Intelligent equipment wearing detection method based on multiple sign sensors

Abstract

The invention discloses an intelligent device wearing detection method based on multiple physical sign sensors. The method is suitable for an intelligent bracelet configured with multiple types of physical sign sensors. The detection method comprises the following steps: acquiring sign data collected by at least two types of sign sensors; according to the currently obtained physical sign data type, at least two detection items are selected from five detection items of temperature detection, motion detection, heart rate detection, ambient light detection and wearing sequence detection for detection, and at least two sub-wearing probabilities are obtained; inputting the at least two sub-wearing probabilities into a pre-constructed wearing probability detection model to obtain a wearing probability a; adaptively acquiring a corresponding detection threshold b according to the selected detection item; and comparing the current detection value a with the detection threshold value b, if a is greater than or equal to b, indicating that the user is worn, otherwise, indicating that the user is not worn. According to the invention, at least two types of sign sensors are combined to detect the wearing condition of the intelligent equipment, the detection precision is improved, and the misjudgment condition is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent wearable device detection, and more specifically to an intelligent device wearing detection method based on a multi-vital sign sensor. Background Art

[0002] With the development of science and technology, people's detection of their own physical signs is no longer limited to specific medical equipment. Instead, they can use smart devices to detect physical signs anytime and anywhere, such as smart bracelets, smart watches, smart clothes, etc., which is convenient and fast. For smart devices, it is very important to identify whether the user is wearing them.

[0003] At present, there are two commonly used technologies for identifying whether a user is wearing a smart device: capacitive detection and optical detection. Among them, capacitive detection technology determines whether a user is wearing a device by the change in capacitance when the user wears or takes off the device. This technology can detect whether the user is wearing it to a certain extent, but when the user wears the device first and then turns it on, the capacitance value does not change, and the device will determine that the user is not wearing it; optical detection technology uses the device to emit infrared rays and detect the distance between the device and the object to determine whether the user is wearing the device, but when the device is close to the desktop or clothing pocket, there will be misjudgment.

[0004] Therefore, how to provide a smart device wearing detection method based on multiple vital sign sensors that can reduce misjudgment is a problem that technicians in this field urgently need to solve. Summary of the invention

[0005] In view of this, the present invention provides a smart device wearing detection method based on multiple vital sign sensors, which combines at least two vital sign sensors to detect the wearing status of the smart device, thereby improving detection accuracy and reducing misjudgment.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A smart device wearing detection method based on multiple vital sign sensors, which is applicable to a smart bracelet equipped with multiple types of vital sign sensors;

[0008] The detection method comprises:

[0009] Acquire vital sign data collected by at least two types of vital sign sensors;

[0010] According to the currently acquired vital sign data type, at least two items are selected from temperature detection, motion detection, heart rate detection, ambient light detection and wearing sequence detection to perform wearing probability detection, and at least two sub-wearing probabilities are obtained;

[0011] Input at least two sub-wearing probabilities into a pre-built wearing probability detection model to obtain a wearing probability a;

[0012] According to the selected detection item, adaptively obtain the corresponding detection threshold b;

[0013] Compare the wearing probability a with the corresponding detection threshold b. If a ≥ b, it means the device is being worn; otherwise, it means the device is not being worn.

[0014] Furthermore, the vital sign sensor includes: a photoelectric sensor, a temperature sensor and a motion sensor; the photoelectric sensor is used to collect heart rate and ambient light data; two temperature sensors are provided, one of which is arranged at the bottom of the smart bracelet to collect body surface temperature, and the other is arranged on the upper side of the inside of the smart bracelet to collect near-body surface temperature.

