Devices, systems, and methods for non-invasive early warning of physical danger

Abstract

A system, method, and apparatus for providing warnings before a medical condition crisis is disclosed. The system, method, and apparatus may include at least one electromagnetic (EM) field-related parameter sensor, at least one processor, and a communication unit. The processor may be configured to extract EM features from signals received from at least one EM sensor, compare them with pre-stored EM features of signals stored in a database, and diagnose a medical condition or issue a warning of an anticipated medical crisis based on the results of the comparison.

Description

Technical Field

[0001] Cross-Reference to Related Applications This application is a PCT patent application claiming the benefit of priority under 35 U.S.C. § 119(e) from U.S. Provisional Patent Application No. 63 / 533,359 filed on August 18, 2023. The entire content of the aforementioned application is incorporated herein by reference as if fully set forth in this specification in its entirety.

[0002] The present innovation relates to an apparatus, system, and method for early warning of physical crisis. More specifically, the present invention relates to a non-invasive early warning methodology prior to a physical / medical crisis based on detecting changes in the magnetophysiology of a subject.

Background Art

[0003] Medical conditions often require timely and accurate diagnosis to ensure appropriate treatment and management. Conventional diagnostic methods such as blood tests and imaging techniques can be invasive and time-consuming, and may not provide real-time data. These methods often require specialized equipment and trained personnel, which may limit their accessibility and frequency of use. Furthermore, interpretation of results from these conventional methods can be subjective, leading to potential inconsistencies in diagnosis.

[0004] Diabetes is a metabolic disorder that causes hyperglycemia or hypoglycemia. The hormone insulin moves sugar from the blood into cells to be stored or used for energy. In diabetes, the human body either does not produce enough insulin or cannot effectively use the insulin it produces.

[0005] If high blood sugar levels due to diabetes are left untreated, they can damage nerves, eyes, kidneys, and other organs. Type 1 diabetes is an autoimmune disease. The immune system attacks and destroys the cells in the pancreas that produce insulin. It is unknown what triggers this attack. About 10% of people with diabetes have this type. Type 2 diabetes develops when the body becomes insulin-resistant and sugar accumulates in the blood.

[0006] Today, blood levels are monitored through direct blood sampling, which is performed several times a day from the patient.

[0007] There are two main concerns regarding diabetic crisis: when blood sugar levels are too low or too high. This problem is even more serious in patients who have had the disease for a very long time, as they may not even experience any associated symptoms.

[0008] Diabetic hypoglycemia occurs when there is insufficient sugar (glucose) in the blood of a person with diabetes. Since glucose is the primary fuel source for the body and brain, a person cannot function properly when sugar levels are insufficient. Hypoglycemia is defined as a blood glucose level below 70 milligrams / deciliter (mg / dL) or 3.9 mmol / liter (mmol / L). Early signs and symptoms of diabetic hypoglycemia include dizziness, lightheadedness, sweating, hunger, tachycardia, difficulty concentrating, confusion, irritability or mood swings, anxiety or nervousness, and headache.

[0009] Hyperglycemia does not cause symptoms until glucose levels rise significantly, usually exceeding 180-200 milligrams / deciliter (mg / dL) or 10-11.1 millimoles / liter (mmol / L). Symptoms of hyperglycemia develop slowly over several days or weeks. The longer blood glucose levels remain high, the more severe the symptoms become. Early signs and symptoms of hyperglycemia include frequent urination, increased thirst, blurred vision, fatigue, and headaches.

[0010] Both hypoglycemia and hyperglycemia can cause confusion, loss of consciousness, and even death. Hypoglycemia is one of the leading causes of nocturnal death in diabetic patients. [Overview of the project] [Problems that the invention aims to solve]

[0011] Therefore, in order to enable early treatment and prevention of sudden medical crises, there is a need for non-invasive methods for the early detection and warning of anticipated physical / medical crises, such as hypoglycemia and hyperglycemia, prior to the onset of such crises. [Means for solving the problem]

[0012] Electromagnetic field (EMF) sensing can measure neuromuscular, metabolic / mitochondrial activity. Therefore, it may be a useful method for measuring changes in physical activity and serving as an early warning of changes related to blood glucose or other medical conditions.

[0013] The proposed device may include predictive and / or diagnostic capabilities for a multitude of other medical conditions, not limited to diabetes. Other medical conditions that may benefit from such a device may include, but are not limited to, conditions resulting from seizures or other changes to bodily homeostasis; psychiatric conditions such as stress, anxiety, depression, schizophrenia, and the potential onset of psychosis; epilepsy, a group of neurological disorders characterized by seizures whose onset is often unpredictable and which can result in physical harm to the patient; asthma, an inflammatory lung disease characterized by variable and recurrent seizures of airway obstruction and bronchospasm; the onset of infectious diseases such as COVID-19, influenza, and bacterial infections may be measured before the actual onset of symptoms; and prediction of various acute exacerbations such as various cardiac arrhythmias, multiple sclerosis, lupus or systemic lupus erythematosus, autoimmune disorders such as SLE, or other inflammatory conditions. Longer-term fluctuations may also indicate oncological processes.

[0014] Slight disturbances in electrophysiological activity may precede physiological changes measured by current modalities that are biochemical or, more obviously, electrical. For example, when blood glucose levels rise, mitochondrial activity slows down, which may reflect a decrease in EMF activity. Conversely, when insulin is administered, mitochondrial activity may increase, which in turn increases EMF activity, and subsequently lowers blood glucose levels.

[0015] Aspects of the present invention relate to systems, methods, and apparatus for providing advance warning of a medical condition crisis. The systems, methods, and apparatus may include at least one electromagnetic field (EMF) related parameter sensor, at least one processor, and a communication unit.

[0016] According to some embodiments, the processor may be configured to extract EM features from signals received from at least one EM sensor, compare the extracted EM features with pre-stored signal features stored in a database, and issue an alert for an anticipated crisis based on the results of the comparison.

[0017] According to some embodiments, the EMF-related parameter sensor may be selected from a list consisting of an EMF sensor and other physiological measurements such as pulse rate, HRV, body temperature, and blood pressure.

[0018] In some additional or alternative embodiments, the apparatus may further include an EM signal preprocessing module for digitizing and amplifying the EM signal to make it a processor-readable signal.

[0019] The device may be a non-invasive wearable device configured to be removably attached to various parts of the subject's body (e.g., arms and legs).

[0020] Apparatus according to some embodiments may include a non-temporary computer-readable storage medium in which program instructions are stored, and the program instructions are executable by at least one processor to receive signals indicating EMF measured by electromagnetic field (EMF)-related parameter sensors over a predetermined period of time, and in the inference stage, to apply a trained machine learning model that can diagnose or warn of a medical condition before a medical condition crisis occurs, based on the received signals.

[0021] A machine learning model may be trained to identify signal patterns indicating an anticipated medical crisis, and in the training phase, the machine learning model may be based on a training set that may include signals from (one or more) EMF-related parameter sensors and, for each of the signals, a label indicating a medical diagnosis or medical condition crisis event.

