A method and system for signal pattern recognition in a mobile smart cell health repair watch

Abstract

This application discloses a signal pattern recognition method and system for a mobile smart cell health repair watch, relating to the field of mobile smartwatch technology. The method includes the following steps: continuously acquiring and analyzing wrist micro-movement data to obtain a first index; continuously acquiring and analyzing baseline conductivity data collected by a skin conductance sensor to obtain a second index; using the first index with a value lower than a preset stability threshold and / or the second index with a value higher than a preset hydration threshold as contextual information; temporally correlating the contextual information with physiological signals synchronously acquired by the skin conductance sensor and a microfluidic biochip integrated on the watch band to obtain a marker signal indicating potential physical interference and / or false physiological interference; and adaptively adjusting the processing strategy for physiological signals based on the contextual information and the marker signal to suppress interference and improve the reliability of physiological signal recognition. This application can reduce the probability of false alarms or missed alarms.

Description

Technical Field

[0001] This application relates to the field of mobile smartwatch technology, and in particular to a method and system for signal pattern recognition in a mobile smart cell health repair watch. Background Technology

[0002] The mobile smart cell health repair watch continuously monitors the user's physiological signals through a skin conductance sensor and a microfluidic biochip integrated on the watchband. It aims to identify early characteristic patterns of health conditions such as chronic fatigue syndrome in order to provide health early warning.

[0003] However, in actual daily use, the chemical components contained in frequently used personal care products (such as hand cream and hand sanitizer) gradually accumulate on the surface of the sensor electrodes and chip sampling area. This can lead to non-physiological baseline drift and response distortion in the skin conductance signal, and cause deviations in the detection results of the microfluidic chip. This type of interference is persistent and cumulative, and is highly superimposed on real physiological changes in both the time and frequency domains.

[0004] Because the signal recognition models built into the devices are usually trained on relatively ideal "clean" data, they are unable to effectively distinguish between these morphologically simulated false signals caused by long-term chemical pollution and real early pathophysiological signals. This ultimately creates a dilemma for the system: either it misjudges the interference as a health risk, leading to frequent false alarms; or it raises the recognition threshold to reduce false alarms, which may result in missing real weak pathological signals and missing the opportunity for early intervention. Summary of the Invention

[0005] This application proposes a signal pattern recognition method and system for a mobile smart cell health repair watch, aiming to solve the technical problem that existing mobile smart cell health repair watches suffer from decreased reliability of physiological signal recognition due to interference from external chemical substances such as personal care products in complex daily environments, leading to false alarms or missed alarms.

[0006] In a first aspect, this application provides a method for signal pattern recognition in a mobile smart cell health repair watch, comprising the following steps: Based on the mechanical sensor integrated in the watch strap, the wrist micro-movement data generated by the user while wearing the watch are continuously collected and analyzed to obtain a first index for characterizing the stability of the skin contact between the skin conductance sensor and the skin. By continuously acquiring and analyzing baseline conductivity data collected by the skin conductance sensor integrated on the watch strap, a second index is obtained to characterize local skin hydration. The first index with a value lower than the preset stability judgment threshold and / or the second index with a value higher than the preset hydration judgment threshold are used as scenario information; The scenario information is temporally correlated with physiological signals synchronously collected by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference. Based on the context information and the labeled signal, the processing strategy for the physiological signal is adaptively adjusted to suppress interference and improve the reliability of physiological signal recognition.

[0007] According to some embodiments of this application, the step of continuously collecting and analyzing wrist micro-movement data generated by the user while wearing the watch, based on a mechanical sensor integrated within the watch strap, to obtain a first index characterizing the stability of the skin contact between the skin conductance sensor and the skin includes: Based on the mechanical sensor integrated into the watch strap, it continuously collects wrist micro-movement data generated by the user while wearing the watch; Based on an environmental vibration reference sensor installed within the rigid body of the watch, environmental vibration data is continuously acquired; The wrist micro-movement data is processed based on the environmental vibration data to remove environmental background noise and obtain denoised wrist micro-movement data. Feature analysis was performed on the denoised wrist micro-movement data to obtain a first index for characterizing the stability of the skin contact between the skin conductance sensor and the skin.

[0008] According to some embodiments of this application, the step of processing the wrist micro-motion data based on the environmental vibration data to remove environmental background noise and obtain denoised wrist micro-motion data includes: The environmental vibration data is used as a reference input, and the wrist micro-movement data is used as the main input. Both are input into the adaptive filter integrated in the watch. The environmental vibration data and the wrist micro-motion data are processed by the adaptive filter to obtain an estimate of the noise component in the wrist micro-motion data corresponding to the environmental vibration data. The noise component estimate is subtracted from the wrist micro-movement data to obtain the denoised wrist micro-movement data.

[0009] According to some embodiments of this application, the step of performing feature analysis on the denoised wrist micro-motion data to obtain a first index for characterizing the contact stability between the skin conductance sensor and the skin includes: Temporal and frequency domain features were extracted from the denoised wrist micro-movement data. Based on the time-domain and frequency-domain features, a first index is generated through weighted summation or a fuzzy logic system to quantify the stability of the skin contact between the electrodermal response sensor and the skin. The first index is configured to decrease when the value of the time-domain feature exceeds a preset time-domain stability range or the value of the frequency-domain feature exceeds a preset frequency-domain stability range.

[0010] According to some embodiments of this application, the step of continuously acquiring and analyzing baseline conductivity data collected by a skin conductance sensor integrated on the watch strap to obtain a second index for characterizing local skin hydration includes: The baseline conductivity data of the skin conductance sensor integrated on the watch strap is continuously acquired at a preset first sampling frequency. The conductivity fluctuation data of the skin conductance sensor within a preset frequency range is continuously acquired at a second sampling frequency higher than the first sampling frequency. Based on the baseline conductivity data and the conductivity fluctuation data, a second index is generated to characterize local skin hydration.

[0011] According to some embodiments of this application, the step of generating a second index for characterizing local skin hydration based on the baseline conductivity data and the conductivity fluctuation data includes: Extract the average value and rate of change of the baseline conductivity data over a preset time period; Extract the fluctuation amplitude and nonlinear characteristics from the conductivity fluctuation data; The average value, rate of change, fluctuation amplitude, and nonlinear characteristics are used as hydration characteristics, and the hydration characteristics are compared with the pre-stored skin hydration electrical characteristic model to obtain the comparison results. Based on the comparison results, a quantization value between 0 and 1 is generated through a fuzzy logic system or weighted summation calculation as a second index to characterize local skin hydration; the quantization value is configured to increase when the hydration feature indicates an increase in skin hydration.

[0012] According to some embodiments of this application, the step of temporally correlating the scenario information with physiological signals synchronously acquired by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference includes: The context information is compared with the timestamps carried by the physiological signals synchronously collected by the skin conductance response sensor and the microfluidic biochip integrated on the watch strap, and the context information and the physiological signals are aligned on the time axis to obtain the correlation result; the physiological signals include the skin conductance response signal collected by the skin conductance response sensor and the biochemical indicator signal collected by the microfluidic biochip; Based on the correlation results, the signal artifacts in the physiological signals that correspond to the context information are marked to obtain a marked signal used to indicate potential physical interference and / or pseudo-physiological interference.