[0015] Furthermore, the process of obtaining the sub-wearing probability under the temperature detection project includes:

[0016] Under different wearing categories, segmented intervals are set for the body surface temperature, and the temperature difference segmented intervals between the body surface temperature and the near-surface temperature are set under each segmented interval of the body surface temperature. Each temperature difference segmented interval under each segmented interval of the body surface temperature corresponds to a sub-wearing probability; the wearing categories under the temperature detection item include: indoor wearing, indoor not wearing, outdoor wearing, outdoor not wearing and direct sunlight;

[0017] Based on the first probability classification model, the corresponding body surface temperature segmentation interval, temperature difference segmentation interval and corresponding wearing category are matched for the currently acquired body surface temperature and temperature difference, and the first sub-wearing probability under the temperature detection item is obtained;

[0018] Among them, as the wearing time increases, the segmented interval thresholds of the body surface temperature, the segmented interval thresholds of the temperature difference and the sub-wearing probabilities of the corresponding intervals are adaptively adjusted.

[0019] Furthermore, according to the overall wearing probability a finally given by the wearing detection, the temperature sequence corresponding to the time of the good wearing category is taken into the training interval, and the sub-wearing probability value corresponding to each temperature data interval is updated;

[0020] Probability update methods include:

[0021] Take the maximum value of the sub-wearing probability a1 and the overall wearing probability a of the temperature detection item at the corresponding time point;

[0022] Or if the overall probability a is greater than the sub-wearing probability a1 of the temperature detection item, then take the average of the two, otherwise keep the atomic wearing probability a1;

[0023] If the overall probability a is greater than the sub-wearing probability a1 of the temperature detection item, it is updated, but the updated sub-wearing probability determines that the temperature-related data worn should be within the range of the experimental range; if it is not within the corresponding range, the wearing period is recorded as a specific environment, and at the same time, the first probability classification model is updated.

[0024] Furthermore, the process of obtaining the sub-wearing probability under the motion detection project includes:

[0025] Under different sports categories, the amount of exercise in different time periods is divided into exercise intervals, and each exercise interval corresponds to a sub-wearing probability; the sports categories include: stationary, active, wearing and taking off;

[0026] Based on the second probability classification model, the corresponding exercise amount interval and exercise category are matched for the currently acquired exercise data to obtain the second sub-wearing probability under the exercise detection item.

[0027] Furthermore, the first probability classification model and the second probability classification model are: LR model or SVC model; the wearing probability detection model is a logistic regression model, an svm model or a lightweight TCN / CNN model.

[0028] Furthermore, the process of obtaining the sub-wearing probability under the heart rate detection project includes:

[0029] Determine in advance the sub-wearing probabilities corresponding to different approximate heart rate value-related feature thresholds or intervals under different wearing categories under the heart rate detection project;

[0030] Match the corresponding wearing category for the currently acquired PPG waveform data of a specific duration; the wearing categories under the heart rate detection project include: stationary, regular exercise, irregular exercise and not worn;

[0031] All pulse rate nodes in the PPG waveform data are calculated to obtain relevant features of the approximate heart rate value, and the relevant features are matched with a predetermined threshold or interval of relevant features of the approximate heart rate value to obtain the third sub-wearing probability; the relevant features are one or more of the mean, standard deviation, skewness, kurtosis and longest continuous interval range.

[0032] Furthermore, the process of obtaining the wearing probability under the ambient light detection project includes:

[0033] Set different sub-wearing probabilities for the ambient light detection results at different time nodes in the past period of time;

[0034] The current ambient light collection time is matched with the corresponding time node to obtain the fourth sub-wearing probability of the ambient light detection item.

[0035] Furthermore, the process of obtaining the sub-wearing probability under the wearing sequence detection project includes:

[0036] Obtain the motion state, wearing time point and removal time point in the historical time interval, where different motion states at different time points correspond to a sub-wearing probability;

[0037] The corresponding time point is matched to the current motion data collection time, and the fifth sub-wearing probability under the wearing sequence detection item is obtained.

[0038] Furthermore, the detection threshold b is set to two, namely b1 and b2, where b1 is less than b2, b1 is used to determine whether it is worn, and b2 is used to determine whether it is worn well. When a≥b1, it means that it is worn, otherwise it is not worn; when a≥b2, it means that it is worn well, and when b1≤a≤b2, it means that it is worn poorly.