[0022] According to some embodiments, a method for issuing a warning before a medical condition crisis may include, by a signal analyzer processor, receiving EM-related signals measured over a predetermined period from one or more electromagnetic (EM) field-related parameter sensors in close proximity to the user's body; extracting EM features from the received signals; comparing the extracted EM features with pre-stored EM features to detect anomalies in the extracted EM features; and issuing a warning if the detected anomaly is associated with a pre-defined medical condition crisis.

[0023] Embodiments of the present invention may include a device for providing warnings before a medical condition crisis occurs. The device may include at least one EMF sensor adapted to measure at least one electromagnetic field (EMF) related parameter, at least one processor, and a communication unit. The at least one processor may be configured to extract one or more EM features from signals received from at least one EM sensor, compare the one or more extracted EM features with pre-stored (e.g., stored in a database) signal features, and issue a diagnosis or warning of an expected crisis based on the results of the comparison.

[0024] According to some embodiments, the at least one EMF sensor may be a magnetic sensor, and the EMF-related parameter may be a magnetic field amplitude.

[0025] Additionally or alternatively, the device may further comprise an EM signal pre-processing module configured to digitize and amplify EM signals to generate processor-readable signals.

[0026] According to some embodiments, the device may be wearable and may be configured to be releasably attached to a limb of a user.

[0027] The device may further comprise a non-transitory computer-readable storage medium having program instructions stored thereon. The program instructions may be executable by at least one processor to: receive a signal indicative of a measurement of at least one EMF-related parameter obtained by the at least one EMF sensor over a predefined period of time; and apply, based on the received signal, a trained machine learning model to issue a warning prior to an onset of a medical condition crisis.

[0028] Additionally or alternatively, the program instructions may be further configured to receive a training dataset. The training dataset may comprise (i) one or more signals indicative of measurements of at least one EMF-related parameter, and (ii) for each of the one or more signals, a label indicative of a medical condition or a medical crisis event. During a training phase, the at least one processor may train the machine learning model using the labels to identify signal patterns indicative of an impending medical crisis within the one or more signals.

[0029] Embodiments of the present invention may include a method for issuing a medical diagnosis or medical warning before a medical condition crisis occurs. Embodiments of the method may include a signal analyzer processor receiving EM-related signals measured over a predetermined period of time from one or more EM-field-related sensors located in close proximity to the user's body. Embodiments of the method may further include the processor extracting EM features from the received signals and comparing the extracted EM features with pre-stored EM features to detect anomalies in the extracted EM features. Embodiments of the method may then issue a warning if the detected anomaly is associated with a pre-defined medical condition crisis.

[0030] Embodiments of the present invention may include a system for classifying the medical condition of a target patient. Embodiments of the system may include one or more magnetic sensors, each adapted to generate magnetic sensor signals that describe the characteristics of a magnetic field in the target patient; a preprocessing module adapted to generate a digitized sampled version of the magnetic sensor signals; and at least one processor.

[0031] At least one processor may be configured to calculate a value of at least one EM feature representing the temporal evolution of the magnetic sensor signal, based on a digitized sampled version of the magnetic sensor signal. The at least one processor may then classify the state of a target patient according to one or more health criteria based on the at least one calculated EM feature value.

[0032] According to some embodiments, at least one processor may be further configured to compute the spectral distribution of at least one magnetic sensor signal of one or more magnetic sensors. Thus, at least one EM feature may represent the temporal evolution of the spectral distribution over a given period.

[0033] Additionally or alternatively, at least one processor may be further configured to decompose a digitized sampled version of the magnetic sensor signal into components selected from a list consisting of a trend component, a seasonal component, and a residual component. At least one EM feature is obtained by selecting from the said list of components (e.g., trend, seasonality, and / or residual component).

[0034] Additionally or alternatively, embodiments of the system may include a transmission module. At least one processor may be further configured to determine the persistence of a target patient's condition over a predefined period of time, and, following the determination of persistence, to issue a notification of the target patient's condition to at least one computing device via the transmission module.

[0035] According to some embodiments, at least one processor may be configured to acquire at least one pre-stored EM feature value for each of at least one subjects in a cohort of subjects. The at least one pre-stored EM feature value may be labeled according to one or more health criteria. The at least one processor may compare the at least one pre-stored EM feature value with a calculated EM feature value for a target patient and classify the target patient's condition based on the comparison.

[0036] Additionally or alternatively, at least one processor may be configured to classify the status of a target patient by (a) obtaining a pre-trained ML-based classification model to classify the status of a subject according to one or more health criteria based on at least one EM feature, and (b) inferring an ML-based classification model based on at least one EM feature of a target patient to classify the status of a target patient according to one or more health criteria.

[0037] Additionally or alternatively, at least one processor may be configured to obtain measurements of the concentration of biological substances in a subject from at least one biosensor. Based on the measurements, at least one processor may generate annotation data elements and label the subject according to one or more health criteria. At least one processor may then use the annotation data elements as supervisory information to train an ML-based classification model.

[0038] According to some embodiments, at least one biosensor may be a glucometer adapted to measure glucose levels in the subject's blood. One or more health criteria may be selected from a list consisting of elevated glucose levels, decreased glucose levels, steady glucose levels, hyperglycemia, hypoglycemia, and normal glucose levels.

[0039] According to some embodiments of the system, at least one processor may be communicatively connected to a medical device such as an insulin pump. The at least one processor may be further configured to communicate with the associated insulin pump in order to administer insulin based on a classification of the target patient's condition.

[0040] The subject matter considered to be the present invention is specifically pointed out and explicitly claimed in the final part of this specification. However, the present invention, along with its object, features and advantages, as well as with respect to both organization and operation, will be best understood by referring to the following detailed description in conjunction with the accompanying drawings. [Brief explanation of the drawing]

[0041] [Figure 1] This is a high-level block diagram of an exemplary computing device according to an embodiment of the present invention. [Figure 2A] This block diagram shows an overview of a system and apparatus according to embodiments of the present invention. [Figure 2B]This is a block diagram showing components of an analytical module that may be included in the system and / or apparatus of the present invention. [Figure 3] A flowchart of the method according to one embodiment of the present invention is shown. [Figure 4A] This is a diagram of a single spectrogram according to several embodiments of the present invention. [Figure 4B] This is a diagram of a bound spectrogram according to several embodiments of the present invention. [Figure 4C] This is an actual data acquisition spectrogram according to one embodiment of the present invention. [Figure 5A] The time-frequency maps of an empty room according to several embodiments of the present invention are shown. [Figure 5B] The time-frequency maps of a healthy patient according to several embodiments of the present invention are shown. [Figure 6A] This graph shows the temporal evolution of trends in (i) glucose level measurements and (ii) magnetic sensor measurements according to several embodiments of the present invention. [Figure 6B] This graph shows the cross-correlation between glucose level measurements and the trends in the magnetic sensor measurements shown in Figure 6A. [Figure 7] This table shows the performance parameters of embodiments of the present invention for predicting glucose-related health conditions such as hypoglycemia and hyperglycemia. [Modes for carrying out the invention]

[0042] For the sake of simplicity and clarity in the illustrations, it should be understood that the elements shown in the diagrams are not necessarily drawn to scale. For example, the dimensions of some elements may be exaggerated relative to others for clarity. Furthermore, where appropriate, reference numbers may be repeated between drawings to indicate corresponding or similar elements.

