Personality Detection Method, Device, Electronic Device, Storage Medium and Program Product

Abstract

The present disclosure provides a personality detection method, apparatus, electronic device, storage medium, and program product. The personality detection method includes: obtaining state data of a target object, where the state data includes human heart rate variability data of the object detected by a millimeter-wave radar in a task state of the object performing a detection task; using a personality detection model to process the state data of the target object to obtain predicted personality information of the target object, where the personality detection model is trained by a supervised method. This method can provide a relatively reliable basis for personality detection by obtaining state data including human heart rate variability data. By using a millimeter-wave radar to detect human heart rate variability data, it helps to ensure the authenticity, accuracy, and reliability of the obtained data. By using a personality detection model to process the state data, it is possible to mine the laws of the state data with the help of the model and achieve reliable prediction of personality information.

Description

Technical Field

[0001] The present disclosure generally relates to the field of data processing technologies, and more specifically, to a personality detection method, apparatus, electronic device, storage medium, and program product. Background Art

[0002] Coronary Heart Disease (CHD), also known as coronary atherosclerotic heart disease, is a common cardiovascular disease among middle-aged and elderly people. As a typical psychosomatic disease, CHD not only affects the physical health of patients but also has a significant impact on their psychological state. In recent years, CHD has gradually become the "number one killer" threatening people's health, and its high incidence and high mortality rate have brought huge social and economic burdens. The families of patients not only face huge economic pressure but also need to invest a lot of time and energy in nursing. Research shows that Type A personality and Type D personality are of great significance in the detection of CHD. Through risk assessment, high-risk groups can be identified early, personalized treatment and rehabilitation plans can be formulated, and the treatment effect and patient compliance can be improved. Understanding the personality type helps doctors and psychologists develop comprehensive treatment plans, combine psychological and behavioral interventions, and improve the overall health level. In addition, through the systematic detection and research of these personality types, the accumulated data can reveal the complex relationship between psychological factors and CHD, provide a scientific basis for prevention and treatment strategies, and thus promote the comprehensive prevention and management of cardiovascular diseases.

[0003] A common way to distinguish personality is through scale detection, which usually includes using a series of questionnaires to quantitatively evaluate personality traits. Scale detection methods have significant reference value because they can systematically capture and quantify an individual's performance in different personality dimensions. However, inevitably, during the scale detection process, some test takers may exhibit ambiguous or even deceptive filling behaviors, which may be due to various reasons. For example, test takers may want to show their idealized side and thus tend to choose options that are generally recognized by society when answering questionnaires; or they may be reluctant to answer questions involving personal sensitive information frankly due to concerns about privacy. Such behaviors will lead to data deviation and thus have a certain wrong impact on the final evaluation results. Summary of the Invention

[0004] The present disclosure provides a personality detection method, apparatus, electronic device, storage medium, and program product for solving at least one of the above problems.

[0005] According to the first aspect of the embodiments of the present disclosure, a personality detection method is provided. The personality detection method includes: obtaining state data of a target object, where the state data includes human heart rate variability data of the object detected by a millimeter-wave radar in the task state of the object performing a detection task; using a personality detection model to process the state data of the target object to obtain predicted personality information of the target object, where the personality detection model is trained by a supervised method.

[0006] Optionally, the detection task includes at least one of the following: a rest state detection task, a diaphragmatic breathing state detection task, a passive stress state detection task, where the passive stress state detection task includes at least one of the following: a passive stress state detection task after rest, a passive stress state detection task after diaphragmatic breathing, and the passive stress state detection task is a task of inducing a stress response in a corresponding object through an emotion induction material.

[0007] Optionally, the state data further includes questionnaire data, which is obtained by the object answering a preset questionnaire, and the preset questionnaire includes at least one scale related to personality detection.

[0008] Optionally, the questionnaire data includes the scores of each scale in the at least one scale related to personality detection. Wherein, using the personality detection model to process the state data of the target object to obtain the predicted personality information of the target object includes: performing time-frequency domain feature extraction processing on the human heart rate variability data corresponding to each detection task to obtain the time domain features and frequency domain features corresponding to each detection task; performing fusion processing on the time domain features, frequency domain features corresponding to each detection task, and the scores of each scale to obtain a fusion vector; inputting the fusion vector into the personality detection model to obtain the predicted personality information.

[0009] Optionally, the questionnaire data includes the scores of each of the at least one personality detection-related scale. The training samples of the personality detection model include the status data of the reference object and the true personality information. Among them, the personality detection model is trained through the following steps: Obtain the status data of the reference object; Use the personality detection model to be trained to process the status data of the reference object to obtain the predicted scale scores and predicted personality information of the reference object, where the predicted scale scores are the predicted values of the scores of the reference object answering the at least one personality detection-related scale; According to the predicted personality information and the true personality information of the reference object, determine the value of the weighted cross-entropy loss function as the first loss value, where the weighted cross-entropy loss function is obtained by adding a personality category weight on the basis of the cross-entropy loss function, and the personality category weight is negatively correlated with the proportion of the personality category corresponding to the true personality information of the reference object in all training samples; According to the predicted scale scores of the reference object and the scores of the corresponding scales in the status data, determine the value of the correlation loss function as the second loss value, where the correlation loss function is the difference between a preset constant and a preset correlation coefficient; According to the first loss value and the second loss value, determine the total loss value; Based on the total loss value, adjust the parameters of the personality detection model to be trained to obtain the personality detection model.

[0010] Optionally, the at least one personality detection-related scale includes at least one of the following: Type A Behavior Pattern Questionnaire, Type D Personality Questionnaire, Trait Coping Style Questionnaire, Self-Rating Depression Scale, Self-Rating Anxiety Scale, Anger and Anger Expression Scale, Perceived Social Support Scale, Rumination Response Scale; and / or the preset questionnaire further includes a psychosomatic disease questionnaire.

[0011] According to the second aspect of the embodiments of the present disclosure, a personality detection device is provided. The personality detection device includes: an acquisition unit configured to acquire the status data of a target object, where the status data includes the human heart rate variability data of the corresponding object detected in the task state, and the human heart rate variability data is detected using a millimeter wave radar; a processing unit configured to use the personality detection model to process the status data of the target object to obtain the predicted personality information of the target object, where the personality detection model is trained by a supervised method.

[0012] Optionally, the detection task includes at least one of the following: rest state detection task, abdominal breathing state detection task, passive stress state detection task, where the passive stress state detection task includes at least one of the following: passive stress state detection task after rest, passive stress state detection task after abdominal breathing, and the passive stress state detection task is a task of inducing a stress response in the corresponding object through emotional induction materials.

[0013] Optionally, the status data further includes questionnaire data, which is obtained by an object answering a preset questionnaire, and the preset questionnaire includes at least one scale related to personality detection.

[0014] Optionally, the questionnaire data includes the scores of each scale in the at least one scale related to personality detection, and the processing unit is further configured to: perform time-frequency domain feature extraction processing on the human heart rate variability data corresponding to each detection task to obtain the time domain features and frequency domain features corresponding to each detection task; perform fusion processing on the time domain features, frequency domain features corresponding to each detection task, and the scores of each scale to obtain a fusion vector; input the fusion vector into the personality detection model to obtain the predicted personality information.