[0039] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a smart device wearing detection method based on a multi-vital sign sensor, which has the following beneficial effects:

[0040] 1. The present invention comprehensively utilizes signals collected by two or more vital sign sensors to determine whether the smart device is worn, thereby enhancing the reliability and accuracy of the wearing detection results;

[0041] 2. The present invention can adaptively select detection values ​​and judgment thresholds according to the physical sign data to determine whether the device is worn, thereby enhancing the adaptability of the detection;

[0042] 3. The present invention performs abnormal processing on the heart rate data, and sets the index threshold to participate in the wearing detection and judgment, so as to remove the interference of external factors;

[0043] 4. The present invention introduces the timing results of ambient light detection and user wearing detection into the wearing detection judgment procedure, increases historical considerations and sensitivity to changes in device status, and improves the accuracy of detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0045] Figure 1 This is a flow chart of the smart device wearing detection method based on multiple vital sign sensors provided by the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] like Figure 1 As shown, an embodiment of the present invention discloses a smart device wearing detection method based on multiple vital sign sensors, which is applicable to a smart bracelet equipped with multiple types of vital sign sensors; wherein the vital sign sensors include: a photoelectric sensor, a temperature sensor and a motion sensor; the photoelectric sensor is used to collect heart rate and ambient light data; two temperature sensors are provided, one of which is provided at the bottom of the smart bracelet for collecting body surface temperature, and the other is provided on the upper side of the inside of the smart bracelet for collecting near-body surface temperature;

[0048] The detection method comprises:

[0049] Acquire vital sign data collected by at least two types of vital sign sensors;

[0050] According to the currently acquired vital sign data type, at least two items are selected from the five detection items of temperature detection, motion detection, heart rate detection, ambient light detection and wearing sequence detection to perform wearing probability detection, and multiple sub-wearing probabilities are obtained;

[0051] At least two sub-wearing probabilities are input into a pre-built wearing probability detection model to obtain a wearing probability a; the wearing probability detection model is a logistic regression model, an SVM model or a lightweight TCN / CNN model; each sub-wearing probability can be weighted summed or directly summed.

[0052] According to the selected detection item, adaptively obtain the corresponding detection threshold b;

[0053] Compare the wearing probability a with the corresponding detection threshold b. If a ≥ b, it means the device is being worn; otherwise, it means the device is not being worn.

[0054] In one embodiment, the detection threshold is set to two, namely b1 and b2, where b1 is less than b2, b1 is used to determine whether it is worn, and b2 is used to determine whether it is worn well. When a≥b1, it means that it is worn, otherwise it is not worn; when a≥b2, it means that it is worn well, and when b1≤a≤b2, it means that it is not worn well.

[0055] In a specific embodiment, the process of determining the current detection value a includes:

[0056] 1. The process of obtaining the sub-wearing probability under the temperature detection project includes:

[0057] Under different wearing categories, segmented intervals are set for the body surface temperature, and the temperature difference segmented intervals between the body surface temperature and the near-surface temperature are set under each segmented interval of the body surface temperature. Each temperature difference segmented interval under each segmented interval of the body surface temperature corresponds to a sub-wearing probability; the wearing categories under the temperature detection item include: indoor wearing, indoor not wearing, outdoor wearing, outdoor not wearing and direct sunlight;

[0058] Based on the first probability classification model, the corresponding body surface temperature segmentation interval, temperature difference segmentation interval and corresponding wearing category are matched for the currently acquired body surface temperature and temperature difference, and the first sub-wearing probability under the temperature detection item is obtained;

[0059] Among them, as the wearing time increases, the segmented interval thresholds of the body surface temperature, the segmented interval thresholds of the temperature difference and the sub-wearing probabilities of the corresponding intervals are adaptively adjusted.

