of the subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and wherein the processing module is configured to process the one or more signals from the one or more sensors to determine:(i) a magnetocardiographic (MCG) signal of the subject; and(ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

2. The biomagnetic sensor system of claim 1 , wherein the biomagnetic sensor system is a wearable biomagnetic sensor system.

3. The biomagnetic sensor system of claim 1 or 2, wherein each of the one or more signals representative of a sensed magnetic field is influenced by electrical activity and mechanical activity of the heart of the subject.

4. The biomagnetic sensor system of any preceding claim, wherein the sensed magnetic field at or adjacent to a location on the body of thesubject is a sensed magnetic field adjacent to biological tissue of the subject.

5. The biomagnetic sensor system of any preceding claim, wherein the, or each, sensor unit is configured to in use contact an external surface of the body of the subject.

6. The biomagnetic sensor system of claim 5, wherein the, or each, sensor is positioned within the, or each, sensor unit so that the, or each, sensor in use contacts an external surface of the body of the subject.

7. The biomagnetic sensor system of any preceding claim, wherein the, or each, sensor unit is configured to be attachable to a torso portion of the body of the subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to the torso portion on the body of the subject.

8. The biomagnetic sensor system of any preceding claim, comprising one or more securing elements arranged to attach the, or each, sensor unit to the body of the subject.

9. The biomagnetic sensor system of claim 8, wherein each of the one or more securing elements is a strap.

10. The biomagnetic sensor system of any preceding claim, wherein the, or each, sensor unit is arranged in layers and comprises a plurality of layers, at least one of the plurality of layers being a sensor layer, the plurality of layers being arranged so that in use the sensor layer is the layer most proximal to the subject.11 . The biomagnetic sensor system of any preceding claim, wherein the, or each, sensor unit is arranged within a housing.

12. The biomagnetic sensor system of claim 11 , wherein the, or each, housing comprises an opening proximal to the one or more sensors in the, or each, sensor unit.

13. The biomagnetic sensor system of claim 11 or 12, wherein the system is arranged so that in use the, or each, opening is positioned between the, or each, sensor of each sensor unit and the body of the subject.

14. The biomagnetic sensor system of any one of claims 11 to 13, wherein the housing of the, or each, sensor unit comprises one or more attachment portions for attaching a securing element thereto.

15. The biomagnetic sensor system of any preceding claim, wherein the, or each sensor unit is a magnetoresistance (MR) sensor unit.

16. The biomagnetic sensor system of any preceding claim, wherein the, or each sensor unit is a tunnel magnetoresistance (TMR) sensor unit.

17. The biomagnetic sensor system of claim 15 or 16, wherein the, or each TMR sensor unit comprises a plurality of TMR sensors, each TMR sensor comprising an array of magnetic tunnelling junctions fabricated on a substrate.

18. The biomagnetic sensor system of any preceding claim, wherein the one or more sensor units is a plurality of sensor units, each sensor unit being configured to output a signal on a respective measurement channelderived from the one or more signals indicative of a sensed magnetic field output from the one or more sensors therein.

19. The biomagnetic sensor system of claim 18, comprising at least three sensor units and at least three measurement channels.

20. The biomagnetic sensor system of any preceding claim, comprising a common processing unit; wherein the one or more sensor units is a plurality of sensor units and each of the biomagnetic sensor units are connected to the common processing unit; and wherein each sensor unit of the plurality of sensor units is configured to output a respective signal indicative of a respective sensed magnetic field to the common processing unit.21 . The biomagnetic sensor system of claim 20, arranged so that the plurality of sensor units is arranged azimuthally around the common processing unit.

22. The biomagnetic sensor system of any preceding claim, wherein the, or each, sensor unit is configured so that the, or each, sensor in use contacts the body of a subject either:(i) directly; or(ii) indirectly by contacting the body of the user via a layer of clothing up to 1 cm in thickness.

23. The biomagnetic sensor system of any preceding claim, wherein the processing module is configured to: use a first artificial neural network to generate:(i) data relating to the MCG signal of the subject; and(ii) data relating to the SCG signal of the subject and / or data relating to the BCG signal of the subject; and process the data generated by the first artificial neural network to determine:(i) the magnetocardiographic (MCG) signal of the subject; and(ii) the seismocardiographic (SCG) signal of the subject and / or the ballistocardiographic (BCG) signal of the subject.

24. The biomagnetic sensor system of claim 23, wherein the processing module is configured to use a second artificial neural network to classify the data output from the first artificial neural network.

25. The biomagnetic sensor system of claim 24, wherein the second artificial neural network is trained to classify input data as MCG signal data, SCG signal data and / or BCG signal data.

26. The biomagnetic sensor system of any one of claims 23 to 25, wherein the first artificial neural network is configured to perform a method of constrained independent component analysis on the one or more signals to determine:(i) the magnetocardiographic (MCG) signal of the subject; and(ii) the seismocardiographic (SCG) signal of the subject and / or the ballistocardiographic (BCG) signal of the subject.

27. The biomagnetic sensor system of claim 26, wherein the method of constrained independent component analysis is a method of constrained FastICA.