[0013] According to some embodiments of this application, the step of marking signal artifacts in the physiological signal corresponding to the context information based on the correlation result to obtain a marked signal for indicating potential physical interference and / or pseudo-physiological interference includes: Based on the correlation results, within the time period corresponding to the scenario information, signal spikes or rapid drifts in the skin conductance response signal with instantaneous change rate exceeding a preset physiological change rate threshold are identified, as well as abnormal fluctuations in the biochemical indicator signal with reading amplitude changes exceeding a preset biochemical change threshold are identified. The signal spikes or rapid drifts and the abnormal fluctuations are marked to obtain a marked signal for indicating potential physical interference and / or pseudo-physiological interference.

[0014] According to some embodiments of this application, the step of adaptively adjusting the processing strategy for the physiological signal based on the context information and the labeled signal to suppress interference and improve the reliability of physiological signal recognition includes: Based on the context information and the marker signal, the type of interference indicated by the marker signal is determined; the type of interference includes potential physical interference and pseudo-physiological interference. Based on the type of interference, the processing strategy for the physiological signal is adaptively adjusted to suppress interference and improve the reliability of physiological signal recognition. The processing strategy includes: for physiological signals marked as potential physical interference, triggering and switching to an adaptive filtering algorithm to process the physiological signal; for physiological signals marked as false physiological interference, reducing the weight of the physiological signal in subsequent pattern recognition.

[0015] Secondly, this application also provides a mobile smart cell health repair watch signal pattern recognition system, comprising: The first index acquisition module is used to continuously collect and analyze wrist micro-movement data generated by the user while wearing the watch based on the mechanical sensor integrated in the watch strap, and obtain a first index to characterize the stability of the skin contact between the skin conductance sensor and the skin. The second index acquisition module is used to continuously acquire and analyze the baseline conductivity data collected by the skin conductance sensor integrated on the watch strap to obtain a second index for characterizing local skin hydration. The scenario information generation module is used to take a first index whose value is lower than a preset stability judgment threshold and / or a second index whose value is higher than a preset hydration judgment threshold as scenario information. The signal correlation module is used to temporally correlate the scenario information with the physiological signals synchronously collected by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference. The processing strategy adjustment module is used to adaptively adjust the processing strategy for the physiological signal based on the context information and the labeled signal, so as to suppress interference and improve the reliability of physiological signal recognition.

[0016] According to the technical solution of the embodiments of this application, it has at least the following beneficial effects: This application effectively solves the problem of decreased reliability of physiological signal recognition due to interference from external chemical substances such as personal care products in the prior art by constructing an innovative framework of multi-dimensional scene perception, accurate interference marking and adaptive processing strategy. It significantly improves the monitoring accuracy and reliability of mobile smart cell health repair watch in daily complex environment, reduces the probability of false alarm or missed alarm, and provides users with more accurate and reliable health management services.

[0017] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0018] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0019] Figure 1 This is a flowchart illustrating a signal pattern recognition method for a mobile smart cell health repair watch, as provided in an embodiment of this application.

[0020] Figure 2 This is a schematic diagram of the architecture of a mobile intelligent cell health repair watch signal pattern recognition system provided in an embodiment of this application. Detailed Implementation

[0021] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0022] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] Existing mobile smart cell health repair watches face a dilemma of false alarms and missed alarms because they cannot effectively distinguish between persistent signal interference caused by chemical residues from personal care products and real early pathophysiological signals.

[0024] In this regard, such as Figure 1 As shown, this application discloses a signal pattern recognition method for a mobile smart cell health repair watch, including the following steps: S110, based on a mechanical sensor integrated in the watch strap, continuously collects and analyzes wrist micro-movement data generated by the user while wearing the watch, and obtains a first index to characterize the stability of the skin contact between the skin conductance sensor and the skin. S120 continuously acquires and analyzes baseline conductivity data collected by the skin conductance sensor integrated on the watch strap to obtain a second index for characterizing local skin hydration. S130, take the first index whose value is lower than the preset stability judgment threshold and / or the second index whose value is higher than the preset hydration judgment threshold as scenario information; S140, the scenario information is temporally correlated with the physiological signals synchronously collected by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference. S150, based on the context information and the labeled signal, adaptively adjust the processing strategy for the physiological signal to suppress interference and improve the reliability of physiological signal recognition.

[0025] The key terms and implementation environments involved will now be explained.

[0026] The "Mobile Smart Cell Health Repair Watch" is a wearable device whose main function is to continuously monitor the user's physiological health, especially physiological signals related to chronic fatigue syndrome. The watch typically consists of a rigid body and a flexible strap, the inner side of which integrates various sensors, such as mechanical sensors, skin conductance sensors, and microfluidic biochips.

[0027] "Mechanical sensors" typically refer to inertial measurement units such as accelerometers and gyroscopes, used to capture minute movements of the wrist, which may reflect the stability of the sensor's contact with the skin. "Skin conductance sensors" are used to measure skin conductivity, the changes of which are closely related to sweat gland activity and the state of the autonomic nervous system. "Microfluidic biochips" are miniaturized laboratory chips capable of real-time or near-real-time analysis of biomarkers in sweat or other bodily fluids.

[0028] "Wrist micro-motion data" refers to the data on the subtle movements of the wrist during daily activities, collected by mechanical sensors. "Baseline conductivity data" refers to the skin conductivity values ​​measured by the skin conductance sensor under relatively stable conditions, reflecting the skin's fundamental electrical properties. "Physiological signals" refer to the skin conductance signals collected by the skin conductance sensor and the biochemical indicator signals collected by the microfluidic biochip; these signals are key data for assessing the user's health status.

[0029] "Contextual information" is a core concept introduced in this application. It combines the stability of the sensor's contact with the skin and the local skin hydration to determine whether the current physiological signal acquisition environment is interfered with. "Labeled signal" is the result of labeling possible artifacts or abnormalities in the physiological signal based on the contextual information, aiming to distinguish between real physiological changes and false changes caused by interference.

[0030] The implementation environment of this application is typically a mobile smart cell health repair watch worn by the user in daily life, with the watch continuously running monitoring and analysis programs in the background. All data collection, analysis, and processing are completed either internally within the watch or in collaboration with external devices such as smartphones.

[0031] This application proposes a signal pattern recognition method for a mobile smart cell health repair watch. Firstly, based on a mechanical sensor integrated within the watch band, it continuously collects and analyzes wrist micro-movement data generated by the user while wearing the watch to obtain a first index characterizing the contact stability between the skin conductance sensor and the skin. For example, the mechanical sensor can be a triaxial accelerometer, capable of recording real-time acceleration changes of the wrist in the X, Y, and Z directions. By analyzing this acceleration data, such as calculating its root mean square value or standard deviation, the intensity of wrist activity can be quantified. When wrist activity is vigorous, the contact stability between the sensor and the skin may decrease, leading to a decrease in the value of the first index. As an alternative implementation, the mechanical sensor can also be a piezoelectric thin-film sensor integrated within the watch band. When the pressure between the watch band and the skin changes, the piezoelectric thin film generates an electrical signal. By analyzing the amplitude or frequency of these electrical signals, the degree of adhesion between the sensor and the skin can be assessed, thereby generating the first index.