[0043] The following detailed description includes numerous specific details to provide a complete understanding of the invention. However, it will be understood by those skilled in the art that the invention may be carried out without these specific details. In other instances, well-known methods, procedures, and components are not described in detail so as not to obscure the invention.

[0044] Embodiments of the present invention are not limited thereto, but descriptions using terms such as “processing,” “computing,” “calculating,” “determining,” “establishing,” “analyzing,” and “checking” may refer to one or more operations and / or processes of a computer, computing platform, computing system, or other electronic computing device that manipulates data represented as physical (e.g., electronic) quantities in computer registers and / or memory, and / or converts it to other data similarly represented as physical quantities in computer registers and / or memory, or to a non-temporary storage medium of other information capable of storing instructions for performing operations and / or processes. Embodiments of the present invention are not limited thereto, but as used herein, the terms “plurality” and “a plurality” may include, for example, “multiple” or “two or more.” Throughout this specification, the terms “plurality” or “a plurality” may be used to describe two or more components, devices, elements, units, parameters, etc. As used herein, the term “set” may include one or more items. Unless expressly stated otherwise, the method embodiments described herein are not restricted to any particular order or sequence. Furthermore, some of the described method embodiments or elements thereof may occur or be performed simultaneously, at the same time, or in parallel.

[0045] Apparatus, systems, and methods according to embodiments of the present invention can monitor changes in an electromagnetic field (EMF) and / or other physiological symptoms such as pulse rate, heart rate variability (HRV), BP, pulse oximetry, and body temperature, also referred herein to as the user's EMF-related physiological parameters, and based on the identified changes, can predict one or more medical diagnoses or medical condition crises and provide a diagnosis or warning before a crisis occurs.

[0046] Refer to Figure 1, which shows a high-level block diagram of an exemplary computing device according to an embodiment of the present invention. The computing device 100 may include at least one processor or controller 105, such as a central processing unit processor (CPU). For example, the at least one processor or controller 105 may be a single high-performance CPU or a combination of a CPU and a dedicated digital signal processor (DSP) to enhance computational efficiency.

[0047] The computing device 100 may further include a chip or any suitable computing device or computing device capable of housing an operating system 115, memory 120, executable code 125, storage 130, input device 135, and output device 140.

[0048] At least one processor or controller 105 may be configured to perform and / or perform or function as various modules, units, etc., as described herein. Two or more computing devices 100 may be included, and one or more computing devices 100 may function as various components, for example, the components shown in Figures 2A and 2B. For example, the analysis module 203 in Figures 2A and 2B described herein may include components of computing device 100, or may be implemented as software components, hardware components, or any combination thereof.

[0049] For example, by executing executable code 125 stored in memory 120, the controller 105 may be configured to perform a method for predicting and alarming medical crisis as described herein. For example, the controller 105 may analyze amplified electromagnetic signals received from EM sensors on the user's body and / or EMF-related physiological signals received from current standard body monitoring sensors, and may use a communication unit to warn or alarm the user and / or caregiver if changes in electromagnetic fields and / or EMF / physiological measurements (and / or other EMF-related parameters) indicate a medical diagnosis or anticipated crisis or seizure as described herein.

[0050] The operating system 115 may include any code segment (for example, similar to the executable code 125 described herein) designed and / or configured to perform tasks including coordinating, scheduling, arbitrating, supervising, controlling, or otherwise managing the operation of the computing device 100, such as scheduling the execution of software programs or enabling software programs or other modules or units to communicate. The operating system 115 may be a commercial operating system.

[0051] Memory 120 may be, for example, random access memory (RAM), read-only memory (ROM), dynamic RAM (DRAM), synchronous DRAM (SD-RAM), double data rate (DDR) memory chip, flash memory, volatile memory, non-volatile memory, cache memory, buffer, short-term memory unit, long-term memory unit, or other suitable memory unit or storage unit, or may include these. Memory 120 may be, or include, multiple, possibly different memory units. Memory 120 may be a non-temporarily readable medium of a computer or processor, or a non-temporarily stored medium of a computer, such as RAM.

[0052] The executable code 125 may be any executable code, such as an application, program, process, task, or script. The executable code 125 may optionally be executed by the controller 105 under the control of the operating system 115. For example, the executable code 125 may be an application that extracts EM features from EM signals and / or other EMF-related parameter signals received from (one or more) EM sensors, compares the extracted EM features with a database, and issues a warning or alarm, as further described herein, if the extracted EM features are associated with features indicating a medical diagnosis or anticipated medical crisis.

[0053] For clarity, Figure 1 shows a single executable code 125, but a system according to an embodiment of the present invention may include multiple executable code segments similar to the executable code 125, which are loaded into memory 120 and cause the controller 105 to execute the methods described herein. For example, a unit or module described herein (e.g., the analysis module 203 in Figure 2) may be the controller 105 and the executable code 125, or may include them.

[0054] The storage 130 may be, for example, a hard disk drive, a floppy disk drive, a compact disc (CD) drive, a CD recordable (CD-R) drive, a Blu-ray disc (BD), a Universal Serial Bus (USB) device, or other suitable removable and / or fixed storage units, or may include them. Content may be stored in the storage 130 and may be loaded from the storage 130 into the memory 120 where the content can be processed by the controller 105. In some embodiments, some of the components shown in Figure 1 may be omitted. For example, the memory 120 may be a non-volatile memory having the storage capacity of the storage 130. Thus, although shown as a separate component, the storage 130 may be incorporated into or included in the memory 120.

[0055] According to some embodiments, the computing device 100 may include a cloud-based computer or server adapted to provide online storage and / or analysis of data in system 10.

[0056] The input device 135 may be, or include, a mouse, keyboard, touchscreen or pad, or any suitable input device. It will be recognized that any number of suitable input devices may be operably connected to the computing device 100 as shown in block 135. The output device 140 may include one or more displays or monitors, speakers, and / or any other suitable output devices. It will be recognized that any number of suitable output devices may be operably connected to the computing device 100 as shown in block 140. Any applicable input / output (I / O) devices may be connected to the computing device 100 as shown in blocks 135 and 140. For example, the input device 135 and / or the output device 140 may include a wired or wireless network interface card (NIC), a printer, a Universal Serial Bus (USB) device, or an external hard drive.

[0057] Embodiments of the present invention may include articles such as a non-temporary-readable medium of a computer or processor, or a non-temporary storage medium of a computer or processor, such as memory, a disk drive, or a USB flash memory, which, when executed by a processor or controller, encodes, contains, or stores instructions, such as computer executable instructions, that perform the methods disclosed herein. For example, an article may include a storage medium such as memory 120, computer executable instructions such as executable code 125, and a controller such as controller 105.