[0015] Optionally, the questionnaire data includes the scores of each scale in the at least one scale related to personality detection, and the training samples of the personality detection model include the status data and true personality information of a reference object. Among them, the personality detection model is trained through the following steps: obtain the status data of the reference object; use the personality detection model to be trained to process the status data of the reference object to obtain the predicted scale scores and predicted personality information of the reference object, where the predicted scale scores are the predicted values of the scores of the reference object answering the at least one scale related to personality detection; according to the predicted personality information and true personality information of the reference object, determine the value of the weighted cross-entropy loss function as the first loss value, where the weighted cross-entropy loss function is obtained by adding a personality category weight on the basis of the cross-entropy loss function, and the personality category weight is negatively correlated with the proportion of the personality category corresponding to the true personality information of the reference object in all training samples; according to the predicted scale scores of the reference object and the scores of the corresponding scales in the status data, determine the value of the correlation loss function as the second loss value, where the correlation loss function is the difference between a preset constant and a preset correlation coefficient; according to the first loss value and the second loss value, determine the total loss value; based on the total loss value, adjust the parameters of the personality detection model to be trained to obtain the personality detection model.

[0016] Optionally, the at least one scale related to personality detection includes at least one of the following: Type A Behavior Pattern Questionnaire, Type D Personality Questionnaire, Trait Coping Style Questionnaire, Self-Rating Depression Scale, Self-Rating Anxiety Scale, Anger and Anger Expression Scale, Perceived Social Support Scale, Indulgence Response Scale; and / or the preset questionnaire further includes a psychosomatic disease questionnaire.

[0017] According to a third aspect of the embodiments of the present disclosure, there is provided an electronic device, including: at least one processor; at least one memory storing computer-executable instructions, wherein when the computer-executable instructions are run by the at least one processor, the at least one processor is caused to execute a personality detection method according to an exemplary embodiment of the present disclosure.

[0018] According to a fourth aspect of the embodiments of the present disclosure, there is provided a computer-readable storage medium, and when instructions in the computer-readable storage medium are run by at least one processor, the at least one processor is caused to execute a personality detection method according to an exemplary embodiment of the present disclosure.

[0019] According to a fifth aspect of the embodiments of the present disclosure, there is provided a computer program product, including computer instructions, and when the computer instructions are run by at least one processor, the at least one processor is caused to execute a personality detection method according to an exemplary embodiment of the present disclosure.

[0020] The technical solutions provided by the embodiments of the present disclosure at least bring the following beneficial effects:

[0021] According to the personality detection method, device, electronic device, and storage medium of the present disclosure, by executing a specified detection task, different reactions of an object can be stimulated, and these reactions can be objectively reflected from the physiological data of the object, reducing the influence of the subjective factors of the object on the detection result. Through a large number of studies, it is found that relatively single physiological data such as blood pressure and heart rate do not show significant differences among objects with different personalities, while the physiological data of human heart rate variability (referring to the change in the heartbeat interval), which has dynamic characteristics, shows obvious differences among objects with different personalities. By obtaining state data including this data, a relatively reliable basis can be provided for personality detection. In addition, by using a millimeter-wave radar to detect the physiological data of human heart rate variability, there is no need for electrodes, patches, or other devices that directly contact the skin, thereby reducing the interference to the target object, reducing the pressure and discomfort of the target object, having a non-invasive feature, and being able to monitor the target object in a natural state, which helps to ensure the authenticity, accuracy, and reliability of the obtained data. Finally, by using a personality detection model obtained through supervised training to process the state data, the rules of the state data can be mined with the help of the model, so as to process the state data and achieve reliable prediction of personality information.

[0022] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings herein are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present disclosure, and are used together with the specification to explain the principles of the present disclosure and do not constitute an improper limitation to the present disclosure.

[0024] Figure 1 is a flowchart of a personality detection method according to an exemplary embodiment of the present disclosure;

[0025] Figure 2 is a schematic flowchart of a rest state detection task according to a specific embodiment of the present disclosure;

[0026] Figure 3 is a schematic flowchart of a diaphragmatic breathing state detection task according to a specific embodiment of the present disclosure;

[0027] Figure 4 is a schematic flowchart of a passive stress state detection task after rest according to a specific embodiment of the present disclosure;

[0028] Figure 5 is a schematic flowchart of a passive stress state detection task after diaphragmatic breathing according to a specific embodiment of the present disclosure;

[0029] Figure 6 is a block diagram of a personality detection device according to an exemplary embodiment of the present disclosure;

[0030] Figure 7 is a block diagram of an electronic device according to an exemplary embodiment of the present disclosure. Detailed Embodiments

[0031] In order to enable those of ordinary skill in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such used data may be interchanged under appropriate circumstances so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following examples do not represent all embodiments consistent with the present disclosure. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0033] It should be noted here that "at least one of a number of items" as used in the present disclosure all represents three parallel cases, namely, "any one of the number of items", "a combination of any multiple of the number of items", and "all of the number of items". For example, "including at least one of A and B" includes the following three parallel cases: (1) including A; (2) including B; (3) including A and B. Another example, "performing at least one of step one and step two" means the following three parallel cases: (1) performing step one; (2) performing step two; (3) performing step one and step two.

[0034] Next, a personality detection method, apparatus, electronic device, and storage medium according to an exemplary embodiment of the present disclosure will be described in detail with reference to the accompanying drawings.

[0035] Figure 1 It is a flowchart of a personality detection method according to an exemplary embodiment of the present disclosure. This personality detection method can be executed by an electronic device with corresponding computing capabilities.

[0036] Referring to Figure 1 , in step S101, state data of the target object is acquired.

[0037] The state data includes the heart rate variability (HRV) data of the object detected by a millimeter-wave radar under the task state of the object performing the detection task.

[0038] By performing a specified detection task, different reactions of the object can be stimulated, and these reactions can be objectively reflected from the physiological data of the object, reducing the influence of the subjective factors of the object on the detection result. Through a large number of studies, it is found that relatively single physiological data such as blood pressure and heart rate do not show significant differences among objects with different personalities, while the physiological data with dynamic characteristics, namely the heart rate variability data of the human body (referring to the change in the heartbeat interval), shows obvious differences among objects with different personalities. By acquiring the state data including this data, a relatively reliable basis can be provided for personality detection. It should be understood that when performing the detection task, the heart rate signal of the object is directly detected, and the interval duration between two adjacent heartbeats can be calculated therefrom, and a sequence of interval durations can be obtained as the heart rate variability data of the human body.