[0032] Baseline conductivity data collected by a skin conductance sensor integrated into the watch band is continuously acquired and analyzed to derive a second index characterizing local skin hydration. For example, the skin conductance sensor can be an electrode array consisting of two or more electrodes. By applying a weak, constant voltage or current to the skin and measuring the corresponding current or voltage response, the skin's conductivity value can be obtained. Baseline conductivity data typically refers to the average or trend of skin conductivity over a period of time. When skin hydration is high, the skin conductivity value usually increases. Alternatively, the skin conductance sensor can also characterize hydration by measuring the skin's dielectric constant, as the water content in the skin directly affects its dielectric properties. By analyzing these conductivity or dielectric constant data, a quantified second index can be generated; the higher the index value, the higher the local skin hydration.

[0033] A first index with a value below a preset stability threshold and / or a second index with a value above a preset hydration threshold are used as context information. If the first index is below the stability threshold (e.g., 0.5, range 0-1), it indicates unstable sensor-skin contact; or if the second index is above the hydration threshold (e.g., 0.8, range 0-1), it indicates excessive local skin hydration. Both of these situations may indicate the presence of external interference. These identified unstable or high hydration states are defined as context information.

[0034] The contextual information is temporally correlated with physiological signals synchronously collected by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain marker signals indicating potential physical interference and / or pseudo-physiological interference. For example, if the contextual information indicates unstable sensor contact during a certain time period, abnormal spikes or drifts in the skin conductance signals collected during that time period may be marked as potential physical interference. Similarly, if the contextual information indicates excessive local skin hydration, unexpected increases or decreases in biochemical indicators collected by the microfluidic biochip during that time period may be marked as pseudo-physiological interference. This temporal correlation can be achieved by comparing the timestamps of the contextual information and physiological signals to ensure that interference information accurately corresponds to the affected physiological signals.

[0035] Based on the contextual information and the labeled signals, the system adaptively adjusts the processing strategy for the physiological signals to suppress interference and improve the reliability of physiological signal recognition. For example, if the labeled signals indicate potential physical interference (such as poor sensor contact), the system can trigger an adaptive filtering algorithm to denoise the affected skin conductance response signals to eliminate artifacts caused by physical motion. If the labeled signals indicate false physiological interference (such as abnormal biochemical indicators due to excessive skin hydration), the system can reduce the weight of biochemical indicator signals in subsequent pattern recognition during that time period, or exclude them from the analysis, thereby avoiding misjudgments caused by external chemical substances. In this way, the system can flexibly adjust the signal processing method according to the specific type and degree of interference, ensuring that the final physiological signal analysis results are more accurate and reliable.

[0036] The overall working principle of this application is to construct a multi-level, adaptive physiological signal recognition framework to effectively cope with various interferences that mobile smart cell health repair watches may encounter in actual wearing environments.

[0037] This application, by introducing a first index and a second index, can assess the stability of sensor-skin contact and local skin hydration in real time, thereby forming contextual information and providing crucial context for subsequent interference identification. Furthermore, this application temporally correlates the contextual information with physiological signals and generates labeled signals, enabling the identification of specific artifacts or anomalies appearing in physiological signals under specific interference scenarios. Based on the contextual information and labeled signals, this application adaptively adjusts the processing strategy for physiological signals. This adaptive processing method allows this application to effectively suppress interference and improve the reliability of physiological signal identification, significantly outperforming the "one-size-fits-all" signal processing methods in the prior art.

[0038] In summary, this application effectively solves the problem of decreased reliability of physiological signal recognition due to interference from external chemical substances such as personal care products in existing technologies by constructing an innovative framework of multi-dimensional context perception, precise interference labeling, and adaptive processing strategies. It improves the monitoring accuracy and reliability of mobile smart cell health repair watches in complex daily environments, reduces the probability of false alarms or missed alarms, and provides users with more accurate and reliable health management services.

[0039] In some embodiments of this application, the step of continuously collecting and analyzing wrist micro-movement data generated by the user while wearing the watch, based on a mechanical sensor integrated within the watch strap, to obtain a first index characterizing the stability of the skin contact between the skin conductance sensor and the skin preferably includes: Based on the mechanical sensor integrated into the watch strap, it continuously collects wrist micro-movement data generated by the user while wearing the watch; Based on an environmental vibration reference sensor installed within the rigid body of the watch, environmental vibration data is continuously acquired; The wrist micro-movement data is processed based on the environmental vibration data to remove environmental background noise and obtain denoised wrist micro-movement data. Feature analysis was performed on the denoised wrist micro-movement data to obtain a first index for characterizing the stability of the skin contact between the skin conductance sensor and the skin.

[0040] The wrist micro-motion data refers to the mechanical vibration or displacement signals generated by the user's subtle wrist movements, posture changes, or alterations in contact with the sensor while wearing the watch. This data is continuously collected by mechanical sensors integrated into the watch band, such as accelerometers, gyroscopes, or piezoelectric sensors. The environmental vibration reference sensor is typically housed within the rigid body of the watch and its purpose is to acquire vibration information from the surrounding environment independently of wrist micro-motions, serving as a reference for background noise. This sensor can be another accelerometer or vibration sensor, its location and installation method dictating that it primarily responds to environmental vibrations rather than the fine movements of the user's wrist.

[0041] The wrist micro-movement data is processed based on the acquired environmental vibration data to effectively separate and remove environmental background noise from the raw data. This processing can employ various signal processing techniques, such as adaptive filtering, wavelet denoising, or frequency domain filtering, with the core objective of identifying and eliminating noise components related to the environmental vibration data. This results in cleaner, more accurate denoised wrist micro-movement data that reflects the user's wrist micro-movement characteristics.

[0042] Feature analysis of the denoised wrist micro-movement data involves extracting quantitative features related to the stability of the skin conductance sensor in contact with the skin from this purified data. These features can include time-domain features (such as root mean square, variance, and peak factor) and frequency-domain features (such as energy and spectral centroid of a specific frequency band). A comprehensive analysis of these features generates a first quantitative index that accurately characterizes the stability of the skin conductance sensor in contact with the skin.

[0043] This application's solution effectively addresses the problem of wrist micro-motion data being easily interfered with by environmental noise in traditional methods by introducing an environmental vibration reference sensor and denoising the acquired wrist micro-motion data. By using environmental vibration data as a reference, signal processing techniques can be used to accurately estimate and remove the environmental background noise component from the wrist micro-motion data. Because the environmental background noise is effectively removed, subsequent feature analysis of the denoised wrist micro-motion data can more accurately reflect the user's true wrist micro-movements and the contact stability between the skin conductance sensor and the skin. This method ensures that the calculation of the first index is based on more reliable input data, thereby improving its representation accuracy.

[0044] In a further embodiment of this application, the step of processing the wrist micro-motion data based on environmental vibration data to remove environmental background noise and obtain denoised wrist micro-motion data preferably includes: The environmental vibration data is used as a reference input, and the wrist micro-movement data is used as the main input. Both are input into the adaptive filter integrated in the watch. The environmental vibration data and the wrist micro-motion data are processed by the adaptive filter to obtain an estimate of the noise component in the wrist micro-motion data corresponding to the environmental vibration data. The noise component estimate is subtracted from the wrist micro-movement data to obtain the denoised wrist micro-movement data.