[0060] Specifically, because the human extremity surface temperature is affected by multiple conditions such as ambient temperature, surface clothing coverage, blood perfusion, and activity, it is necessary to adaptively adjust the segmented interval threshold of the surface temperature, the segmented interval threshold of the temperature difference, and the sub-wearing probability of the corresponding interval. The specific method is:

[0061] Initial wearing test, the relationship between temperature related data and probability value is F ori (T1, T2, T3) are the temperature series characteristics such as body surface temperature, ambient temperature and temperature difference;

[0062] According to the overall wearing probability a finally given by the wearing test, the temperature sequence corresponding to the time of the good wearing category is taken into the training interval, and the sub-wearing probability value corresponding to each temperature data interval is updated;

[0063] The number of updates is limited to num k ; The update is performed when the user binds a new device or when the temperature suddenly changes during the wearing detection process.

[0064] There are many ways to update the probability, including:

[0065] Take the maximum value of the sub-wearing probability a1 and the overall wearing probability a of the temperature detection item at the corresponding time point;

[0066] Or if the overall probability a is greater than the sub-wearing probability a1 of the temperature detection item, then take the average of the two, otherwise keep the atomic wearing probability a1;

[0067] If the overall probability a is greater than the sub-wearing probability a1 of the temperature detection item, it will be updated, but the updated sub-wearing probability determines that the temperature-related data worn should be within the range of the experimental range; if it is not within the corresponding range, the wearing period will be recorded as a specific environment (extreme cold, hot, etc.), and at the same time, the first probability classification model will be updated.

[0068] 2. The process of obtaining the sub-wearing probability under the motion detection project includes:

[0069] Under different sports categories, the amount of exercise in different time periods is divided into exercise intervals, and each exercise interval corresponds to a sub-wearing probability; the sports categories include: stationary, active, wearing and taking off;

[0070] Based on the second probability classification model, the corresponding exercise amount interval and exercise category are matched for the currently acquired exercise data to obtain the second sub-wearing probability under the exercise detection item.

[0071] The first probability classification model and the second probability classification model are: LR model or SVC model.

[0072] 3. The process of obtaining the sub-wearing probability under the heart rate detection project includes:

[0073] Determine in advance the sub-wearing probabilities corresponding to different approximate heart rate value-related feature thresholds or intervals under different wearing categories under the heart rate detection project;

[0074] Match the corresponding wearing category for the currently acquired PPG waveform data of a specific duration; the wearing categories under the heart rate detection project include: stationary, regular exercise, irregular exercise and not worn;

[0075] All pulse rate nodes in the PPG waveform data are calculated to obtain relevant features of the approximate heart rate value, and the relevant features are matched with a predetermined threshold or interval of relevant features of the approximate heart rate value to obtain the third sub-wearing probability; the relevant features are one or more of the mean, standard deviation, skewness, kurtosis and longest continuous interval range.

[0076] In the embodiment of the present invention, there are two ways to determine the wearing probability of the third child. One is to determine by similarity, specifically:

[0077] The preset user characteristics are user characteristics P2 of no heart and other diseases that may cause the stability of heart rate / pulse wave detection.

[0078] For users who do not have heart or other diseases that may cause heart rate / pulse wave detection stability, the preset wearing value (i.e., the characteristics related to the approximate heart rate value) X h Similarity>P1 h , similarity with the preset user ≥ P2;

[0079] For users with corresponding problems, they and users with heart and other diseases that may cause heart rate / pulse wave detection stability wear value feature X u Similarity>P1 u , similarity with the preset user <P2。

[0080] The other is to judge by threshold, specifically:

[0081] For users who do not have heart or other diseases that may cause heart rate / pulse wave detection stability, their mean, standard deviation, etc. are usually in the interval [a1, a2], [b1, b2], ..., and the relevant features are matched with the preset interval to obtain the third child wearing probability, and three-level labels of good quality, general quality and poor quality can also be defined.

[0082] 4. The process of obtaining the wearing probability under the ambient light detection project includes:

[0083] Set different sub-wearing probabilities for the ambient light detection results at different time nodes in the past period of time;

[0084] The current ambient light collection time is matched with the corresponding time node to obtain the fourth sub-wearing probability of the ambient light detection item.