[0058] Some embodiments may be provided in a computer program product which may include a non-temporary machine-readable medium on which instructions are stored, which can be used to program a computer, controller, or other programmable device to perform the methods disclosed herein. Embodiments of the present invention may include a non-temporary machine-readable medium of a computer or processor that encodes, includes, or stores instructions, such as computer executable instructions, when executed by a processor or controller, to perform the methods disclosed herein, or articles such as a non-temporary storage medium of a computer or processor, such as memory, a disk drive, or a USB flash memory. The storage medium may include, but is not limited to, any type of disk or programmable storage device which includes any type of disk which includes semiconductor devices such as read-only memory (ROM) and / or random-access memory (RAM), flash memory, electrically erasable programmable read-only memory (EEPROM), or any type of medium which is suitable for storing electronic instructions. For example, in some embodiments, memory 120 is a non-temporary machine-readable medium.

[0059] System 10 according to some embodiments of the present invention may include, but are not limited to, a plurality of CPUs or any other suitable multipurpose or specific processor or controller (e.g., a controller similar to controller 105), a plurality of input units, a plurality of output units, a plurality of memory units, and a plurality of storage units. System 10 may further include other suitable hardware and / or software components.

[0060] In some embodiments, System 10 may include, or may not include, a personal computer, a desktop computer, a laptop computer, a workstation, a server computer, a network device, or any other suitable computing device. For example, System 10 as described herein may include one or more devices such as computing device 100.

[0061] Herein, Figure 2 is a block diagram illustrating an overview of an apparatus 200 or system 10 adapted to perform non-invasive monitoring of a subject's magnetic physiology according to an embodiment of the present invention. As detailed herein, embodiments of the present invention may subsequently provide early warnings before physical or medical danger based on the detection of changes in the subject's monitored magnetic physiology.

[0062] The apparatus 200 may include one or more electrophysiological sensors 201, such as electromagnetic (EM) sensors, adapted to measure an electrophysiological signal 201S. Additionally or alternatively, the sensors 201 may measure EMF-related signals or parameters 201S, as detailed herein.

[0063] According to some embodiments, the (one or more) EMF sensor 201 may include low-frequency EMF sensors such as currently available Nivio modules or xMR modules, which may comprise multiple ultra-high sensitivity magnetic sensors. The (one or more) EMF sensors 201 may be arranged in an array to detect biomagnetism in a subject, organ, or tissue of a subject. Such ultra-high sensitivity magnetic sensors 201 may be capable of detecting changes of less than 1 picotesla in the EMF of a subject, tissue, or organ.

[0064] The arrangement of one or more EMF sensors, such as the magnetic sensor 201 described above, in the device may enable non-invasive detection of magnetic field characteristics in a target patient, which can exhibit a variety of physiological states, as detailed herein.

[0065] Sensor 201 may be operably connected to a signal preprocessing module 202, which is adapted to generate a digital representation 202D of signal 201S readable by analysis module 203 for digital analysis. For example, preprocessing module 202 may employ adaptive amplification to amplify the EM or EM-related signal 201S to match the required signal amplitude and dynamic range. Preprocessing module 202 may sample (one or more) amplified signals 201S according to a predetermined sampling rate. Additionally or alternatively, preprocessing module 202 may use different types of analog-to-digital converters (ADCs) and / or filters to adapt to the required signal characteristics and noise levels, thereby generating a digital representation 202D of signal 201S.

[0066] As shown in Figure 2A, the apparatus 200 may further include an analysis module 203 adapted to analyze pre-processed EMF-related signals 202D to identify the subject's condition 203C.

[0067] Block diagram showing components of an analysis module 203 that may be included in the system 10 and / or apparatus 200 of the present invention (for example, the same as the analysis module 203 in Figure 2A). See also Figure 2B.

[0068] As detailed herein, the analysis module 203 may be implemented as a software module, a hardware module, or any combination thereof. For example, the analysis module 203 may include at least one processor or controller (e.g., 105 in Figure 1) configured to implement a method for classifying the status of a target patient, as detailed herein.

[0069] According to some embodiments, the analysis module 203 may be configured to calculate one or more EM features (e.g., 22F, 24F) that represent the temporal evolution of the sensor signal 201S (digital representation 202D of the signal 201S), including parameters such as changes in amplitude, frequency, and phase over time.

[0070] The analysis module 203 can then classify the target patient's condition based on these EM features (e.g., 22F, 24F) using predefined health criteria 28C, which can then be adjusted to detect a range of medical conditions.

[0071] For example, embodiments of the present invention may be adapted to classify medical conditions related to glucose levels in the blood of a target patient representing a clinical state. In this example, one or more health criteria 28C may include, for example, an expected increase in glucose levels in the subject's blood, an expected decrease in glucose levels, an expected steady glucose level, an expected hyperglycemic state, an expected hypoglycemic state, and an expected normal glucose level state.

[0072] According to some embodiments, the analysis module 203 may be configured to calculate the spectral distribution 22F of at least one sensor signal 201S (digital representation 202D of the signal 201S) of one or more magnetic sensors 201. For example, as shown in Figure 2B, the analysis module 203 may include a spectrogram generation module 22. The spectrogram generation module 22 may be adapted to calculate the spectral distribution 22F of at least one sensor signal 201S as a spectrogram based on the digital representation 202D of the (one or more) measurement signals 201S.

[0073] As is known in the art, a spectrogram can be a data structure, such as a matrix of values, that represents the spectrum of a signal's frequency as it changes over time. For example, the horizontal axis of a spectrogram may represent time, the vertical axis may represent frequency, and the values ​​of the entries in the matrix may represent the amplitude of the signal at each corresponding frequency and time point.

[0074] It will be understood that the spectrogram 22F (or any derivative thereof) may be, or may represent, the temporal evolution of the calculated spectral distribution 22F over a given period. As detailed herein, the analysis module 203 may then classify the state 203C of the target patient by using the spectrogram 22F (e.g., the evolution of the spectral distribution 22F) as a distinguishing EM feature.

[0075] As is known in the art, decomposing a signal into components, namely trend components, seasonal components, and residual components, can refer to the process of breaking down a time-series signal into three distinct parts: a trend component that captures the long-term progression or direction of the data; a seasonal component that captures repeating patterns or cycles within the data over a specific period; and a residual component that captures the remaining variability of the data not explained by the trend or seasonal components. Such decomposition can facilitate the analysis and understanding of underlying patterns and behaviors within the signal.

[0076] As shown in Figure 2B, the analysis module 203 may include a signal decomposition module 24 configured to decompose a digitized sampled version 202D of (one or more) sensor signals 201S into constituent components (for example, to generate the decomposition 24F). The constituent components 24F may include, for example, a trend component 24F-1, a seasonality component 24F-2, and a residual component 24F-3.

[0077] As detailed herein, the analysis module 203 may then use one or more components 24F (e.g., 24F-1, 24F-2, 24F-3) as distinguishing EM features for classifying the condition 203C of the target patient.

[0078] As shown in Figure 2B, the analysis module 203 may include or be associated with a database 30DB. Database 30DB may maintain pre-stored EM feature values ​​(e.g., 22F, 24F) for each subject (e.g., human subjects) of at least one subject in a cohort of subjects. The pre-stored EM feature values ​​22F, 24F may be labeled or annotated (e.g., by a human expert) according to one or more health criteria 28C.