[0045] The environmental vibration data refers to information characterizing the mechanical vibrations of the surrounding environment, continuously acquired by an environmental vibration reference sensor installed within the rigid body of the watch. The wrist micro-motion data refers to the subtle movement data of the user's wrist continuously collected by a mechanical sensor integrated within the watch strap. Using the environmental vibration data as a reference input means treating it as a representation of a known noise source, used to guide the adaptive filter in identifying and eliminating similar noise components in the main input. Using the wrist micro-motion data as the main input means processing it as a mixed signal containing the target signal and noise.

[0046] The adaptive filter is a digital filter that can automatically adjust its parameters based on the statistical characteristics of the input signal. Its core lies in continuously adjusting the filter coefficients through an iterative algorithm (such as the Least Mean Square (LMS) algorithm or the Recursive Least Squares (RLS) algorithm) to minimize the power of the output error signal. In this embodiment, the adaptive filter is configured to receive the environmental vibration data as a reference input and attempt to predict and subtract noise components related to the environmental vibration from the wrist micro-motion data.

[0047] The noise component estimation refers to the noise component caused by environmental vibration in the wrist micro-motion data, which is estimated in real time by the adaptive filter based on the correlation between the reference input (the environmental vibration data) and the main input (the wrist micro-motion data). This estimate is dynamically adjusted and can reflect real-time changes in environmental noise.

[0048] Subtracting the noise component estimate from the wrist micro-motion data means removing the noise component estimated by the adaptive filter from the original wrist micro-motion data, thereby obtaining a cleaner and more accurate denoised wrist micro-motion data. The purpose is to retain as much effective information related to the skin conductance sensor's contact stability with the skin as possible while eliminating environmental interference.

[0049] Through the above technical solution, this application can significantly improve the accuracy and robustness of wrist micro-motion data denoising. Compared with the fixed parameter filtering method, the adaptive filter can better adapt to dynamically changing environmental noise, effectively avoiding the problems of incomplete or excessive noise stripping. Therefore, cleaner and more reliable wrist micro-motion data can be obtained, enabling the first index calculated based on this data to more accurately characterize the stability of the skin conductance sensor's contact with the skin. This provides a more solid foundation for subsequent physiological signal recognition and interference suppression, thereby improving the reliability of the entire signal pattern recognition method.

[0050] In a further embodiment of this application, the step of performing feature analysis on the denoised wrist micro-motion data to obtain a first index for characterizing the contact stability between the skin conductance sensor and the skin preferably includes: Temporal and frequency domain features were extracted from the denoised wrist micro-movement data. Based on the time-domain and frequency-domain features, a first index is generated through weighted summation or a fuzzy logic system to quantify the stability of the skin contact between the electrodermal response sensor and the skin. The first index is configured to decrease when the value of the time-domain feature exceeds a preset time-domain stability range or the value of the frequency-domain feature exceeds a preset frequency-domain stability range.

[0051] In practical applications, various time-domain and frequency-domain features can be extracted from denoised wrist micro-movement data. Time-domain features include, but are not limited to, mean, variance, standard deviation, peak value, root mean square value, zero-crossing rate, kurtosis, and skewness. These features aim to describe the amplitude, energy, and variation characteristics of wrist micro-movement data along the time axis. Frequency-domain features, such as dominant frequency, bandwidth, energy distribution, and spectral entropy, can be obtained from time-domain data through signal processing methods like Fourier transform. These features reveal the periodicity, vibration modes, and energy distribution of wrist micro-movement data along the frequency axis. The extraction of these multi-dimensional features aims to comprehensively capture various manifestations of wrist micro-movements, thus providing a rich data foundation for accurately characterizing the stability of skin contact between the electrodermal response sensor and the skin.

[0052] When generating the first quantized index, a weighted summation or a fuzzy logic system can be used. Using a weighted summation method, multiple extracted time-domain and frequency-domain features are assigned different weights, and then the weighted feature values ​​are summed to obtain a comprehensive quantized value as the first index. The weights can be set based on extensive experimental data, empirical knowledge, or optimized through machine learning algorithms to ensure that the first index more accurately reflects the actual contact stability. As another implementation method, a fuzzy logic system can handle uncertainty and fuzzy information. It takes time-domain and frequency-domain features as input, and through preset fuzzy rules and fuzzy inference mechanisms, outputs a quantized value within a specific range (e.g., 0 to 1) as the first index. For example, fuzzy variables such as "large micro-motion amplitude" and "rapid frequency change" can be defined, and corresponding fuzzy rules can be set to judge contact stability. Generally, a higher first index value indicates more stable contact, while a lower value indicates less stable contact.

[0053] The time-domain stability range refers to the normal fluctuation range of characteristics such as amplitude and rate of change allowed in wrist micro-movement data within the time domain. For example, if the root mean square value or instantaneous rate of change of wrist micro-movement exceeds a preset threshold, it may indicate that the sensor has slipped or disengaged. Similarly, the frequency-domain stability range refers to the normal fluctuation range of characteristics such as energy distribution and dominant frequency allowed in wrist micro-movement data within the frequency domain. For example, an abnormal increase in high-frequency components may indicate severe shaking or friction. This configuration ensures that when any abnormality that may lead to contact instability is detected, the first index can reflect this instability state promptly and accurately, thus providing a reliable basis for subsequent physiological signal processing.

[0054] Through the above technical solution, this application can more accurately and robustly evaluate the stability of the skin contact between the skin conductance sensor and the skin. Compared with only general feature analysis, this application significantly improves the representation accuracy of the first index by extracting multi-dimensional time-domain and frequency-domain features and combining them with weighted summation or fuzzy logic systems for quantification. In particular, when the time-domain or frequency-domain features in the wrist micro-movement data exceed the preset stability range, the value of the first index is actively reduced, which enables the system to identify potential poor contact or unstable states in a timely and sensitive manner. This refined stability assessment mechanism provides a more reliable foundation for the accurate identification and interference suppression of subsequent physiological signals, thereby effectively improving the reliability of the overall signal pattern recognition of the mobile smart cell health repair watch.

[0055] The following is a specific example to illustrate this.

[0056] Suppose that while wearing a mobile smart cell health repair watch, a user experiences a violent wrist tremor. Mechanical sensors integrated into the watch band continuously collect wrist movement data, which is then processed by an environmental vibration reference sensor to obtain denoised wrist movement data. The system then extracts time-domain features, such as root mean square (RMS) and peak values, and frequency-domain features, such as dominant frequency and energy distribution, from this denoised data. If the RMS value of the wrist movement data suddenly exceeds a preset upper limit of the time-domain stability range, or if abnormal high-frequency energy concentration appears in its spectrum, according to the configuration of this application, even if other features are normal, the system will immediately reduce the value of the generated first index. For example, the first index, which originally indicated stable contact, might drop from 0.9 to 0.3, clearly indicating poor contact stability between the current skin conductance sensor and the skin. This reduced first index is then used as part of the contextual information to guide adjustments to subsequent physiological signal processing strategies, such as triggering stronger filtering algorithms or reducing the weight of physiological signals during that time period, to effectively suppress signal artifacts or pseudo-physiological interference caused by unstable contact.