[0039] Specifically, the core characteristics of Type A Personality include high competitiveness, a sense of time urgency, and a tendency to be irritable. High competitiveness means that individuals show strong achievement motivation and a competitive awareness in life and work. They are always eager for success, pursue excellence, set high standards, and are not only strict with themselves but also have high expectations for others. The sense of time urgency is manifested as individuals always feeling that time is not enough, liking to complete tasks quickly, being difficult to relax, often handling multiple tasks simultaneously, pursuing high efficiency, and being extremely uneasy about procrastination. The tendency to be irritable refers to individuals being easily impatient and angry, especially when encountering obstacles or delays. They have a low tolerance for setbacks and often show impatient and irritable emotions. This personality trait makes individuals often in a state of tension and anxiety when facing stress and challenges, thereby affecting their mental health and interpersonal relationships.

[0040] Type D Personality, that is, "distressed personality", is a psychological trait whose core characteristics include persistent negative emotions and social inhibition. Persistent negative emotions mean that individuals often experience negative emotions such as anxiety, depression, anger, and irritability. These emotions are not only intense but also persistent, and are particularly sensitive to stress and setbacks in life. Social inhibition is manifested as individuals avoiding and suppressing their emotions and thoughts in social situations, being afraid to express their true selves, worried about being rejected or criticized by others, and thus tending to avoid social interaction and emotional expression. This personality trait makes individuals often feel isolated and helpless when facing social and emotional challenges, thereby affecting their mental health and interpersonal relationships.

[0041] According to relevant research, the data of human heart rate variability can be an important indicator for distinguishing Type D Personality and Type A Personality because the data of human heart rate variability reflects the functional state of the autonomic nervous system, and its changes can reveal the differences in psychological and physiological stress responses. Specifically, time-frequency domain feature extraction can be performed on the data of human heart rate variability, and with the help of these time-domain features and frequency-domain features, the changes in heart rate intervals can be quantified and represented from different perspectives. It has been found through research that individuals with generally lower frequency-domain features usually show higher activity of the sympathetic nervous system and poorer regulation of the parasympathetic nervous system, which is consistent with the characteristics of high stress, tension, and hostility of Type A Personality; while Type D Personality individuals are often accompanied by negative emotions and social inhibition, and these emotional states will also lead to a decrease in frequency-domain features, but there are still regular differences in their different frequency-domain features (such as but not limited to the frequency-domain features of different frequency bands). Therefore, by measuring and analyzing the changes in heart rate intervals, an objective physiological means can be provided to distinguish different personality types.

[0042] In addition, traditional monitoring methods such as electrocardiogram and pulse wave monitoring require the use of electrodes, patches, or other monitoring devices that directly contact the skin, which can easily interfere with the subject and may affect the accuracy of the detection results. Millimeter-wave radar is a device that emits electromagnetic waves with a frequency in the millimeter-wave band towards the target and works by receiving the electromagnetic waves reflected by the target. Millimeter-wave radar also has the characteristics of high precision and high resolution, and can capture the subtle changes in heart rate and respiration by detecting and analyzing the tiny chest movements, thereby obtaining the human heart rate variability data. Millimeter-wave radar does not require electrodes, patches, or other devices that directly contact the skin, thus reducing the interference to the target subject, reducing the stress and discomfort of the target subject, and having a non-invasive feature, which can monitor the target subject in a natural state and help ensure the authenticity, accuracy, and reliability of the acquired data. This non-contact detection is especially suitable for those people who are skin-sensitive, suffering from skin diseases, or infants and young children who are not suitable for wearing traditional monitoring devices. The non-contact technology also has the advantage of reducing the spread of germs in public health and clinical environments, improving the safety and hygiene level.

[0043] Continue to refer to Figure 1 , in step S102, use the personality detection model to process the status data of the target subject to obtain the predicted personality information of the target subject.

[0044] The personality detection model is trained by a supervised method and can use the model to discover the rules of the status data, thereby processing the status data to achieve reliable prediction of the personality information.

[0045] Optionally, the training samples of the personality detection model include the status data of the reference object and the true personality information, and the true personality information is obtained by following up the reference object. Follow-up refers to an observation method in which a hospital regularly understands the changes in the patient's condition and guides the patient's rehabilitation by means of communication or other means for patients who have visited the hospital. In other words, the present disclosure selects objects that have been followed up, such as coronary heart disease follow-up patients, or regular physical examination personnel who are not ill. Of course, it can also be other personnel whose stable true personality can be determined through follow-up. These objects are used as reference objects. Taking coronary heart disease follow-up patients as an example, due to different personalities, especially type A and type D personalities, there are different impacts on the prognosis of coronary heart disease. Since the reference objects have been followed up for a long time, they have detailed information on treatment plans and prognosis, and there is often a relatively stable judgment on their personality types. Therefore, relatively reliable true personality information of the reference objects can be obtained, and the influence of human factors (for the detection scheme using scales, such as the subjective factors when the detection object answers, and the manual analysis of the scale evaluator) on personality assessment can be basically eliminated. Using the true personality information obtained in this way as the personality label in training can obtain high-quality training data, which helps to improve the accuracy of the trained personality detection model from the source. It should be understood that for reference objects without coronary heart disease, although there is no information such as treatment plans and prognosis, relatively stable personality detection results can also be obtained through long-term follow-up, thereby significantly reducing the error of single personality detection and ensuring the sample quality. As an example, the true personality information may include type A personality information and type D personality information.

[0046] Next, the personality detection method according to an exemplary embodiment of the present disclosure will be further introduced.

[0047] Optionally, the detection tasks include at least one of the following: rest state detection task, abdominal breathing state detection task, passive stress state detection task, where the passive stress state detection task includes at least one of the following: passive stress state detection task after rest, passive stress state detection task after abdominal breathing, and the passive stress state detection task is a task that triggers the stress response of the corresponding object through emotional induction materials. By adopting at least one of the above detection tasks, the heart rate variability data of the human body in different states can be obtained, thereby enriching the physiological description of the target object.

[0048] As an example, each detection task can be a short-term task (such as 5 minutes) or a long-term task (such as 24 hours), and the present disclosure does not limit this.

[0049] Short-term tasks are mainly used to evaluate the short-term responses of the autonomic nervous system, reflecting the dynamic balance between the sympathetic and parasympathetic nerves in a short period. In the selection of frequency-domain features of human heart rate variability data, total power (TotalPower, TP, frequency band ≤ 0.4 Hz), very low frequency band (Very Low Frequency, VLF, frequency band 0.0033 - 0.04 Hz) power, low frequency band (Low Frequency, LF, frequency band 0.04 - 0.15 Hz) power, high frequency band (High Frequency, HF, frequency band 0.15 - 0.4 Hz) power, the ratio of low frequency band power to high frequency band power (LF / HF), normalized low frequency band power (LFnorm), and normalized high frequency band power (HFnorm) can be adopted. Among them, the high frequency band power can reflect the parasympathetic nerve tone. The low frequency band power can reflect the combined effects of the sympathetic and parasympathetic nerves. The very low frequency band power usually reflects slower physiological changes, involving aspects such as body temperature regulation, hormone regulation, chronic stress, and immune system activities, and is related to the long-term regulation of the autonomic nervous system. Especially in the state of chronic stress or disease, the very low frequency band power may decrease, indicating abnormal autonomic nerve function. The ratio of low frequency band power to high frequency band power reflects the balance between the sympathetic and parasympathetic nerves. The units of normalized low frequency band power (LFnorm) and normalized high frequency band power (HFnorm) are (nu), and their calculation method is to divide the absolute value of the measured low frequency band power or high frequency band power by the difference obtained by subtracting the very low frequency band power from the total power, and then multiply by 100.