[0079] In the case of monitoring diabetic patients, the pre-stored EM feature values ​​in the database 30DB may include spectral distribution 22F values ​​and / or decomposition component 24F values ​​labeled according to the diagnosed state of each subject (e.g., normal, hypoglycemia, and hyperglycemia).

[0080] According to some embodiments, the analysis module 203 may compare at least one pre-stored EM feature value (22F, 24F) with the calculated EM feature value (22F, 24F) of the target patient and classify or categorize the target patient's condition 203C based on this comparison. Examples of such comparisons are provided herein (for example, in relation to Figure 7).

[0081] Additionally or alternatively, as shown in Figure 2B, the analysis module 203 may include at least one ML-based classification model 28 which may be configured to classify the state 203 of a target patient based on one or more EM feature values ​​(e.g., 22F and / or 24F) according to one or more health criteria.

[0082] According to some embodiments, the ML-based classification model 28 may be based on the application of evolutionary computational methods, such as genetic algorithms, to identify the optimal order of computational operations and machine learning models for accurately and effectively detecting correlation signals.

[0083] Embodiments of the present invention may apply various preprocessing operations, such as filtering, transformation, and statistical testing, to the data to prepare it for the machine learning stage. Numerous tools are available, each with its own parameters, and several tools may be combined hierarchically. The output from this preprocessing stage is then fed into a second stage, in which a machine learning model 28 may be constructed. Again, many tools are available, each with its own specific parameters.

[0084] To find the best combination of parameters, the inventors employed an iterative trial-and-error search process aimed at finding the best combination from trillions of possible options. A genetic algorithm can monitor this stage and intelligently optimize the search process, rather than thoroughly testing all possibilities by brute force, which would be impossible to achieve within a reasonable timeframe.

[0085] When applied to available data, the iterative search process identified preprocessing techniques such as trend detection, seasonality detection, anomaly detection, and data normalization tools as beneficial. During the machine learning phase, supervised machine learning methods for time series forecasting proved effective.

[0086] Additionally or alternatively, the ML-based classification model 28 may be configured to classify the target patient's condition 203 based on a digitized sampled version 202D of (one or more) sensor signals 201S.

[0087] According to some embodiments (as shown in Figure 2B), the ML-based classification model 28 may include several separate models (e.g., shown in 28-1, 28-2) each dedicated to classifying target subjects based on their respective unique types of EM features or inputs (e.g., 22F, 24F, 202D). The results of the individual models (e.g., 28-1, 28-2) can then be summarized (e.g., weighted by classification confidence values) to obtain a classification value 203C. Alternatively, the ML-based classification model 28 may include a unified classification model adapted to generate a classification value 203C based on various types of EM feature values ​​(e.g., 22F and / or 24F) and / or digitized sensor signals 202D.

[0088] As is known in the art, the term “feature” in ML-based models can refer to individual measurable properties or characteristics of the observed phenomenon.

[0089] In this context, “features” may include the values ​​of input EM features (e.g., 22F, 24F) used by the ML model to make predictions or classifications, and / or the values ​​of the digitized sensor signal 202D.

[0090] Additionally or alternatively, the term “feature” may further refer to any derivative of the input variables (e.g., derivatives of EM features 22F, 24F). For example, in the context of a neural network (NN)-based classifier, the term “feature” 22F, 24F may also refer to any combination of the input variables (22F, 24F), such as represented by the weights of individual nodes in the NN.

[0091] ML-based classification models 28 can be implemented using various machine learning algorithms such as decision trees, support vector machines, or neural networks, depending on the specific requirements and complexity of the classification task. ML-based models 28 can be trained on diverse datasets, including a wide range of patient data, ensuring robustness and generalizability across different patient populations.

[0092] According to some embodiments, system 10 may incorporate a federative learning approach in which an ML-based classification model 28 is trained across multiple decentralized devices or servers 100 holding local data samples without exchanging them, thereby enhancing data privacy and security while benefiting from large and diverse datasets. Decentralized devices or servers may include, for example, onboard computing devices such as computing device 100 in Figure 1.

[0093] In another example, a decentralized device or server 100 may include at least one local computing device 100, such as a smartphone device, associated with a specific target.

[0094] In yet another example, the decentralized device or server 100 may include an online (e.g., cloud-based) server that can be adapted to provide personalized analytical services to multiple devices.

[0095] The device 200 may be designed to support a continuous learning method, thereby allowing the ML-based model 28 to be regularly updated with new data to improve accuracy and adapt to evolving medical knowledge. Furthermore, the device 200 may include a user interface (UI, e.g., input device 135 in Figure 1) that can enable healthcare providers to input additional patient information or adjust health criteria to provide a customizable, interactive diagnostic tool.

[0096] According to some embodiments, the device 200 may include or be associated with one or more biosensors 40M configured to measure the biomedical parameters of a subject.

[0097] For example, the biosensor 40 may be configured to measure the concentration of a biological or chemical substance in a subject. In such embodiments, the biosensor 40 may be a glucometer adapted to measure the glucose concentration in the subject's blood, a pulse-oxygenation sensor adapted to measure the oxygenation in the subject's blood, and so on.

[0098] In another example, the biosensor 40 may be configured to measure parameters of electrical signals associated with a subject's heart, such as pulse rate, pulse waveform, and heart rate variability (HRV). In yet another example, the biosensor 40 may be configured to measure physical parameters such as body temperature.

[0099] According to some embodiments, the system 10 (e.g., device 200) may include a diagnostic module 220. The diagnostic module 220 receives EMF-related signals and parameters (e.g., features 22F, 24F) and / or physiological measurements 40M as inputs and compares these inputs to baseline values ​​to reach a diagnosis 220D (e.g., diabetic) or warning 205 / 206 (e.g., a serious condition in the near future) of the subject.

[0100] For example, it is assumed that cancer patients exhibit elevated levels of EMF-related signals (e.g., 202D) or parameters (e.g., features 22F, 24F) compared to a cohort of similar subjects (e.g., similar age, sex, and physique). Therefore, the diagnostic module 220 may compare the measured EMF-related signals and / or parameters to predetermined baseline values ​​in a cohort of healthy or diseased subjects to derive a cancer diagnosis 220D, which may be output as a notification 207 to the relevant computer device.

[0101] In another example, it is assumed that an epileptic subject exhibits changing levels of EMF-related signals (e.g., 202D) or parameters (e.g., features 22F, 24F) before and during a seizure. Thus, the analysis module 203 may compare the measured EMF-related signals and / or parameters to the patient's previously established baseline values ​​to issue an alert 205 / 206 for an approaching seizure.

[0102] Additionally or alternatively, the ML model 28 may accept physiological measurements 40M as input, along with EMF-related signals and parameters (e.g., features 22F, 24F), to classify the subject's state 203C.

[0103] Additionally or alternatively, system 10 may train an ML model 28 to classify the subject's state 203C based on EMF-related parameters (e.g., features 22F, 24F) using physiological measurements 40M as supervisory data.