[0057] In a specific embodiment of this application, the step of continuously acquiring and analyzing baseline conductivity data collected by the skin conductance sensor integrated on the watch strap to obtain a second index for characterizing local skin hydration preferably includes: The baseline conductivity data of the skin conductance sensor integrated on the watch strap is continuously acquired at a preset first sampling frequency. The conductivity fluctuation data of the skin conductance sensor within a preset frequency range is continuously acquired at a second sampling frequency higher than the first sampling frequency. Based on the baseline conductivity data and the conductivity fluctuation data, a second index is generated to characterize local skin hydration.

[0058] The first sampling frequency refers to the frequency at which the baseline conductivity value output by the skin conductance sensor is periodically collected. Its purpose is to obtain long-term or relatively stable trend information of skin conductance. For example, this first sampling frequency can be set to a low frequency, such as 1-5 times per second, to effectively capture macroscopic changes in skin conductance. The second sampling frequency refers to the frequency at which the conductance fluctuation data of the skin conductance sensor is collected within a specific frequency range. This frequency is configured to be higher than the first sampling frequency. Its purpose is to capture rapid, transient changes in skin conductance, which are often related to the skin's microscopic physiological activities or rapid responses to external stimuli, and can more precisely reflect the dynamic characteristics of skin hydration. For example, the second sampling frequency can be set to 10-50 times per second to ensure effective capture of high-frequency conductance fluctuations. In practical applications, baseline conductance data can be understood as the skin's ability to conduct current under relatively stable conditions, reflecting the water content of the stratum corneum. Conductance fluctuation data, on the other hand, refers to rapid, small-amplitude changes in conductance values ​​caused by various physiological activities (such as sweat gland activity, vasomotor activity, etc.) or external disturbances, based on the baseline conductance. By combining these two types of data, skin hydration can be comprehensively assessed from both macroscopic and microscopic perspectives.

[0059] This application's solution, by combining baseline conductivity data and conductivity fluctuation data at different sampling frequencies, can capture multi-dimensional information about skin hydration more comprehensively and precisely, thereby significantly improving the accuracy and reliability of the second index used to characterize local skin hydration. This multi-frequency, multi-dimensional data acquisition and analysis method makes the assessment of skin hydration more accurate, providing more reliable contextual information for subsequent physiological signal pattern recognition, and thus improving the overall reliability of signal recognition.

[0060] In a more specific embodiment of this application, the step of generating a second index for characterizing local skin hydration based on the baseline conductivity data and the conductivity fluctuation data preferably includes: Extract the average value and rate of change of the baseline conductivity data over a preset time period; Extract the fluctuation amplitude and nonlinear characteristics from the conductivity fluctuation data; The average value, rate of change, fluctuation amplitude, and nonlinear characteristics are used as hydration characteristics, and the hydration characteristics are compared with the pre-stored skin hydration electrical characteristic model to obtain the comparison results. Based on the comparison results, a quantization value between 0 and 1 is generated through a fuzzy logic system or weighted summation calculation as a second index to characterize local skin hydration; the quantization value is configured to increase when the hydration feature indicates an increase in skin hydration.

[0061] In practical applications, baseline conductivity data reflects the overall conductivity level of the skin. Its average value provides a macroscopic indication of skin hydration, while the rate of change captures the trend of hydration changing over time. The preset duration can be set according to the actual application scenario and data sampling frequency, for example, it can be from several minutes to several hours.

[0062] Electrical conductivity fluctuation data, especially those acquired at higher sampling frequencies, can reveal dynamic changes in skin microstructure and physiological activity. Fluctuation amplitude can reflect the intensity of skin's response to stimulation or the activity of internal physiological processes, while nonlinear features, such as fractal dimension and Lyapunov exponent, can more deeply characterize the complexity and dynamics of skin electrical conductivity changes. These features are closely related to the skin's micro-hydration state.

[0063] The average value, rate of change, fluctuation amplitude, and nonlinear characteristics are used as hydration features. These extracted features collectively constitute a multi-dimensional index set for a comprehensive assessment of local skin hydration. The hydration features are compared with a pre-stored skin hydration electrical characteristic model to obtain the comparison results. This pre-stored model is based on extensive experimental data and physiological knowledge, and includes typical patterns of various electrical characteristics under different hydration states. By comparing the currently collected hydration features with this model, the similarity or difference between the current skin hydration and known states can be quantified.

[0064] Based on the comparison results, a quantized value between 0 and 1 is generated using a fuzzy logic system or weighted summation calculation as a second index to characterize local skin hydration. The fuzzy logic system can handle uncertainty and fuzziness, mapping multiple input features (comparison results) to a continuous output value (second index), making it particularly suitable for complex assessments of physiological signals. Weighted summation calculation, by assigning different weights to different features, comprehensively reflects their contribution to hydration. This quantized value provides a standardized and easily understood index of skin hydration. The quantized value is configured to increase when the hydration feature indicates increased skin hydration; this configuration ensures the intuitiveness of the second index—a higher value represents higher skin hydration—consistent with general understanding and facilitating subsequent system judgment and processing.

[0065] This application's solution comprehensively captures multi-scale information about local skin hydration by extracting macroscopic average values ​​and rates of change from baseline conductivity data, and microscopic fluctuation amplitudes and nonlinear characteristics from high-frequency conductivity fluctuation data. These multi-dimensional hydration features are then input into a pre-stored skin hydration electrical characteristic model for comparison to objectively assess the current skin hydration status. The comparison results are transformed into a quantified second index between 0 and 1 using a fuzzy logic system or weighted summation calculation. This comprehensive feature extraction and intelligent evaluation mechanism effectively overcomes the limitations of single indicators or simple data combinations in reflecting skin hydration, thus providing a more accurate and robust characterization of skin hydration.

[0066] The following is a specific example to illustrate this.

[0067] Assume that the skin conductance sensor continuously acquires baseline conductance data at a first sampling frequency of 1 Hz, and acquires conductance fluctuation data in the frequency range of 0.1 Hz to 10 Hz at a second sampling frequency of 100 Hz.

[0068] First, the mean and rate of change of the baseline conductance data from the most recent 5 minutes are calculated. Simultaneously, the maximum fluctuation amplitude and nonlinear features based on wavelet transform, such as approximate entropy, are extracted from the conductance fluctuation data from the most recent 1 minute.

[0069] Subsequently, these four features (average value, rate of change, fluctuation amplitude, and nonlinear feature) are input into a pre-trained fuzzy logic system. This fuzzy logic system contains multiple fuzzy rules, such as: "If the average value is high, the rate of change is low, the fluctuation amplitude is small, and the nonlinear feature value is low, then the degree of hydration is high"; "If the average value is low, the rate of change is high, the fluctuation amplitude is large, and the nonlinear feature value is high, then the degree of hydration is low".

[0070] Based on these rules, the fuzzy logic system will output a quantized value between 0 and 1. For example, when skin hydration is normal, this value might be around 0.5; when skin hydration is significantly increased, such as with excessive sweating, this value might be close to 1; and when skin hydration is significantly decreased, such as with dry skin, this value might be close to 0. This quantized value serves as a second index characterizing local skin hydration, used for subsequent contextual information judgment.

[0071] In a preferred embodiment of this application, the step of temporally correlating the scenario information with physiological signals synchronously acquired by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference includes: The context information is compared with the timestamps carried by the physiological signals synchronously collected by the skin conductance response sensor and the microfluidic biochip integrated on the watch strap, and the context information and the physiological signals are aligned on the time axis to obtain the correlation result; the physiological signals include the skin conductance response signal collected by the skin conductance response sensor and the biochemical indicator signal collected by the microfluidic biochip; Based on the correlation results, the signal artifacts in the physiological signals that correspond to the context information are marked to obtain a marked signal used to indicate potential physical interference and / or pseudo-physiological interference.