[0050] The normal value range of the power spectrum recorded after lying quietly for 5 minutes is as follows:

[0051] Total power: 3466 ± 1018 ms2 / Hz.

[0052] Very low frequency band power: 1170 ± 416 ms2 / Hz.

[0053] Normalized low frequency band power: 54 ± 4 nu.

[0054] High frequency band power: 975 ± 203 ms2 / Hz.

[0055] Normalized high frequency band power: 29 ± 3 nu.

[0056] The ratio of low frequency band power to high frequency band power: 1.5 - 2.0.

[0057] The long-term task is used to evaluate the comprehensive autonomic nerve activity throughout the day, can capture the circadian rhythm and long-term physiological changes, and can reflect the overall situation of the subject. When selecting statistical indicators of human heart rate variability, total power, ultra-low frequency (ULF, frequency band ≤ 0.003Hz) power, very low frequency band power, low frequency band power, and high frequency band power can be used.

[0058] Optionally, in addition to the human heart rate variability data, the status data further includes questionnaire data, which is obtained by the subject answering a preset questionnaire. The preset questionnaire includes at least one scale related to personality detection. Although there is a risk of data deviation in scale detection, it is still a relatively mature method for quantitatively evaluating personality traits at present. By combining the human heart rate variability data with strong objectivity and authenticity with the questionnaire data including the scale detection results, multi-dimensional information can be provided, so as to conduct a comprehensive and comprehensive analysis and obtain a more accurate and detailed personality analysis result. By adopting this multi-level and non-invasive evaluation method, personality traits can be identified and analyzed more effectively, which not only brings the possibility of personalized mental health intervention, but also provides a scientific basis for personalized disease treatment.

[0059] As an example, the questionnaire data may include the detection results of each scale, such as the scores of each scale, or the personalities detected by each scale. It may also include the answers given by the subject to the specified questions in each scale. When multiple scales are used, the corresponding detection results can be spliced into a vector as the questionnaire data. Of course, the questionnaire data may also include the answering data of other preset questionnaires other than the scales, and the present disclosure does not limit this.

[0060] Optionally, the questionnaire data includes the scores of each scale in at least one scale related to personality detection. Step S102 includes: performing time-frequency domain feature extraction processing on the human heart rate variability data corresponding to each detection task to obtain the time domain features and frequency domain features corresponding to each detection task; performing fusion processing on the time domain features, frequency domain features corresponding to each detection task and the scores of each scale to obtain a fusion vector; inputting the fusion vector into a personality detection model to obtain predicted personality information. By fusing the time domain features and frequency domain features corresponding to the human heart rate variability data with the scores of each scale, a fusion vector containing rich information can be obtained, which is convenient for the personality detection model to comprehensively analyze and process the overall information, so as to learn the implicit relationship between different elements, which helps to improve the accuracy of the personality detection result. As an example, the fusion processing may be splicing processing, that is, splicing these data into a fusion vector, or using a fusion model to process these data to obtain a fused feature vector, and the present disclosure does not limit this.

[0061] Optionally, the questionnaire data still includes the scores of each of the above scales, and the personality detection model is trained through the following steps: obtaining the status data of the reference object; using the personality detection model to be trained to process the status data of the reference object to obtain the predicted scale scores and predicted personality information of the reference object, where the predicted scale scores are the predicted values of the scores of the reference object answering at least one scale related to personality detection; determining the value of the weighted cross-entropy loss function as the first loss value according to the predicted personality information and the true personality information of the reference object, where the weighted cross-entropy loss function is obtained by adding a personality category weight to the cross-entropy loss function, and the personality category weight is negatively correlated with the proportion of the personality category corresponding to the true personality information of the reference object in all training samples; determining the value of the correlation loss function as the second loss value according to the predicted scale scores of the reference object and the scores of the corresponding scales in the status data, where the correlation loss function is the difference between a preset constant and a preset correlation coefficient; determining the total loss value according to the first loss value and the second loss value; and adjusting the parameters of the personality detection model to be trained based on the total loss value to obtain the personality detection model.

[0062] This embodiment specifically describes the loss function used in the training process. On the one hand, based on the comparison between the conventional prediction results (i.e., predicted personality information) and sample labels (i.e., true personality information) and the use of the cross-entropy loss function, by increasing the personality category weight negatively correlated with the proportion of the true personality category (i.e., the personality category corresponding to the true personality information) in all training samples, that is, on the basis of calculating the calculated value of each training sample using the existing cross-entropy loss function, then multiplying by the corresponding personality category weight, and taking the obtained product as the new calculated value of each training sample, and then calculating the average value of the new calculated values of all training samples as the first loss value. Such a processing method can weaken the influence of a single training sample of the personality category with a high proportion on the first loss value, so as to balance the influence of training samples of different categories when there are significant differences in the number of samples of different personality categories. On the other hand, considering the correlation between human heart rate variability data and scale scores, by making the personality detection model additionally output the predicted scores of each scale during the training phase and calculating the correlation measure (i.e., the value of the preset correlation coefficient) between the predicted scale scores and the true scale scores (i.e., the scores of each scale included in the state data), and then obtaining the difference between the preset constant and the correlation measure as the second loss value, it can guide the model to make the predicted scale scores approach the true scale scores during the training process, so that the model can learn the correlation between human heart rate variability data and scale scores, which helps to improve the accuracy of personality prediction. It should be understood that in the inference phase, that is, when applying the trained personality detection model for personality detection, there is no need to output the predicted scale scores anymore. By comprehensively using the first loss value and the second loss value, the advantages of the above two aspects can be taken into account, thus significantly improving the objectivity and accuracy of personality prediction.

[0063] As an example, the personality category weight is, for example, but not limited to, the reciprocal of the proportion of the corresponding true personality category in all training samples; the preset correlation coefficient used in the correlation loss function is, for example, but not limited to, the Pearson correlation coefficient, the canonical correlation coefficient, the cosine similarity, and the preset constant used in the correlation loss function is, for example, but not limited to 1; when determining the total loss value, for example, but not limited to, calculating the weighted sum value of the first loss value and the second loss value, so as to flexibly adjust the influence degree of the first loss value and the second loss value by adjusting the weight.

[0064] Optionally, at least one scale related to personality detection includes at least one of the following: Type A Behavior Pattern Questionnaire, D Personality Questionnaire, Trait Coping Style Questionnaire, Self-Rating Depression Scale, Self-Rating Anxiety Scale, Anger and Anger Expression Scale, Comprehension of Social Support Scale, Obsessive Response Scale. By using at least one of the above preset scales, the personality of the target object can be measured by scale from different angles, thus providing rich information.