[0104] For example, the biosensor 40 may include a glucometer 40M adapted to measure blood glucose levels in the blood of diabetic patients. During the training phase, the device 200 may receive the measured values ​​40M of the concentration of a substance (e.g., glucose) and generate annotation data elements 40L that label subjects according to one or more health criteria (e.g., normal, hypoglycemia, hyperglycemia). The analysis module 203 may then employ a training method such as a backpropagation training method and use the annotation data elements 40L as supervisory information to train an ML-based classification model.

[0105] In a subsequent inference stage, the analysis module 203 can be configured to acquire a trained ML-based classification model 28 and classify the target patient's condition by inferring it based on at least one EM feature of the target patient (e.g., 22F, 24F, or any derivative thereof), thereby classifying the target patient's condition 203C according to one or more health criteria 28C.

[0106] According to some embodiments, the device 200 may include a power supply (not shown) and a transmitting module 204, such as a wireless communication unit, adapted to transmit one or more notifications or alerts 204T of the status 203C of the person being monitored (for example, to a doctor's or caregiver's computing device).

[0107] According to some embodiments, the device 200 can communicate with a computing device, such as the computing device 100 in Figure 1, via a dedicated application for further data analysis, and generate alerts as needed.

[0108] For example, the device 200 may determine the persistence of the target patient's condition 203 (for example, over a predetermined period of time). Following this determined persistence, the device 200 may issue a notification 204T of the target patient's condition 203C to at least one computing device 100 via the transmission module 204.

[0109] One or more sensors 201, a preprocessing module 202, an analysis module 203, and a power supply and transmission unit 204 (also referred to herein as the “communication unit”) may be placed within a wearable mount 208 that can be detachably attached to a patient, for example, around the limbs or muscles. For example, the wearable mount 208 may have sensors 201 attached to the biceps, wrist, calf, or thigh of the subject or patient being monitored.

[0110] According to some embodiments, the sensor 201 may be positioned on two or more locations on the patient's body, for example, by using two or more wearable mounts 208, or by using a wearable mount having multiple sensors 201 located on different locations.

[0111] The device 200 may sense the patient's electrophysiological activity over a given time frame and then analyze that activity to learn or monitor the parameters of the subject's electrophysiological activity.

[0112] According to some embodiments, the non-invasive warning device 200 may be positioned near the patient, for example, by a wearable mount 208. EMF-related sensors, such as (one or more) EM sensors and / or (one or more) HRV sensors 201, may measure signals 201S representing the patient's magnetic field and / or HRV.

[0113] The preprocessing module 202 may amplify the signal 201S and / or convert the physiological signal readings 201S received from (one or more) sensors 201 into readable digitized EMF signals or data 202D for the processor or controller of the analysis module 203 (e.g., 105 in Figure 1).

[0114] The processor 105 may be adapted to analyze the digitized EMF signal or data 202D to determine the patient's condition. For example, the analysis module 203 may or may include a machine learning (ML) model configured to predict and / or indicate, based on the digitized EMF signal 202D, whether the patient shows signals representing an early sign of a medical condition crisis, hyperglycemia, or hypoglycemia (i.e., high or low glucose concentration in the blood).

[0115] If such an instruction occurs, the device 200 may send a warning or notification 204T of state 203C to one or more computing devices (e.g., 100 in Figure 1) via a transmitting module (e.g., a wireless communication module) 204.

[0116] For example, the device 200 may send a notification 204T of state 203C to a local application 205 installed on the same computing device (e.g., a smartphone) as the analysis module 203.

[0117] Additionally or alternatively, the device 200 may send a notification 204T of condition 203C to an application 206 on a remotely connected computing device 100 to alert the patient and / or caregiver of the abnormal condition 203C that is occurring and to request further action to re-stabilize the patient (e.g., stabilize the glucose level in the patient's blood).

[0118] According to some embodiments, the system 10 may include a device 200 and / or one or more remote computing devices 100 (e.g., user / patient, caregiver). The system 10 may inform the patient's physician of the current situation 203C 204T, enabling intervention if an alarm is not handled or treatment is ineffective.

[0119] According to some embodiments, the analysis module 203 may read digital signals or data 202D from the preprocessing module 202 and compare the current reading with previous crisis data using the patient database 30DB. This allows the analysis module 203 to determine the patient's current condition 203C.

[0120] Additionally or alternatively, the device 200 may further check state 203 based on the persistence of its appearance. For example, the transmitting module 204 may issue a warning notification 204T when state 203C appears or persists for a predetermined duration.

[0121] Changes in the EM signal can range from extremely low frequencies, such as 0.1 Hz to 10 kHz, to picotesla frequencies.

[0122] According to some embodiments, an ML algorithm can be trained with EM signals and / or other physiological parameters acquired prior to a medical diagnosis, event, or crisis. During the training phase, the ML model is provided with signals acquired a predefined time before known medical events, as well as signals not associated with medical events. Unsupervised training may be used to train the ML model to identify signals (e.g., EMF signals) that are associated with or predict a patient's medical diagnosis or event.

[0123] According to some embodiments, the system 10 may include or be associated with an insulin pump 50 (for example, it may be connected in a communicative manner). At least one processor 105 may be configured to communicate with the associated insulin pump 50 to administer insulin to a subject based on the classification 203C of the target patient's condition, for example, if the classification 203C indicates an expected state of hyperglycemia.

[0124] The following methods, described in Figure 3 of this specification, can illustrate the function of the analysis module 203, but are not limited to these.

[0125] Herein, we refer to Figure 3, a flowchart of a method for issuing a warning before a medical condition such as a hyperglycemic crisis occurs, according to some embodiments of the present invention. In step 305, the patient may be monitored for a given period of time during which the patient maintains normal blood glucose levels while maintaining normal daily activities, and this stage is called the “data collection” or training stage. During this stage, EM signals and patterns 202D, as well as other measurements, may be collected from sensors 40, such as a blood glucose sensor 40.

[0126] The EM signal and pattern 202D may be labeled 40L according to the physiological measurement 40M received from the sensor 40S, for example, as a pattern associated with a normal or abnormal condition 203C.

[0127] In this example of a diabetic patient, the labeled data 40L may include the patient's "normal activity" (e.g., normal glucose levels) or "abnormal activity" (e.g., "hyperglycemia" or "hypoglycemia").

[0128] According to some embodiments of the present invention, the analysis module 203 may use ML technology to extract EM features from the EM signal 202D (step 310) to define patterns representing the normal and / or abnormal activity of the patient.

[0129] According to some embodiments, the analysis module 203 may use unsupervised machine learning techniques to divide those patterns into different activities that the patient is accustomed to performing. If the algorithmic classification achieves a sufficiently high level of accuracy (for example, if the correlation between the perceived signals or patterns is greater than 0.5 (r≧0.5) with the medical event), the collected patterns may be stored as a criterion for further analysis.

[0130] As shown in step 305, during the inference phase, system 10 may record the patient's electrophysiological state over a given time interval (e.g., 1 minute). Subsequently, system 10 may extract relevant EM patterns that can indicate each of the patient's predefined activity levels (step 310).