[0085] 5. The process of obtaining the sub-wearing probability under the wearing sequence detection project includes:

[0086] Obtain the motion state, wearing time point and removal time point in the historical time interval, where different motion states at different time points correspond to a sub-wearing probability;

[0087] The corresponding time point is matched to the current motion data collection time, and the fifth sub-wearing probability under the wearing sequence detection item is obtained.

[0088] The reason for using the above probability data in the past period of time for judgment is that according to a product usage idea: users will not wear and take off the device frequently, and the time series model can fit the characteristics of wearing, taking off and frequent removal. Among them, the probabilities of taking off, not wearing and frequent removal and good wearing should all be set in the range of not wearing or not wearing well.

[0089] The present invention can comprehensively utilize two or more types of vital sign sensors for wearing detection, which can specifically include four situations, namely:

[0090] ① If only the data collected by the temperature sensor and the motion sensor are obtained, temperature detection, motion detection and wearing sequence detection are performed, the sub-wearing probabilities under the three detection items are summed, and the detection threshold b is adaptively adjusted to perform wearing detection.

[0091] ② If only the data collected by the temperature sensor and the photoelectric sensor are obtained, temperature detection, heart rate detection and ambient light detection are performed, the sub-wearing probabilities under the three detection items are summed, and the detection threshold b is adaptively adjusted to perform wearing detection.

[0092] ③ If only the data collected by the motion sensor and the photoelectric sensor are obtained, motion detection, heart rate detection, ambient light detection and wearing sequence detection are performed, the sub-wearing probabilities under the four detection items are summed, and the detection threshold b is adaptively adjusted to perform wearing detection.

[0093] ④ If the data collected by the temperature sensor, motion sensor and photoelectric sensor are obtained, temperature detection, motion detection, heart rate detection, ambient light detection and wearing sequence detection are performed, the sub-wearing probabilities under the five detection items are summed, and the detection threshold b is adaptively adjusted to perform wearing detection.

[0094] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0095] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A smart device wearing detection method based on multi-vital sign sensors, characterized in that: It is applicable to a smart bracelet equipped with multiple types of vital sign sensors; the detection method includes: Acquire vital sign data collected by at least two types of vital sign sensors; According to the currently acquired vital sign data type, at least two items are selected from temperature detection, motion detection, heart rate detection, ambient light detection and wearing sequence detection to perform wearing probability detection, and at least two sub-wearing probabilities are obtained; Input at least two sub-wearing probabilities into a pre-built wearing probability detection model to obtain a wearing probability a; According to the selected detection item, adaptively obtain the corresponding detection threshold b; Compare the wearing probability a with the corresponding detection threshold b. If a ≥ b, it means the device is being worn; otherwise, it means the device is not being worn.

2. According to claim 1, a smart device wearing detection method based on multi-vital sign sensors is characterized in that: The vital sign sensor includes: a photoelectric sensor, a temperature sensor and a motion sensor; the photoelectric sensor is used to collect heart rate and ambient light data; two temperature sensors are provided, one of which is arranged at the bottom of the smart bracelet to collect body surface temperature, and the other is arranged on the upper side of the inside of the smart bracelet to collect near-body surface temperature.

3. According to claim 1, a smart device wearing detection method based on multi-vital sign sensors is characterized in that: The process of obtaining the sub-wearing probability under the temperature detection project includes: Under different wearing categories, segmented intervals are set for the body surface temperature, and the temperature difference segmented intervals between the body surface temperature and the near-surface temperature are set under each segmented interval of the body surface temperature. Each temperature difference segmented interval under each segmented interval of the body surface temperature corresponds to a sub-wearing probability; the wearing categories under the temperature detection item include: indoor wearing, indoor not wearing, outdoor wearing, outdoor not wearing and direct sunlight; Based on the first probability classification model, the corresponding body surface temperature segmentation interval, temperature difference segmentation interval and corresponding wearing category are matched for the currently acquired body surface temperature and temperature difference, and the first sub-wearing probability under the temperature detection item is obtained; Among them, as the wearing time increases, the segmented interval thresholds of the body surface temperature, the segmented interval thresholds of the temperature difference and the sub-wearing probabilities of the corresponding intervals are adaptively adjusted.