[0065] Specifically, the Type A Behavior Pattern Questionnaire (TABP) was developed under the auspices of Professor Zhang Boyuan. It consists of 60 questions and is divided into three parts: the Time Hurriedness (TH, 25 questions), which represents a sense of time urgency, time pressure, and doing things quickly; the Competition and Hostility (CH, 25 questions), which includes characteristics such as being competitive, suspicious or hostile, and lacking patience; and the Lie (L, 10 questions). It uses a scoring system of yes (1) or no (0), and the score is the sum of CH and TH. The correlation between the ratings by others and self - ratings of this scale is 0.57, and the test - retest reliability at an interval of 2 months is 0.51.

[0066] Diagnostic criteria have been determined according to the Chinese norm. Scores from 1 - 18 indicate Type B, 19 - 26 indicate slightly Type B, 27 - 28 indicate the intermediate type, 29 - 36 indicate slightly Type A, and 37 - 50 indicate Type A.

[0067] The Type D Personality Scale (DS14) was developed by Denollet et al. in 1998 as a tool to assess Type D personality. It was revised to DS14 in 2005 and consists of 14 questions, with 7 questions each to measure Negative Affectivity (NA) and Social Inhibition (SI), such as "I often feel unhappy", "I often make a mountain out of a molehill over trivial matters", "I feel very constrained in social interactions", etc. It uses a five - level scoring system from 0 (highly inconsistent) to 4 (highly consistent).

[0068] The score ranges of both the NA and SI sub - scales are from 0 to 28. When using it, a cut - off point of 10 is used. Subjects with scores of NA and SI both ≥ 10 are considered to have a tendency towards Type D personality.

[0069] The Trait Coping Style Questionnaire (TCSQ) was developed by Professor Jiang Qianjin as the first self - developed coping scale in China to evaluate the relatively stable coping strategies related to personality traits of the subjects. This questionnaire has a certain cross - situational consistency and is related to physical and mental health. It includes 20 items, and the subjects make a five - level choice response of 1 - 5 for each item. The scale includes two dimensions: Negative Coping (NC) and Positive Coping (PC).

[0070] The Self-Rating Depression Scale (SDS) is the most commonly used depression assessment scale in clinical practice and is one of the scales recommended by the US Department of Education, Health and Welfare. The scale consists of 20 items, reflecting emotional states (such as crying, depression, etc.), physical functions (such as loss of appetite, easy fatigue), mental dullness (mental confusion, agitation), and psychological disorders (such as emptiness, self-deprecation, suicidal tendency, etc.). It is scored on a four-point scale from 1 (none) to 4 (persistent). Calculate the depression index, which ranges from 0.25 to 1.0. The higher the index, the more severe the depression.

[0071] The Self-Rating Anxiety Scale (SAS) is the most commonly used anxiety assessment scale in clinical practice, including 20 items, scored on a four-point scale from 1 (none) to 4 (persistent). When tested simultaneously with the Hamilton Anxiety Scale (HAMA), the Pearson correlation coefficient of the total scores of the two scales is 0.365, and the Spearman rank correlation coefficient is 0.341, indicating good validity.

[0072] Regarding the anger and anger expression scale, the emotion of anger is closely related to the occurrence and development of coronary heart disease. Many prospective studies have shown that anger is likely to trigger coronary heart disease and affect the prognosis of coronary heart disease. To assess the anger emotion and anger expression of the subjects, the anger subscale in the Type C Behavior Scale (CB) and the two subscales of anger-in and anger-out were selected. This scale was introduced into China by Professor Zhang Yao and is widely used in clinical practice. The anger scale contains 10 items, and both the anger-in and anger-out scales contain 6 items. Each question is answered on a four-point scale from 1 (almost never) to 4 (often).

[0073] Regarding the Perceived Social Support Scale (PSSS), social support is an important mediating factor influencing the relationship between psychological stress and physical and mental health. Social support includes objective, visible, or actual support, as well as the respect, support, and understanding perceived by individuals. The Perceived Social Support Scale (PSSS) mainly assesses the social support perceived by individuals. This scale was adapted by Blumenthal et al. and introduced and revised by Jiang Qianjin et al. It measures the respect, understanding, and support that individuals subjectively experience from family, friends, and others. Domestic research shows that the social support scale has good reliability and validity. This scale contains 12 items and is divided into two subscales: in-family support and out-of-family support. Specific items include "I share joys and sorrows with some people", "My family can actually and specifically help me", "I can discuss my problems with my friends". Each item uses a seven-point rating method from 1 to 7.

[0074] Regarding the Ruminative Responses Scale (RRS), research shows that rumination is related to the generation and maintenance of various negative emotions. In particular, anxiety, depression, and anger are closely related to rumination and are important cognitive factors in the process of stress coping. To evaluate the rumination response of the object, the most commonly used Ruminative Responses Scale is adopted in this disclosure. This scale is a subscale of the Ruminative Style Questionnaire (RSQ) and mainly assesses the degree of attention of the object to the symptoms, meaning, causes, and consequences of a sad state, which is an important cognitive factor for the maintenance and intensification of depressive emotions (Nolen-Hoeksema, 1991). This scale contains 22 items, and each item is rated on a four-point scale (1 never or occasionally; 2 sometimes; 3 often; 4 always), mainly involving the attention and evaluation of the symptoms, meaning, causes, and consequences of sad and depressive emotions.

[0075] Optionally, the preset questionnaire further includes a psychosomatic disease questionnaire. Since the psychosomatic health status of the subject is related to personality, coping style, emotion, and cognition, and will also affect the emotional state, in order to study the psychological mechanism of the influence of Type A behavior and Type D personality on the prognosis of coronary heart disease, it is necessary to understand the psychosomatic health status of the subjects. For this purpose, according to common typical psychosomatic diseases (hypertension, diabetes, coronary heart disease, ulcer disease, neurodermatitis, allergy), a psychosomatic disease questionnaire was compiled. The survey content mainly includes the psychosomatic disease history of the subjects and the psychosomatic disease history of their families. At the same time, the smoking and drinking situations were investigated. It should be understood that for the questionnaire including the psychosomatic disease questionnaire, or other non-scale types and questionnaires only used for information collection, the information collected may not participate in the processing of the personality detection model, but only be saved as background information, or it can be further extracted to obtain key data and added to the questionnaire data to participate in the processing of the personality detection model. For the latter, the specific participation method is, for example, when obtaining the fusion vector introduced above, making the key data also participate in the fusion processing to make it a part of the model input; or taking the population to which the object belongs as the key data, such as healthy people and people with psychosomatic diseases, and training personality detection models for different populations respectively, and then determining the applicable personality detection model according to the key data of the target object during personality detection, and using the determined model to perform personality detection. The present disclosure does not limit the specific use of the information collected by the information collection questionnaire.