[0131] According to some embodiments, an AI algorithm may be used to define which of the predefined activities the current state corresponds to, as seen in step 315. If the current state deviates from a predefined activity (e.g., indicating a potential crisis in step 320), and this finding is repeated through several observations (steps 325-340), or if the deviation increases over time, the system may issue a warning 204T to the patient (step 345).

[0132] If these deviations persist for an extended period, the system may also issue a warning 204T to a computing device 100 associated with a designated person, such as a family member, the patient's doctor, or medical support.

[0133] The inventors conducted the following experiment to demonstrate the effectiveness of the present invention in checking the condition of a subject.

[0134] In these experiments, sensor 201 included a pair of commercially available detectors (e.g., TDK, Nivio xMR). The Nivio-based sensor 201 was capable of operating under geomagnetic conditions in a DC magnetic field range of less than ±60 μT using the recommended circuitry. Sensor 201 was able to measure with high sensitivity at the picotesla level, i.e., less than 1 / 1,000,000 of the Earth's magnetic field.

[0135] Sensor 201 was sensitive to a frequency range of 0.1 Hz to 1 kHz, and the measurable AC magnetic field achieved was up to ±250 nanotesla (nT).

[0136] The sensor 201 was enclosed in a plastic casing and isolated from external magnetic fields in all but one direction. The sensor 201 was secured to the patient's body with a soft bandage so that the unisolated side could detect the magnetic signal 201S from as close to the body as possible. The preprocessing module 202 acquired the signal 201S in differential mode, for example, showing the difference between the readings of the two detectors, to eliminate the influence of ambient noise sources.

[0137] Two baseline measurements were taken.

[0138] For the initial baseline measurements, device 200 was operated in an empty patient room. Additional surrounding electronic devices were measured separately. No signals were detected within the range of interest.

[0139] For the second baseline measurement, patients without diabetes were monitored as a reference for the diabetic patients who were participants in the experiment.

[0140] The sampling rate of signal 202D was 20 kilohertz (kHz). Measurements were taken at 6-hour intervals. A one-dimensional continuous wavelet transform was applied to signal 202D.

[0141] Here, we refer to Figures 4A and 4B, which are schematic diagrams of spectrograms. As is known in the art, a spectrogram shows the temporal changes in the spectral (e.g., frequency-related) characteristics of a signal over time.

[0142] It will be understood that long-term spectrograms may undergo overall normalization and adaptation. Therefore, spectrograms observed at short intervals may have higher-grained frequency data compared to spectrograms at longer intervals. Accordingly, embodiments of the present invention may employ a process referred to herein as “spectrogram stitching” to support the analysis of long-term spectral data.

[0143] According to some embodiments of the present invention, Figure 4A shows a single spectrogram, while Figure 4B shows a stitched spectrogram, enabling a continuous representation of the spectral characteristics of a signal.

[0144] See also Figure 4C, which is an image showing a stitched spectrogram of actual data obtained by embodiments of the present invention.

[0145] To process this amount of data, the measurement interval was divided into 50 equal intervals. Each was processed separately, and the results were stitched together on a time-frequency map. The map is a mosaic of 50 spectrograms.

[0146] Here, we refer to Figures 5A and 5B, which show time-frequency maps of an empty room and a healthy patient, respectively, serving as reference maps. By comparing Figures 5A and 5B, it will be clear that in the absence of a patient, the acquired data has very few features. In other words, the majority of the acquired signal 201S was obtained from a human subject and not as a result of ambient noise.

[0147] See also Figures 6A and 6B. Figure 6A is a graph showing the temporal evolution of (i) glucose level measurement 40M and (ii) trend 24F-1 of the measurement from the magnetic sensor 201, according to several embodiments of the present invention. Figure 6B is a graph showing the cross-correlation between the glucose level measurement 40M and the trend 24F-1 of the measurement from the magnetic sensor 201, as shown in Figure 6A.

[0148] As shown in Figure 6A, at point A, the monitored subject received a rapid administration of insulin without food. Between point A and point B, the subject reported feeling unwell, and at point B, EMF-related parameters (e.g., feature 22F / 24F, in this example, trend feature 24F-1) began to rise. Approximately 400 seconds later, at point C, the subject's blood glucose level began to decrease.

[0149] In other words, at point B, the EMF-related parameters (e.g., feature 24F) predict the upcoming change in the subject's blood glucose approximately 7 minutes earlier. Therefore, embodiments of the present invention can result in the prediction of a change in the subject's state 203C, i.e., the prediction of a decrease in blood glucose.

[0150] Between point C and point D, embodiments of the present invention may generate a prediction of the subject's condition 203C as an anticipated medical crisis. In this case, the medical crisis 203C is a state of hypoglycemia (e.g., a blood glucose concentration of less than 70 milligrams (mg) / deciliter (dl)), which actually manifested at point D.

[0151] In other words, EMF-related parameters (e.g., feature 24F) have also been shown to predict impending crisis conditions, such as hypoglycemia, approximately 35–40 minutes before the onset of the condition. Thus, embodiments of the present invention can provide warnings for anticipated conditions 203C, such as crisis conditions, well before the crisis manifests.

[0152] At point D, the subject was given food, which resulted in another predicted change in the subject's medical condition 203C, namely a rapid increase in blood glucose, which actually occurred between points D and E. As shown at point F, this increase in blood glucose led to another predicted crisis condition 203C, this time a state of hyperglycemia (e.g., blood glucose concentration above 180 mg / dl).

[0153] By observing Figures 6A and 6B, it will be understood that there is a significant cross-correlation between the EMF-related parameters of the measured EM or magnetic field provided by the EM or magnetic sensor 201 (in this case, feature 24F, e.g., trend 24F-1) and the ground truth glucose values ​​provided by the biosensor 40 (e.g., glucometer) after insulin injection and food intake.

[0154] Furthermore, it will be understood that significant changes in the monitored characteristic 24F (e.g., trend 24F-1) may well precede expected changes (e.g., decreases and increases) in the subject's state (e.g., blood glucose levels).

[0155] See also Figure 7, which is a table showing the performance parameters of embodiments of the present invention when predicting glucose-related crisis health conditions 203C (e.g., hypoglycemia and hyperglycemia).

[0156] As shown in Figure 7, a gap of approximately 1000 seconds (between 1078 and 886) was typically observed between the crisis prediction 203C made by System 10 and the actual occurrence of the crisis. The performance parameters of System 10 in generating these predictions (e.g., accuracy, specificity, and F1) are also provided.

[0157] The "threshold" values ​​for each state (hypoglycemic crisis and hyperglycemic crisis) were obtained from a database 30 representing multiple measurements. These multiple measurements may be associated with the historical measurements of a single subject, e.g., a target subject of interest. Additionally or alternatively, the multiple measurements may be associated with a cohort of subjects with similar characteristics (e.g., sex, age, weight, and health status or traits). Embodiments of the present invention may calculate the thresholds shown in Figure 7 to predict 203C using the performance parameters described above.