4. The method for detecting the wearing of a smart device based on a multi-vital sign sensor according to claim 3, characterized in that: The process of adaptively adjusting the sub-wearing probability under the temperature detection item includes: According to the overall wearing probability a finally given by the wearing test, the temperature sequence corresponding to the time of the good wearing category is taken into the training interval, and the sub-wearing probability value corresponding to each temperature data interval is updated; Probability update methods include: Take the maximum value of the sub-wearing probability a1 and the overall wearing probability a of the temperature detection item at the corresponding time point; Or if the overall probability a is greater than the sub-wearing probability a1 of the temperature detection item, then take the average of the two, otherwise keep the atomic wearing probability a1; If the overall probability a is greater than the sub-wearing probability a1 of the temperature detection item, it is updated, but the updated sub-wearing probability determines that the temperature-related data worn should be within the range of the experimental range; if it is not within the corresponding range, the wearing period is recorded as a specific environment, and at the same time, the first probability classification model is updated.

5. A smart device wearing detection method based on multi-vital sign sensors according to claim 1 or 3, characterized in that: The process of obtaining the sub-wearing probability under the motion detection project includes: Under different sports categories, the amount of exercise in different time periods is divided into exercise intervals, and each exercise interval corresponds to a sub-wearing probability; the sports categories include: stationary, active, wearing and taking off; Based on the second probability classification model, the corresponding exercise amount interval and exercise category are matched for the currently acquired exercise data to obtain the second sub-wearing probability under the exercise detection item.

6. The method for detecting the wearing of a smart device based on a multi-vital sign sensor according to claim 5, characterized in that: The first probability classification model and the second probability classification model are: LR model or SVC model; the wearing probability detection model is a logistic regression model, an SVM model or a lightweight TCN / CNN model.

7. The method for detecting the wearing of a smart device based on a multi-vital sign sensor according to claim 1, characterized in that: The process of obtaining the sub-wearing probability under the heart rate detection project includes: Determine in advance the sub-wearing probabilities corresponding to different approximate heart rate value-related feature thresholds or intervals under different wearing categories under the heart rate detection project; Match the corresponding wearing category for the currently acquired PPG waveform data of a specific duration; the wearing categories under the heart rate detection project include: stationary, regular exercise, irregular exercise and not worn; All pulse rate nodes in the PPG waveform data are calculated to obtain relevant features of the approximate heart rate value, and the relevant features are matched with a predetermined threshold or interval of relevant features of the approximate heart rate value to obtain the third sub-wearing probability; the relevant features are one or more of the mean, standard deviation, skewness, kurtosis and longest continuous interval range.

8. The method for detecting the wearing of a smart device based on a multi-vital sign sensor according to claim 1, characterized in that: The process of obtaining the wearing probability under the ambient light detection project includes: Set different sub-wearing probabilities for the ambient light detection results at different time nodes in the past period of time; The current ambient light collection time is matched with the corresponding time node to obtain the fourth sub-wearing probability of the ambient light detection item.

9. The method for detecting the wearing of a smart device based on a multi-vital sign sensor according to claim 1, characterized in that: The process of obtaining the sub-wearing probability under the wearing sequence detection project includes: Obtain the motion state, wearing time point and removal time point in the historical time interval, where different motion states at different time points correspond to a sub-wearing probability; The corresponding time point is matched to the current motion data collection time, and the fifth sub-wearing probability under the wearing sequence detection item is obtained.

10. The method for detecting the wearing of a smart device based on a multi-vital sign sensor according to claim 1, characterized in that: The detection threshold b is set to two, namely b1 and b2, where b1 is less than b2. b1 is used to determine whether it is worn, and b2 is used to determine whether it is worn well. When a≥b1, it means that it is worn, otherwise it is not worn; when a≥b2, it means that it is worn well, and when b1≤a≤b2, it means that it is not worn well.

Images