[0076] Next, a specific embodiment of a detection task is introduced. In this specific embodiment, the detection task is a short-term task and includes four detection tasks in different states. In addition, there is a significant correlation between the two scale dimensions of Type A personality and Type D personality, but the connotations of the dimensions they measure are very different. In order to more accurately distinguish different personalities, according to the two dimensions of the Type A behavior type questionnaire (i.e., Type A and Type B) and the two dimensions of the Type D personality scale (i.e., Type D and type d, where type d refers to non-Type D), the subjects are divided into four groups: AD group, BD group, Ad group, and Bd group. It should be understood that the division into four groups here is for the convenience of briefly distinguishing and describing the task performances of different objects, and it is not necessary to first determine the group to which the target object belongs according to the human heart rate variability data when performing the personality detection method of the exemplary embodiment of the present disclosure.

[0077] 1. Detection task in the resting state.

[0078] As Figure 2 shown, ask the target object to lean on a high-back chair with armrests and close his eyes and rest for ten minutes, and then use a millimeter-wave radar to detect and record his human heart rate variability data. Rest for five minutes and record the second human heart rate variability data.

[0079] In this task, the AD group was lower than the BD group in terms of TP. The TP, VLF, and LF of D personality subjects were all lower than those of d personality subjects.

[0080] 2. Abdominal breathing state detection task.

[0081] As Figure 3 shown, ask the target subject to lean on a high-back chair with armrests and close their eyes to rest for ten minutes. Instruct the target subject to practice abdominal breathing seven times per minute. After one minute of practice, ask the target subject to perform deep abdominal breathing for five minutes. Use a millimeter-wave radar to detect and record the human heart rate variability data. After resting for five minutes, repeat the above steps to record the second set of human heart rate variability data.

[0082] In this task, there was a significant difference between the AD group and the BD group in terms of HF in the abdominal breathing state. The HF of AD-type subjects was significantly lower than that of BD-type subjects, and there was a marginal significance in terms of LFnorm and HFnorm. The TP and HF of D personality subjects were lower than those of d personality subjects.

[0083] 3. Passive stress state detection task after rest.

[0084] As Figure 4 shown, ask the target subject to lean on a high-back chair with armrests and close their eyes to rest for ten minutes. Ask the target subject to watch the sad emotion induction video material for five minutes. Use a millimeter-wave radar to detect and record the human heart rate variability data. After resting for five minutes, repeat the above steps, replace the sad emotion induction video material, and record the second set of human heart rate variability data.

[0085] In this task, the AD group was marginally significant in terms of LF / HF, LFnorm, and HFnorm. It was higher than the BD group in terms of HFnorm and lower than the BD group in terms of heart rate change, LF / HF, and LFnorm. For young men, the LF / HF and LFnorm of D personality subjects were higher than those of d personality subjects, and the HFnorm of D personality subjects was lower than that of d personality subjects.

[0086] 4. Passive stress state detection task after abdominal breathing.

[0087] As Figure 5 shown, ask the target subject to lean on a high-back chair with armrests and close their eyes to rest for ten minutes. Instruct the target subject to practice abdominal breathing seven times per minute. After one minute of practice, ask the target subject to perform deep abdominal breathing for five minutes. Ask the target subject to watch the sad emotion induction video material for five minutes. Use a millimeter-wave radar to detect and record the human heart rate variability data. After resting for five minutes, repeat the above steps, replace the sad emotion induction video material, and record the second set of human heart rate variability data.

[0088] In this task, the change in heart rate of the Type D personality subjects before and after watching the video is greater than that of the type d personality.

[0089] Next, a specific embodiment is combined to introduce the personality detection method according to the exemplary embodiment of the present disclosure.

[0090] This specific embodiment mainly includes the following steps:

[0091] 1. Data acquisition: Use a millimeter-wave radar to collect the heart rate signals of the target subject under different detection tasks.

[0092] 2. Signal processing: Extract the time-domain features and frequency-domain features of the human heart rate variability data.

[0093] 3. Feature fusion: Map the time-frequency domain features and the scale scores to form a fusion vector.

[0094] 4. Personality detection: Use the trained personality detection model to process the fusion features to obtain the predicted personality traits.

[0095] Next, each will be introduced one by one.

[0096] 1. Data acquisition step.

[0097] In order to stimulate different physiological responses of the target subject, the following multiple detection tasks are designed.

[0098] Rest state detection task (Task 1): Sit quietly and maintain natural breathing.

[0099] Abdominal breathing state detection task (Task 2): Guide the target subject to perform abdominal deep breathing.

[0100] Passive stress state detection task (Task 3): Trigger the stress response of the target subject through emotional induction materials (such as stress pictures or sounds).

[0101] During the execution of each detection task, use a millimeter-wave radar device to non-contactlessly obtain the heart rate signal of the target subject. The millimeter-wave radar detects the minute chest wall displacement caused by the heartbeat by transmitting and receiving electromagnetic waves.

[0102] 2. Signal processing step.

[0103] This step further includes 4 sub-steps.

[0104] 1) Preprocess the collected heart rate signal. Specifically, extract the heart rate time series {R(t)} from the original signal collected by the millimeter-wave radar, and use a filter to remove noise and enhance the signal quality.

[0105] 2) Calculate the heart rate interval (R-R interval) sequence using the following formula:

[0106] RR(n)=R(t n+1 ) - R(t n )。

[0107] 3) Extract time-domain features, specifically calculate the following time-domain features.

[0108] Average heart rate interval:

[0109]

[0110] Standard Deviation of Normal to Normal intervals (SDNN):

[0111]

[0112] Root Mean Square of the Successive Differences (RMSSD):

[0113]

[0114] 4) Extract frequency-domain features. Specifically, first perform a Fast Fourier Transform (FFT) on the heart rate interval sequence RR(n), and then calculate the power of different frequency bands.

[0115] Performing a Fast Fourier Transform on RR(n) can be expressed as:

[0116] PSD(f) = |F{RR(n)}| 2 。

[0117] The power of the very low frequency band (VLF, 0.003 - 0.04 Hz) can be expressed as:

[0118]

[0119] The power of the low frequency band (LF, 0.04 - 0.15 Hz) can be expressed as:

[0120]

[0121] The power of the high frequency band (HF, 0.15 - 0.4 Hz) can be expressed as:

[0122]

[0123] The ratio of the low frequency band power to the high frequency band power (LF / HF) can be expressed as:

[0124]

[0125] 3. Feature fusion.

[0126] On the one hand, it is necessary to obtain the scores of the target user's answers to each scale, including: Type A Behavior Pattern Questionnaire (score S A ), Type D Personality Questionnaire (score S D ), Trait Coping Style Questionnaire (score S C ), Self-Rating Depression Scale (score S Dep ), Self-Rating Anxiety Scale (score S Anx ).

[0127] On the other hand, it is necessary to construct a physiological feature vector (i.e., the feature vector of human heart rate variability data, which will not be explained one by one below). First, construct the feature vector X i :

[0128]

[0129] Then merge the feature vectors of all detection tasks into X:

[0130] X = [X 1 , X 2 , …, X M ,

[0131] where M is the number of tasks.

[0132] Fuse the features of these two aspects to form the fusion vector Z:

[0133] Z = [X, S A , S D , S C , S Dep , S Anx .

[0134] 4. Personality detection.