[0072] Contextual information refers to information formed by the first and / or second indices, used to indicate the user's wearing status of the watch or local skin hydration. Physiological signals refer to skin conductance response signals collected by the skin conductance response sensor and biochemical indicator signals collected by the microfluidic biochip. These signals are timestamped during acquisition to record the time of their generation. By comparing these timestamps, contextual information and physiological signals can be precisely aligned on the timeline, thereby establishing a temporal correlation between them. Through this correlation, signal artifacts that may exist in the physiological signals within the specific time period indicated by the contextual information can be identified. Signal artifacts refer to signal anomalies caused by external interference or non-physiological factors, such as signal distortion caused by unstable wear or changes in skin hydration. Marking these signal artifacts can clearly indicate potential physical interference or pseudo-physiological interference, providing a basis for subsequent signal processing.

[0073] This application's solution first compares and aligns the timestamps of contextual information and physiological signals, ensuring their synchronization in the temporal dimension. This precise temporal correlation enables the system to identify anomalies in physiological signals under specific circumstances, such as artifacts that may appear when wrist micro-motion data indicates unstable wear or baseline conductivity data indicates abnormal skin hydration. By marking these signal anomalies corresponding to contextual information as potential physical interference or pseudo-physiological interference, this application can effectively distinguish between genuine physiological responses and interference caused by external factors, thus providing accurate interference indications for subsequent physiological signal processing.

[0074] In a more preferred embodiment of this application, the step of marking signal artifacts in the physiological signals corresponding to contextual information based on the correlation results to obtain marked signals for indicating potential physical interference and / or pseudo-physiological interference includes: Based on the correlation results, within the time period corresponding to the scenario information, signal spikes or rapid drifts in the skin conductance response signal with instantaneous change rate exceeding a preset physiological change rate threshold are identified, as well as abnormal fluctuations in the biochemical indicator signal with reading amplitude changes exceeding a preset biochemical change threshold are identified. The signal spikes or rapid drifts and the abnormal fluctuations are marked to obtain a marked signal for indicating potential physical interference and / or pseudo-physiological interference.

[0075] Within the time period corresponding to the scenario information, the system is configured to identify specific abnormal patterns in the skin conductance response signal. A "signal spike" typically refers to a significant, instantaneous increase or decrease in the skin conductance response signal within a very short period, with its instantaneous rate of change significantly exceeding a preset physiological rate of change threshold. This spike may be caused by physical interference such as sudden, large movements of the wearer's wrist or a momentary interruption or resumption of sensor contact with the skin. "Rapid drift" refers to a sustained, non-physiological, rapid increase or decrease in the skin conductance response signal over a period of time, with its rate of change also exceeding the preset physiological rate of change threshold. This may be related to slight slippage of the sensor position or rapid changes in the local skin environment. By setting the physiological rate of change threshold, normal physiological fluctuations and abnormal changes caused by interference can be effectively distinguished.

[0076] Simultaneously, the system is configured to identify "abnormal fluctuations" in biochemical indicator signals. Specifically, this refers to changes in the amplitude of biochemical indicator signal readings exceeding a preset biochemical change threshold within the time period corresponding to the scenario information. For example, when a microfluidic biochip collects biochemical indicators, factors such as bubbles, fluid blockage, reagent depletion, or external pressure may cause unexpected sudden increases or decreases in readings, or a sustained deviation from the baseline. By setting a biochemical change threshold, these non-physiological abnormal readings, indicating potential pseudo-physiological interference, can be accurately captured.

[0077] Once the system identifies the aforementioned signal spikes, rapid drifts, or abnormal fluctuations, it will mark these specific signal segments. This marking signal clearly indicates that, at a specific point in time or within a specific time period, the physiological signal may have been affected by potential physical interference or pseudo-physiological interference.

[0078] Compared to basic solutions that rely solely on general labeling based on contextual information, this approach introduces specific signal feature recognition (such as instantaneous rate of change and reading amplitude variations) to more precisely distinguish between genuine physiological changes and artifacts caused by external factors. This not only avoids misjudging normal physiological signals but also ensures the effective capture of key interference events, thus providing more accurate and reliable input for subsequent adaptive adjustments to physiological signal processing strategies. Ultimately, this improves the overall performance and user experience of the signal pattern recognition in the mobile smart cell health repair watch.

[0079] The following example illustrates this.

[0080] Suppose that when a user is wearing a smartwatch, a sudden raising or lowering of their arm causes a violent wrist movement detected by the mechanical sensors within the watchband. This generates a first index below a preset stability threshold, which is identified as contextual information. Simultaneously, the skin conductance response signal collected by the skin conductance sensor may exhibit a signal spike with an instantaneous rate of change far exceeding a preset physiological rate of change threshold (e.g., more than 5 microSiemens per second). According to this scheme, the system identifies this signal spike based on the correlation between contextual information and physiological signals, and marks it as a marker signal indicating potential physical interference.

[0081] For example, in another scenario, after strenuous exercise, the user's local skin hydration increases, causing the second index to exceed a preset hydration threshold. This second index is identified as contextual information. Simultaneously, when the microfluidic biochip integrated into the watch band collects a certain biochemical indicator, sweat entering the microfluidic channel causes continuous abnormal fluctuations in the reading amplitude of the biochemical indicator signal, exceeding a preset biochemical change threshold (e.g., a change of more than 20% within 30 seconds). This solution identifies and marks these abnormal fluctuations as marker signals indicating pseudo-physiological interference. In this way, the system can accurately identify and mark specific signal artifacts corresponding to contextual information, providing precise guidance for subsequent interference suppression and signal processing.

[0082] In some embodiments of this application, the step of adaptively adjusting the processing strategy for physiological signals based on context information and labeled signals to suppress interference and improve the reliability of physiological signal recognition preferably includes: Based on the context information and the marker signal, the type of interference indicated by the marker signal is determined; the type of interference includes potential physical interference and pseudo-physiological interference. Based on the type of interference, the processing strategy for the physiological signal is adaptively adjusted to suppress interference and improve the reliability of physiological signal recognition. The processing strategy includes: for physiological signals marked as potential physical interference, triggering and switching to an adaptive filtering algorithm to process the physiological signal; for physiological signals marked as false physiological interference, reducing the weight of the physiological signal in subsequent pattern recognition.

[0083] Determining the type of interference indicated by the labeled signal refers to the system's judgment of the source and nature of the interference based on contextual information (such as wrist micro-movement data or local skin hydration data) and the characteristics of the labeled signal (such as spikes, drifts, or abnormal fluctuations in physiological signals). Potential physical interference typically refers to signal artifacts caused by external mechanical motion, poor sensor contact, or environmental noise. These interferences often manifest as instantaneous large fluctuations, spikes, or baseline drift in physiological signals. False physiological interference refers to signal changes caused by non-physiological factors (such as changes in skin hydration or environmental temperature) that may resemble genuine physiological responses. These changes may cause a sustained increase or decrease in the baseline of physiological signals, or slow fluctuations.