[0158] According to some embodiments, the analysis module 203 in Figure 2A may analyze EMF-related parameters or features (e.g., 22F, 24F, 202D) of the measured EM signal 201S by comparing the values ​​of these features with thresholds in Figure 7. Based on this comparison, the analysis module 203 may conclude or determine the subject's state 203C. In relation to the example in Figure 7, if the feature value 24F (e.g., 24F-1) exceeds a value of 0.13439, the analysis module 203 may conclude that the subject is expected to enter a hypoglycemic state. In another example, if the feature value 24F (e.g., 24F-1) falls below a value of -0.01496, the analysis module 203 may conclude that the subject is expected to enter a hyperglycemic state.

[0159] As detailed herein, embodiments of the present invention can provide practical applications in the fields of diagnostic support, preventive care, and well-being. Accordingly, embodiments of the present invention can provide improvements over currently available diagnostic techniques by generating non-invasive, reliable, and predictive indicators of a patient's condition.

[0160] The non-limiting examples provided herein primarily relate to health conditions associated with blood glucose levels. The inventors hypothesize that observed changes in a subject's response to EM and / or magnetic fields correlate with metabolic activity at the cellular and / or organelle levels (e.g., mitochondria), as shown, for example, in Figure 6A.

[0161] Therefore, it will be understood that embodiments of the present invention should not be limited to the identification of glucose-related states. In other words, any sign associated with metabolic diseases, neuromuscular diseases, metabolic neoplastic diseases, inflammatory diseases, or infectious diseases may be detected by changes in a person's EMF or EMF fingerprint, which are represented by MF, frequency, waveform, etc.

[0162] Furthermore, given sufficient training data, the EM or magnetic signal can also be identified or predicted by embodiments of the present invention by analyzing it.

[0163] Unless expressly stated otherwise, the method embodiments described herein are not restricted to any particular chronological or chronological order. Furthermore, some of the described method elements may be skipped or repeated during a series of operations of the method.

[0164] While specific features of the present invention have been illustrated and described herein, many modifications, substitutions, alterations, and equivalents will be conceivable to those skilled in the art. Therefore, the appended claims are intended to encompass all such modifications and alterations that fall within the true spirit of the invention.

[0165] Various embodiments are presented. Each of these embodiments may, of course, include features from other embodiments presented, and embodiments not specifically described may include various features described herein.

Claims

1. At least one EMF sensor adapted to measure at least one electromagnetic field (EMF) related parameter, At least one processor, Communication unit and A device for providing warning before a medical condition becomes critical, comprising: The at least one processor is configured to extract one or more EM features from the signal received from the at least one EM sensor, compare the one or more extracted EM features with pre-stored signal features stored in a database, and issue a diagnosis or warning of an expected crisis based on the result of the comparison. Device.

2. The apparatus according to claim 1, wherein the at least one EMF sensor is a magnetic sensor, and the EMF-related parameter is the amplitude of the magnetic field.

3. The apparatus according to any one of claims 1 to 2, further comprising an EM signal preprocessing module configured to digitize and amplify the EM signal to generate a signal readable by the processor.

4. The apparatus according to any one of claims 1 to 3, wherein the apparatus is a wearable device configured to be releasably attached to the user's limbs.

5. The system further comprises a non-temporary computer-readable storage medium in which program instructions are stored, A signal is received indicating a measurement of at least one EMF-related parameter by the at least one EMF sensor over a predetermined period of time. Based on the received signals, a machine learning model trained to issue warnings before a medical condition becomes critical is applied. The apparatus according to any one of claims 1 to 4, which is executable by the at least one processor.

6. The aforementioned program instruction is, (i) receive a training dataset including one or more signals indicating a measurement of at least one EMF-related parameter, and (ii) for each of the one or more signals, a label indicating a medical condition or medical crisis event, and During the training phase, the machine learning model is trained using the labels to identify signal patterns indicating a predicted medical crisis within one or more signals. The apparatus according to claim 5, further configured as follows.

7. A method of issuing a medical diagnosis or medical warning before a medical condition becomes critical, The signal analyzer's processor receives EM-related signals measured over a predetermined period from one or more electromagnetic (EM) field-related sensors located in close proximity to the user's body, Extracting EM features from the received signal, The extracted EM features are compared with pre-stored EM features to detect anomalies in the extracted EM features, The system issues a warning if the detected anomaly is associated with a predefined medical condition crisis. Methods that include...

8. A system for classifying the medical status of a target patient, wherein the system is One or more magnetic sensors, each adapted to generate a magnetic sensor signal indicating the characteristics of the magnetic field in the target patient, A preprocessing module adapted to generate a digitized sampled version of the magnetic sensor signal, At least one processor and Equipped with, The aforementioned at least one processor is Based on the digitized sampled version of the magnetic sensor signal, the value of at least one EM feature representing the temporal evolution of the magnetic sensor signal is calculated. Based on the at least one calculated EM feature value, the condition of the target patient is classified according to one or more health criteria. It is configured in such a way. system.

9. The system according to claim 8, wherein the at least one processor is further configured to calculate the spectral distribution of at least one magnetic sensor signal of the one or more magnetic sensors, and the at least one EM feature represents the temporal evolution of the spectral distribution over a predetermined period of time.

10. The system according to any one of claims 8 to 9, wherein the at least one processor is further configured to decompose the digitized sampled version of the magnetic sensor signal into constituent components selected from a list consisting of a trend component, a seasonal component, and a residual component, and the at least one EM feature is selected from the list of constituent components.

11. The system further comprises a transmission module, and the at least one processor is The persistence of the aforementioned condition in the target patient over a predetermined period of time is determined, Following the determination of persistence, a notification of the target patient's condition is issued to at least one computing device via the transmission module. The system according to any one of claims 8 to 10, further configured as follows.

12. The aforementioned at least one processor is Obtain at least one pre-stored EM feature value for each subject in at least one of the subject cohorts, and at least one pre-stored EM feature value is labeled according to one or more health criteria. The at least one pre-stored EM feature value is compared with the calculated EM feature value of the target patient. Based on the above comparison, the above condition of the target patient is classified. The system according to any one of claims 8 to 11, configured as described above.

13. The aforementioned at least one processor is Based on the aforementioned at least one EM feature, obtain an ML-based classification model that is pre-trained to classify the subject's condition according to one or more health criteria, and In order to classify the state of the target patient according to one or more health criteria, the ML-based classification model is inferred based on at least one EM characteristic of the target patient. The system according to any one of claims 8 to 12, configured to classify the state of the target patient by the means thereof.

14. The aforementioned at least one processor is Obtain measurements of the concentration of biological substances in the subject from at least one biosensor. Based on the measured values, annotation data elements are generated, and the subjects are labeled according to one or more health criteria. The aforementioned annotation data elements are used as supervisory information to train the ML-based classification model. The system according to any one of claims 8 to 13, configured as described above.

15. The system according to claim 14, wherein the at least one biosensor is a glucometer adapted to measure the glucose level in the subject's blood, and the one or more health criteria are selected from a list consisting of elevated glucose levels in the subject's blood, decreased glucose levels, steady glucose levels, hyperglycemia, hypoglycemia, and normal glucose levels.

16. The system according to claim 15, wherein the at least one processor is further configured to communicate with an associated insulin pump to administer insulin based on the classification of the target patient's condition.

Images