[0135] In this step, a Deep Neural Network (DNN) is used as the personality detection model, which contains multiple fully connected layers and activation functions. The input of the model is the fusion vector Z, and the output is the predicted personality information

[0136] The personality detection model is obtained through supervised training. Before starting the training, it is necessary to obtain a number of training samples (each training sample includes the status data of a reference object and the true personality information representing the true personality category of the reference object), and randomly divide these training samples into two data sets: a training set and a test set. The former is used to train the model, and the latter is used to test the trained model.

[0137] Next, the loss function and parameter update algorithm used during training are introduced.

[0138] Regarding the loss function, in order to better adapt to the personality detection task, this specific embodiment designs a composite loss function that combines weighted cross-entropy loss and correlation loss.

[0139] Regarding the weighted cross-entropy loss, since there may be class imbalance in the training set, the weighted cross-entropy loss function is introduced and defined as follows:

[0140]

[0141] where is the true personality class y of the k-th training sample (k) of the class weight, which is used to balance the influence of different personality classes and is defined as:

[0142]

[0143] where represents the prior probability of class y (k) , that is, the sample proportion of this class in the training set, which is the ratio of the number of samples with the true personality class y (k) to the total number of samples in the training set.

[0144] Regarding the correlation loss, considering the correlation between physiological characteristics and scale scores, this specific embodiment introduces correlation loss to improve the effectiveness of the model in fusing multi-modal data. The correlation loss is defined as:

[0145]

[0146] where is the Pearson correlation coefficient between the true scale score S (k) (i.e., the scale score in the state data) and the predicted scale score obtained by the model, and is calculated as follows:

[0147]

[0148] where is the mean of the true scale scores, is the mean of the predicted scale scores.

[0149] Combining the above two loss functions, the total loss function is defined as:

[0150] L(θ) = L wCE (θ) + λL corr (θ).

[0151] Among them, λ is a trade-off parameter used to adjust the ratio between the weighted cross-entropy loss and the correlation loss.

[0152] When updating the parameters, the backpropagation algorithm is used to update the model parameters, and the goal is to minimize the total loss function:

[0153]

[0154] where η is the learning rate, is the gradient of the total loss function with respect to the parameters.

[0155] After the parameters are adjusted, the test set is used for testing, and indicators such as accuracy, precision, recall, and F1 score are used to evaluate the model performance. A high accuracy is achieved, which proves the effectiveness of fusing physiological characteristics and scale scores.

[0156] This specific embodiment proposes a method for detecting personality categories based on millimeter-wave radar and deep learning. Through non-intrusive physiological signal acquisition, feature extraction and fusion, and the construction and application of a deep learning model, objective and accurate prediction of personality categories is achieved.

[0157] Figure 6 is a block diagram of a personality detection device according to an exemplary embodiment of the present disclosure. Referring to Figure 6 , the personality detection device 600 includes an acquisition unit 601 and a processing unit 602.

[0158] The acquisition unit 601 can acquire the state data of the target object. Among them, the state data includes the human heart rate variability data of the corresponding object detected in the task state, and the human heart rate variability data is detected using a millimeter-wave radar.

[0159] The processing unit 602 can use the personality detection model to process the state data of the target object to obtain the predicted personality information of the target object. Among them, the personality detection model is trained by a supervised method.

[0160] Optionally, the detection tasks include at least one of the following: rest state detection task, abdominal breathing state detection task, passive stress state detection task. Among them, the passive stress state detection task includes at least one of the following: passive stress state detection task after rest, passive stress state detection task after abdominal breathing. The passive stress state detection task is a task that induces a stress response in the corresponding object through emotional induction materials.

[0161] Optionally, the state data further includes questionnaire data, which is obtained by the object answering a preset questionnaire. The preset questionnaire includes at least one scale related to personality detection.

[0162] Optionally, the questionnaire data includes the scores of each scale in at least one scale related to personality detection, and the processing unit is further configured to: perform time-frequency domain feature extraction processing on the human heart rate variability data corresponding to each detection task to obtain the time domain features and frequency domain features corresponding to each detection task; perform fusion processing on the time domain features, frequency domain features corresponding to each detection task, and the scores of each scale to obtain a fusion vector; input the fusion vector into the personality detection model to obtain predicted personality information.

[0163] Optionally, the questionnaire data includes the scores of each scale in the at least one scale related to personality detection, and the training samples of the personality detection model include the state data and true personality information of a reference object. Among them, the personality detection model is trained through the following steps: obtaining the state data of the reference object; using the personality detection model to be trained to process the state data of the reference object to obtain the predicted scale scores and predicted personality information of the reference object, where the predicted scale scores are the predicted values of the scores of the reference object answering the at least one scale related to personality detection; determining the value of the weighted cross-entropy loss function as the first loss value according to the predicted personality information and true personality information of the reference object, where the weighted cross-entropy loss function is obtained by adding a personality category weight on the basis of the cross-entropy loss function, and the personality category weight is negatively correlated with the proportion of the personality category corresponding to the true personality information of the reference object in all training samples; determining the value of the correlation loss function as the second loss value according to the predicted scale scores of the reference object and the scores of the corresponding scales in the state data, where the correlation loss function is the difference between a preset constant and a preset correlation coefficient; determining the total loss value according to the first loss value and the second loss value; adjusting the parameters of the personality detection model to be trained based on the total loss value to obtain the personality detection model.

[0164] Optionally, at least one scale related to personality detection includes at least one of the following: Type A Behavior Pattern Questionnaire, Type D Personality Questionnaire, Trait Coping Style Questionnaire, Self-Rating Depression Scale, Self-Rating Anxiety Scale, Anger and Anger Expression Scale, Perceived Social Support Scale, Addiction Response Scale; and / or the preset questionnaire further includes a psychosomatic disease questionnaire.

[0165] Regarding the device in the above embodiments, the specific manners in which each unit performs operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0166] Figure 7 The structural block diagram of an electronic device according to an exemplary embodiment of the present disclosure is shown.

[0167] Refer to Figure 7, the electronic device 700 includes: at least one memory 701 and at least one processor 702. Computer-executable instructions are stored in the at least one memory 701. When the computer-executable instructions are run by the at least one processor 702, the at least one processor is caused to execute the target corresponding method as described in the above exemplary embodiments.

[0168] As an example, the electronic device 700 can be a PC computer, a tablet device, a personal digital assistant, a smart phone, or other devices capable of executing the above instruction set. Here, the electronic device 700 does not have to be a single electronic device 700, and can also be an aggregate of devices or circuits that can individually or jointly execute the above instructions (or instruction sets). The electronic device 700 can also be a part of an integrated control system or a system manager, or can be configured as a portable electronic device 700 that is interconnected with a local or remote (e.g., via wireless transmission) interface.

[0169] In the electronic device 700, the processor 702 can include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller, or a microprocessor. As an example but not a limitation, the processor 702 can also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.

[0170] The processor 702 can run the instructions or code stored in the memory 701. Among them, the memory 701 can also store data. The instructions and data can also be sent and received via the network interface device through the network, where the network interface device can adopt any known transmission protocol.