[0084] Adaptive adjustment of physiological signal processing strategies refers to the system dynamically selecting or adjusting the corresponding signal processing algorithm based on the determined type of interference. For physiological signals marked as potential physical interference, an adaptive filtering algorithm is triggered and switched to process the physiological signal. This adaptive filtering algorithm can be understood as an algorithm that can dynamically adjust filter parameters according to the statistical characteristics of the input signal, such as the Least Mean Square (LMS) algorithm, Recursive Least Squares (RLS) algorithm, or Kalman filtering. Its purpose is to effectively remove noise components related to physical interference while preserving the true information of the physiological signal to the greatest extent possible. For physiological signals marked as false physiological interference, the weight of the physiological signal in subsequent pattern recognition is reduced. The purpose is to reduce the contribution of these signals, which are affected by non-physiological factors, to the final pattern recognition result and avoid misjudgment. In practical applications, reducing the weight can be achieved by adjusting the confidence of features in the machine learning model, assigning lower priority in multimodal data fusion, or introducing a penalty factor in the decision logic.

[0085] This application's solution first distinguishes the types of interference based on contextual information and labeled signals, thereby enabling the selection of the most suitable processing strategy. Specifically, when potential physical interference is identified, such as high-frequency noise or transient spikes caused by wrist micro-movements, an adaptive filtering algorithm can effectively remove these external interference components from physiological signals. This is because the adaptive filter can dynamically learn and cancel noise, maximizing the recovery of the true physiological signal waveform. However, when pseudo-physiological interference is identified, such as baseline drift or abnormal conductivity caused by changes in local skin hydration, this interference is not simply noise but may simulate signal changes in physiological responses. In this case, forcibly applying filtering may inadvertently damage genuine physiological information. Therefore, by reducing the weight of such physiological signals in subsequent pattern recognition, misjudgments or inaccurate health assessments caused by pseudo-physiological interference can be effectively avoided, ensuring that the final pattern recognition results are more reliable and accurate. This differentiated processing strategy makes interference suppression more precise and avoids the potential side effects of a "one-size-fits-all" approach.

[0086] The following is a specific example to illustrate this.

[0087] Assuming a user is wearing a mobile smart cell health repair watch during daily activities, the system continuously collects physiological signals.

[0088] Scenario 1: The user suddenly performs a vigorous wrist movement, such as rapidly swinging their arm. At this moment, the mechanical sensor integrated into the watch band detects significant wrist movement, causing a rapid drop in the first index, which characterizes the stability of the skin contact between the sensor and the skin, and generating scenario information. Simultaneously, the skin response signal collected by the skin response sensor may exhibit large instantaneous spikes or rapid drifts, which are identified as potential physical interference. Based on this, the system determines that this type of interference is potential physical interference and immediately triggers an adaptive filtering algorithm to process the skin response signal, effectively removing motion artifacts caused by wrist movement and restoring the true skin response signal.

[0089] Scenario 2: When a user wears a watch for an extended period in a high-temperature environment, increased sweating and skin hydration occur. In this situation, the baseline conductivity data collected by the skin conductance sensor will show a consistently high conductivity value, causing the second index used to characterize local skin hydration to exceed a preset hydration threshold, generating scenario information. Simultaneously, the skin conductance signal may exhibit slow baseline drift or a persistently high conductivity state, which is flagged as pseudo-physiological interference. Based on this, the system identifies this interference as pseudo-physiological interference and reduces the weight of the skin conductance signal during this time period in subsequent physiological signal pattern recognition (e.g., emotional state recognition or cell health assessment) to avoid misleading recognition results due to non-physiological signal fluctuations caused by changes in skin hydration, thereby improving the overall accuracy of recognition.

[0090] like Figure 2 As shown, this application also discloses a mobile smart cell health repair watch signal pattern recognition system 200, comprising: The first index acquisition module 210 is used to continuously collect and analyze wrist micro-movement data generated by the user while wearing the watch based on the mechanical sensor integrated in the watch strap, and obtain a first index to characterize the stability of the skin contact between the skin conductance sensor and the skin. The second index acquisition module 220 is used to continuously acquire and analyze the baseline conductivity data collected by the skin conductance sensor integrated on the watch strap to obtain a second index for characterizing local skin hydration. The scenario information forming module 230 is used to take a first index whose value is lower than a preset stability judgment threshold and / or a second index whose value is higher than a preset hydration judgment threshold as scenario information. The signal association module 240 is used to temporally correlate the scenario information with the physiological signals synchronously collected by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference. The processing strategy adjustment module 250 is used to adaptively adjust the processing strategy for the physiological signal based on the context information and the labeled signal, so as to suppress interference and improve the reliability of physiological signal recognition.

[0091] The first index acquisition module 210 may include a miniature inertial measurement unit (IMU), such as a three-axis accelerometer and gyroscope, for real-time capture of subtle wrist movements in three-dimensional space. By preprocessing and feature extraction of these raw motion data, such as calculating their root mean square value, standard deviation, or energy within a specific frequency range, the intensity and stability of wrist activity can be quantified, thereby generating the first index.

[0092] The second index acquisition module 220 may include a high-precision conductivity measurement circuit that measures the skin's impedance or conductivity by applying a weak AC or DC signal to the skin. Baseline conductivity data typically refers to the average or trend of skin conductivity over a period of time. Skin conductivity values ​​usually increase when skin hydration is high.

[0093] The scenario information generation module 230 may include a processor and a storage unit, which stores preset stability judgment thresholds and hydration judgment thresholds. Upon receiving the first index and the second index, the processor performs a logical judgment. If the first index is lower than its threshold, or the second index is higher than its threshold, the corresponding scenario information is generated.

[0094] The signal correlation module 240 may include a timestamp comparison unit and a data alignment algorithm. All sensors and modules can share a unified system clock to ensure high consistency of data acquisition timestamps. When contextual information is generated, its timestamp is recorded, and the signal correlation module then retrieves physiological signal data with a similar timestamp and performs precise time window matching to ensure that interference information corresponds accurately to the affected physiological signal.

[0095] The processing strategy adjustment module 250 may include a decision logic unit and a library of various signal processing algorithms. Upon receiving context information and a labeled signal, the decision logic unit selects and triggers a corresponding processing strategy from the algorithm library based on the type of interference indicated by the labeled signal (e.g., potential physical interference or pseudo-physiological interference). For example, for physiological signals labeled as potential physical interference, this module may trigger algorithms such as Kalman filtering, wavelet denoising, or adaptive notch filtering for real-time denoising. For physiological signals labeled as pseudo-physiological interference, the processing strategy adjustment module 250 may employ strategies such as data interpolation, outlier removal, or reducing their contribution weight in subsequent pattern recognition models.

[0096] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0097] The foregoing has provided a detailed description of the preferred embodiments of this application. However, this application is not limited to the above-described embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application. All such equivalent modifications or substitutions are included within the scope defined in this application.