28. The biomagnetic sensor system of claim 26 or 27, wherein the method of constrained independent component analysis comprises:using a first auxiliary measure and a corresponding first compound cost function for determining the MCG signal; using a second auxiliary measure and a corresponding second compound cost function for determining the SCG signal; and / or using a third auxiliary measure and a corresponding third compound cost function for determining the BCG signal.

29. The biomagnetic sensor system of claim 28, wherein each auxiliary measure is determined based on one or more spatial characteristics and / or temporal characteristics of each signal type.

30. A biomagnetic sensor device comprising: one or more sensor units, the, or each, sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; and wherein the, or each, sensor unit is configured to be locatable on the body of a subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and wherein the processing module is configured to process the one or more signals from the one or more sensors to determine:(i) a magnetocardiographic (MCG) signal of the subject; and(ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.31 . A method for cardiac monitoring of a subject using a biomagnetic sensor system, wherein the biomagnetic sensor system comprises:one or more sensor units; the, or each sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; wherein the, or each, sensor unit is configured to be locatable on the body of a subject; the method comprising: locating the, or each, sensor unit on the body of a subject; outputting from the, or each sensor, a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and processing, using the processing module, the one or more signals from the one or more sensors to determine:(i) a magnetocardiographic (MCG) signal of the subject; and(ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

32. The method of claim 31 , comprising locating the, or each, sensor unit on the body of the subject so that the, or each, sensor in use contacts the body of a subject either:(i) directly by contacting the skin of the body portion of the user; or(ii) indirectly by contacting the body portion of the user via a layer of clothing up to 1 cm in thickness.

33. The method of claim 31 or 32, comprising the processing module: using a first artificial neural network to generate:(i) data relating to the MCG signal of the subject; and(ii) data relating to the SCG signal of the subject and / or data relating to the BCG signal of the subject; andprocessing the data generated by the first artificial neural network to determine:(i) the magnetocardiographic (MCG) signal of the subject; and(ii) the seismocardiographic (SCG) signal of the subject and / or the ballistocardiographic (BCG) signal of the subject.

34. The method of claim 33, comprising using a second artificial neural network to classify the data output from the first artificial neural network.

35. The method of claim 34, wherein the second artificial neural network is trained to classify input data as MCG signal data, SCG signal data and / or BCG signal data.

36. The method of any of claims 33 to 35, comprising the first artificial neural network performing a method of constrained independent component analysis on data relating to the one or more signals to determine:(i) the magnetocardiographic (MCG) signal of the subject; and(ii) the seismocardiographic (SCG) signal of the subject and / or the ballistocardiographic (BCG) signal of the subject.

37. The method of claim 36, wherein the method of constrained independent component analysis is a method of constrained FastICA.

38. The method of claim 37, wherein the method of constrained independent component analysis comprises: using a first auxiliary measure and a corresponding first compound cost function for determining the MCG signal; using a second auxiliary measure and a corresponding second compound cost function for determining the SCG signal; and / orusing a third auxiliary measure and a corresponding third compound cost function for determining the BCG signal.

39. The method of claim 38, comprising: minimising the first compound cost function to determine the MCG signal; minimising the second compound cost function to determine the SCG signal; and / or minimising the third compound cost function to determine the BCG signal.

40. The method of claim 38 or 39, wherein each auxiliary measure is determined based on one or more spatial characteristics and / or temporal characteristics of each respective signal type.41 . A method for cardiac monitoring of a subject using a biomagnetic sensor device, wherein the biomagnetic sensor device comprises: one or more sensor units; the, or each sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; wherein the, or each, sensor unit is configured to be locatable on the body of the subject; the method comprising: locating the, or each, sensor unit on the body of the subject; outputting from the, or each sensor, a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and processing, using the processing module, the one or more signals from the one or more sensors to determine:(i) a magnetocardiographic (MCG) signal of the subject; and(ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

42. A method comprising: inputting, to a trained machine learning model, data relating to one or more signals indicative of a sensed magnetic field at or adjacent to a location on a body of a subject, wherein the trained machine learning model is trained to determine, from data relating to the one or more signals:- a magnetocardiographic (MCG) signal of the subject; and- a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject; and using the trained machine learning model to determine:(i) a magnetocardiographic (MCG) signal of the subject; and(ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

43. A system for cardiac health monitoring of a subject comprising: one or more sensor units; the, or each, sensor unit including one or more sensors; and a processing module; wherein the, or each, sensor is configured to sense a magnetic field; and wherein the, or each, sensor unit is configured to be locatable on the body of the subject, such that the, or each, sensor in use outputs a signal indicative of a sensed magnetic field at or adjacent to a location on the body of the subject; and wherein the processing module is configured to process the one or more signals from the one or more sensors to determine:(i) a magnetocardiographic (MCG) signal of the subject; and(ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

44. A method comprising: receiving, at a processing module, data relating to one or more signals; the, or each, signal being indicative of a sensed magnetic field at or adjacent to a location on the body of a subject; the processing module processing the data to determine:(i) a magnetocardiographic (MCG) signal of the subject; and(ii) a seismocardiographic (SCG) signal of the subject and / or a ballistocardiographic (BCG) signal of the subject.

45. A computer readable storage medium bearing software comprising instructions that, when executed by a processing system, cause the processing system to perform the method of claim 44.

Images