[0171] The memory 701 can be integrated with the processor 702. For example, RAM or flash memory can be arranged within an integrated circuit microprocessor, etc. In addition, the memory 701 can include independent devices, such as external disk drives, storage arrays, or other storage devices that can be used by any database system. The memory 701 and the processor 702 can be operatively coupled, or can communicate with each other, for example, through I / O ports, network connections, etc., so that the processor 702 can read the files stored in the memory.

[0172] In addition, the electronic device 700 can also include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, a mouse, a touch input device, etc.). All components of the electronic device 700 can be connected to each other via a bus and / or a network.

[0173] According to an exemplary embodiment of the present disclosure, a computer-readable storage medium storing instructions may also be provided, wherein, when the instructions are run by at least one processor, the at least one processor is caused to execute the target corresponding method as described in the above exemplary embodiment. Examples of such computer-readable storage media include: read-only memory (ROM), programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-R LTH, BD-RE, Blu-ray or optical disc memory, hard disk drive (HDD), solid state drive (SSD), cartridge memory (such as, multimedia card, secure digital (SD) card or extreme digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk, and any other device configured to store a computer program and any associated data, data files, and data structures in a non-transitory manner and to provide the computer program and any associated data, data files, and data structures to a processor or computer such that the processor or computer can execute the computer program. The computer program in the above computer-readable storage medium may run in an environment deployed in computer devices such as clients, hosts, proxy devices, servers, etc. In addition, in one example, the computer program and any associated data, data files, and data structures are distributed on a networked computer system such that the computer program and any associated data, data files, and data structures are stored, accessed, and executed in a distributed manner by one or more processors or computers.

[0174] According to an exemplary embodiment of the present disclosure, a computer program product may also be provided, including computer instructions that, when run by at least one processor, execute the target corresponding method as described in the above exemplary embodiment.

[0175] Those skilled in the art will readily conceive of other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include known common knowledge or conventional technical means in the technical field not disclosed by the present disclosure. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of the present disclosure are pointed out by the appended claims.

[0176] It should be understood that the present disclosure is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is only limited by the appended claims.

Claims

1. A personality detection method, characterized in that: The personality detection method comprises: Acquiring state data of the target object, wherein the state data includes human heart rate variability data of the object obtained by detecting by a millimeter wave radar when the object is in a task state of performing a detection task, and the state data also includes questionnaire data, the questionnaire data is obtained by the object answering a preset questionnaire, the preset questionnaire includes at least one scale related to personality detection, and the questionnaire data includes a score of each scale in the at least one scale related to personality detection; Using the personality detection model, processing the state data of the target object to obtain the predicted personality information of the target object, The training samples of the personality detection model include the state data and real personality information of the reference object, and the personality detection model is trained through the following steps: Acquiring status data of the reference object; Using the personality detection model to be trained, processing the state data of the reference subject, obtaining a predicted scale score and predicted personality information of the reference subject, wherein the predicted scale score is a predicted value of the score of the reference subject answering the at least one scale related to personality detection; Determine, according to the predicted personality information and the real personality information of the reference object, a value of a weighted cross entropy loss function as a first loss value, wherein the weighted cross entropy loss function is obtained by adding a personality category weight to the cross entropy loss function, and the personality category weight is negatively correlated with a proportion of the personality category corresponding to the real personality information of the reference object in all training samples; Determining a value of a correlation loss function as a second loss value according to the predicted scale score of the reference object and the scores of each corresponding scale in the status data, wherein the correlation loss function is a difference between a preset constant and a preset correlation coefficient; Determining a total loss value according to the first loss value and the second loss value; Based on the total loss value, the parameters of the personality detection model to be trained are adjusted to obtain the personality detection model.

2. The personality detection method according to claim 1, characterized in that: The detection task includes at least one of the following: a resting state detection task, an abdominal breathing state detection task, and a passive stress state detection task, wherein the passive stress state detection task includes at least one of the following: a passive stress state detection task after rest, and a passive stress state detection task after abdominal breathing. The passive stress state detection task is a task that induces a stress response of the corresponding object through emotion-inducing materials.

3. The personality detection method according to claim 1 or 2, characterized in that: The at least one scale related to personality testing includes at least one of the following: Type A Behavior Type Questionnaire, Type D Personality Questionnaire, Trait Coping Style Questionnaire, Self-Rating Depression Scale, Self-Rating Anxiety Scale, Anger and Anger Expression Scale, Perceptual Social Support Scale, and Indulgent Reaction Scale; and / or The preset questionnaire also includes a psychosomatic disease questionnaire.

4. The personality detection method according to claim 1 or 2, characterized in that: The using the personality detection model to process the state data of the target object to obtain the predicted personality information of the target object includes: Perform time-domain feature extraction and processing on the human heart rate variability data corresponding to each detection task to obtain the time-domain features and frequency-domain features corresponding to each detection task; The time domain features and frequency domain features corresponding to each detection task and the scores of each scale are fused to obtain a fusion vector; The fusion vector is input into the personality detection model to obtain the predicted personality information.

5. A personality detection device, characterized in that: The personality detection device comprises: an acquisition unit configured to acquire state data of a target object, wherein the state data includes human heart rate variability data of a corresponding object detected in a task state, the human heart rate variability data is obtained by using a millimeter wave radar for detection, the state data also includes questionnaire data, the questionnaire data is obtained by the object answering a preset questionnaire, the preset questionnaire includes at least one scale related to personality detection, and the questionnaire data includes a score of each scale in the at least one scale related to personality detection; a processing unit configured to use a personality detection model to process the state data of the target object to obtain predicted personality information of the target object, The training samples of the personality detection model include the state data and real personality information of the reference object, and the personality detection model is trained through the following steps: Acquiring status data of the reference object; Using the personality detection model to be trained, processing the state data of the reference subject, obtaining a predicted scale score and predicted personality information of the reference subject, wherein the predicted scale score is a predicted value of the score of the reference subject answering the at least one scale related to personality detection; Determine, according to the predicted personality information and the real personality information of the reference object, a value of a weighted cross entropy loss function as a first loss value, wherein the weighted cross entropy loss function is obtained by adding a personality category weight to the cross entropy loss function, and the personality category weight is negatively correlated with a proportion of the personality category corresponding to the real personality information of the reference object in all training samples; Determining a value of a correlation loss function as a second loss value according to the predicted scale score of the reference object and the scores of each corresponding scale in the status data, wherein the correlation loss function is a difference between a preset constant and a preset correlation coefficient; Determining a total loss value according to the first loss value and the second loss value; Based on the total loss value, the parameters of the personality detection model to be trained are adjusted to obtain the personality detection model.

6. An electronic device, characterized in that: include: at least one processor; at least one memory storing computer executable instructions, Wherein, when the computer executable instructions are executed by the at least one processor, the at least one processor is prompted to execute the personality detection method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by at least one processor, the at least one processor is prompted to perform the personality detection method according to any one of claims 1 to 4.

8. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by at least one processor, the at least one processor is prompted to perform the personality detection method according to any one of claims 1 to 4.

Images