Claims

1. A mobile intelligent cell health repair watch signal pattern recognition method, characterized in that, Includes the following steps: Based on the mechanical sensor integrated in the watch strap, the wrist micro-movement data generated by the user while wearing the watch are continuously collected and analyzed to obtain a first index for characterizing the stability of the skin contact between the skin conductance sensor and the skin. By continuously acquiring and analyzing baseline conductivity data collected by the skin conductance sensor integrated on the watch strap, a second index is obtained to characterize local skin hydration. The first index with a value lower than the preset stability judgment threshold and / or the second index with a value higher than the preset hydration judgment threshold are used as scenario information; The scenario information is temporally correlated with physiological signals synchronously collected by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference. Based on the context information and the labeled signal, the processing strategy for the physiological signal is adaptively adjusted to suppress interference and improve the reliability of physiological signal recognition.

2. The mobile intelligent cell health repair watch signal pattern recognition method according to claim 1, characterized in that, The step of continuously collecting and analyzing wrist micro-movement data generated by the user while wearing the watch, based on a mechanical sensor integrated within the watch strap, to obtain a first index characterizing the stability of the skin contact between the skin conductance sensor and the skin includes: Based on the mechanical sensor integrated into the watch strap, it continuously collects wrist micro-movement data generated by the user while wearing the watch; Based on an environmental vibration reference sensor installed within the rigid body of the watch, environmental vibration data is continuously acquired; The wrist micro-movement data is processed based on the environmental vibration data to remove environmental background noise and obtain denoised wrist micro-movement data. Feature analysis was performed on the denoised wrist micro-movement data to obtain a first index for characterizing the stability of the skin contact between the skin conductance sensor and the skin.

3. The signal pattern recognition method for a mobile intelligent cell health repair watch according to claim 2, characterized in that, The step of processing the wrist micro-movement data based on the environmental vibration data to remove environmental background noise and obtain denoised wrist micro-movement data includes: The environmental vibration data is used as a reference input, and the wrist micro-movement data is used as the main input. Both are input into the adaptive filter integrated in the watch. The environmental vibration data and the wrist micro-motion data are processed by the adaptive filter to obtain an estimate of the noise component in the wrist micro-motion data corresponding to the environmental vibration data. The noise component estimate is subtracted from the wrist micro-movement data to obtain the denoised wrist micro-movement data.

4. The signal pattern recognition method for a mobile intelligent cell health repair watch according to claim 2, characterized in that, The step of performing feature analysis on the denoised wrist micro-movement data to obtain a first index for characterizing the contact stability between the skin conductance sensor and the skin includes: Temporal and frequency domain features were extracted from the denoised wrist micro-movement data. Based on the time-domain and frequency-domain features, a first index is generated through weighted summation or a fuzzy logic system to quantify the stability of the skin contact between the electrodermal response sensor and the skin. The first index is configured to decrease when the value of the time-domain feature exceeds a preset time-domain stability range or the value of the frequency-domain feature exceeds a preset frequency-domain stability range.

5. The signal pattern recognition method for a mobile intelligent cell health repair watch according to claim 1, characterized in that, The step of continuously acquiring and analyzing baseline conductivity data collected by the skin conductance sensor integrated on the watch strap to obtain a second index for characterizing local skin hydration includes: The baseline conductivity data of the skin conductance sensor integrated on the watch strap is continuously acquired at a preset first sampling frequency. The conductivity fluctuation data of the skin conductance sensor within a preset frequency range is continuously acquired at a second sampling frequency higher than the first sampling frequency. Based on the baseline conductivity data and the conductivity fluctuation data, a second index is generated to characterize local skin hydration.

6. The signal pattern recognition method for a mobile intelligent cell health repair watch according to claim 5, characterized in that, The step of generating a second index to characterize local skin hydration based on the baseline conductivity data and the conductivity fluctuation data includes: Extract the average value and rate of change of the baseline conductivity data over a preset time period; Extract the fluctuation amplitude and nonlinear characteristics from the conductivity fluctuation data; The average value, rate of change, fluctuation amplitude, and nonlinear characteristics are used as hydration characteristics, and the hydration characteristics are compared with the pre-stored skin hydration electrical characteristic model to obtain the comparison results. Based on the comparison results, a quantization value between 0 and 1 is generated through a fuzzy logic system or weighted summation calculation as a second index to characterize local skin hydration; the quantization value is configured to increase when the hydration feature indicates an increase in skin hydration.

7. The signal pattern recognition method for a mobile intelligent cell health repair watch according to claim 1, characterized in that, The step of temporally correlating the scenario information with physiological signals synchronously acquired by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference includes: The context information is compared with the timestamps carried by the physiological signals synchronously collected by the skin conductance response sensor and the microfluidic biochip integrated on the watch strap, and the context information and the physiological signals are aligned on the time axis to obtain the correlation result; the physiological signals include the skin conductance response signal collected by the skin conductance response sensor and the biochemical indicator signal collected by the microfluidic biochip; Based on the correlation results, the signal artifacts in the physiological signals that correspond to the context information are marked to obtain a marked signal used to indicate potential physical interference and / or pseudo-physiological interference.

8. The signal pattern recognition method for a mobile intelligent cell health repair watch according to claim 7, characterized in that, The step of marking signal artifacts in the physiological signal corresponding to the context information based on the correlation results to obtain marked signals for indicating potential physical interference and / or pseudo-physiological interference includes: Based on the correlation results, within the time period corresponding to the scenario information, signal spikes or rapid drifts in the skin conductance response signal with instantaneous change rate exceeding a preset physiological change rate threshold are identified, as well as abnormal fluctuations in the biochemical indicator signal with reading amplitude changes exceeding a preset biochemical change threshold are identified. The signal spikes or rapid drifts and the abnormal fluctuations are marked to obtain a marked signal for indicating potential physical interference and / or pseudo-physiological interference.

9. The signal pattern recognition method for a mobile intelligent cell health repair watch according to claim 1, characterized in that, The step of adaptively adjusting the processing strategy for the physiological signal based on the context information and the labeled signal to suppress interference and improve the reliability of physiological signal recognition includes: Based on the context information and the marker signal, the type of interference indicated by the marker signal is determined; the type of interference includes potential physical interference and pseudo-physiological interference. Based on the type of interference, the processing strategy for the physiological signal is adaptively adjusted to suppress interference and improve the reliability of physiological signal recognition. The processing strategy includes: for physiological signals marked as potential physical interference, triggering and switching to an adaptive filtering algorithm to process the physiological signal; for physiological signals marked as false physiological interference, reducing the weight of the physiological signal in subsequent pattern recognition.

10. A mobile intelligent cell health repair watch signal pattern recognition system, characterized in that, include: The first index acquisition module is used to continuously collect and analyze wrist micro-movement data generated by the user while wearing the watch based on the mechanical sensor integrated in the watch strap, and obtain a first index to characterize the stability of the skin contact between the skin conductance sensor and the skin. The second index acquisition module is used to continuously acquire and analyze the baseline conductivity data collected by the skin conductance sensor integrated on the watch strap to obtain a second index for characterizing local skin hydration. The scenario information generation module is used to take a first index whose value is lower than a preset stability judgment threshold and / or a second index whose value is higher than a preset hydration judgment threshold as scenario information. The signal correlation module is used to temporally correlate the scenario information with the physiological signals synchronously collected by the skin conductance sensor and the microfluidic biochip integrated on the watch strap to obtain a marker signal for indicating potential physical interference and / or pseudo-physiological interference. The processing strategy adjustment module is used to adaptively adjust the processing strategy for the physiological signal based on the context information and the labeled signal, so as to suppress interference and improve the reliability of physiological signal recognition.

Images