A mental diagnosis system based on convolutional neural network and facial recognition

Abstract

The present invention relates to the technical field of mental diagnosis, and specifically discloses a mental diagnosis system based on convolutional neural network and face recognition, including: a mental state analysis module for analyzing whether the mental state of a target patient is abnormal; a face recognition module for analyzing whether the mental abnormality of the patient is related to the target facial features; an early warning module for detecting the facial features of the patient and predicting the mental state of the patient; by comprehensively analyzing the behaviors, emotions and physiological characteristic values of the patient, as well as the facial features, the present invention can more accurately reflect the mental state of the patient, providing the possibility for personalized diagnosis, helping doctors formulate more accurate treatment plans according to the specific conditions of different patients, and improving the reliability of diagnosis; by continuously detecting the facial features of the patient, the trend of mental abnormality can be detected in time and an alarm can be generated to remind medical staff to intervene in time, which helps to prevent the deterioration of mental diseases and reduce the occurrence of accidents.

Description

Technical Field

[0001] The present invention relates to the technical field of mental diagnosis, and particularly to a mental diagnosis system based on convolutional neural network and face recognition. Background Art

[0002] Mental illness is a complex and broad concept that encompasses a variety of diseases affecting brain function. It is a disease in which, under the influence of various biological, psychological, and social environmental factors, the brain function is disordered, resulting in varying degrees of disorders in mental activities such as cognition, emotion, will, and behavior. Mental illness can be roughly divided into mild mental illness and severe mental illness. Mild mental illness mainly includes anxiety disorders, obsessive-compulsive disorder, depression, and phobias, etc. These diseases are mainly manifested as emotional distress and one-sided thinking and judgment under the domination of strong emotions, but the patient's cognitive, logical reasoning ability, and self-awareness are usually basically intact. While severe mental illness, such as schizophrenia and bipolar disorder, may involve more serious decline in cognitive and logical reasoning ability, and even loss of self-awareness. In addition, mental illness also includes many other types, such as organic mental disorders (such as Alzheimer's disease, vascular dementia, etc.), which are usually caused by factors such as central nervous system diseases, metabolic disorders, and brain injuries; mental and behavioral disorders caused by the use of psychoactive substances (such as alcohol, drugs, etc.); mood disorders, such as manic episodes, bipolar disorder, and depression, etc.; and behavioral and emotional disorders that onset in childhood and adolescence, such as autism, tic disorders, etc.

[0003] The Chinese invention patent with the publication number CN116386845A discloses a schizophrenia diagnosis system based on convolutional neural network and facial dynamic video, including a video data acquisition unit; a video data preprocessing unit; a convolutional neural network unit; and a visualization unit.

[0004] However, traditional mental diagnosis methods often rely on doctors' experience and direct observation of patients. This method is not only time-consuming and laborious, but also easily affected by subjective factors, resulting in inaccurate diagnosis results, and traditional mental diagnosis methods also lack consideration of individual differences among patients. Summary of the Invention

[0005] The purpose of the present invention is to provide a mental diagnosis system based on convolutional neural network and face recognition to solve the above technical problems in the background.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A mental diagnosis system based on convolutional neural network and face recognition, comprising:

[0008] A mental state analysis module, which is used to analyze whether the mental state of a target patient is abnormal;

[0009] A facial recognition module, which is used to analyze whether the mental abnormality of the patient is related to the target facial features;

[0010] An early warning module, which is used to detect the facial features of the patient, so as to predict the mental state of the patient;

[0011] When the target facial features are related to mental abnormalities, obtain the fluctuation parameters of the target facial features of the patient , and according to the relevant parameters of the target facial features , calculate and obtain the early warning parameter YJ;

[0012] Through the formula: , calculate and obtain the early warning parameter YJ, where 、 are preset proportional parameters, and both are greater than 0. i represents the i-th target facial feature, and i takes positive integers;

[0013] The preset early warning threshold is YY, and the early warning parameter YJ is compared and analyzed with the early warning threshold YY;

[0014] If the early warning parameter YJ ≥ the early warning threshold YY, it means that the patient has a tendency of mental abnormality, and a mental abnormality alarm is generated.

[0015] As a further solution of the present invention: The method for analyzing whether the mental state of the target patient is abnormal is:

[0016] Obtain the mental state data of the target patient;

[0017] Calculate the mental state parameter according to the mental state data;

[0018] Compare and analyze the mental state parameters, so as to analyze whether the mental state of the target patient is abnormal.

[0019] As a further solution of the present invention: The method for obtaining the mental state data is:

[0020] Observe the behavior of the target patient and score the behavior of the target patient from 0 to 10. This score is the behavior characterization value, and the behavior characterization value is marked as XW;

[0021] Observe the emotion of the target patient and score the emotion of the target patient from 0 to 10. This score is the emotion characterization value, and the emotion characterization value is marked as QX;

[0022] By observing the physiological conditions of the target patient and scoring the physiological conditions of the target patient on a scale of 0-10, this score is the physiological characterization value, and the physiological characterization value is marked as SL.

[0023] As a further solution of the present invention: perform data processing on the behavior characterization value XW, the emotion characterization value QX, and the physiological characterization value SL, and through the formula: , calculate to obtain the mental state parameter JS, where is a preset proportionality factor, and both are greater than 0.

[0024] As a further solution of the present invention: preset the mental state threshold as JY, and compare and analyze the mental state parameter JS with the mental state threshold JY;

[0025] If the mental state parameter JS < the mental state threshold JY, it is determined that the mental state of the target patient is normal;

[0026] If the mental state parameter JS ≥ the mental state threshold JY, it is determined that the mental state of the target patient is abnormal.

[0027] As a further solution of the present invention: the method for obtaining the fluctuation parameter is:

[0028] Capture an image of the patient's face, obtain the degree and existence time of the target facial feature of the patient, and draw a target facial feature fluctuation graph with the X-axis being the existence time of the target facial feature and the Y-axis being the degree of the target facial feature, and obtain the area enclosed by the fluctuation line and the X-axis, and mark this area as the fluctuation parameter BD.

[0029] As a further solution of the present invention: preset the fluctuation parameter threshold as BY, and compare and analyze the fluctuation parameter BD with the fluctuation parameter threshold BY;

[0030] If the fluctuation parameter BD < the fluctuation parameter threshold BY, it indicates that the existence time of the target facial feature of the patient is short or the degree is low, that is, it is determined that the target facial feature is invalid;

[0031] If the fluctuation parameter BD ≥ the fluctuation parameter threshold BY, it indicates that the existence time of the target facial feature of the patient is long or the degree is high, that is, it is determined that the target facial feature is valid.

[0032] As a further solution of the present invention: the method for obtaining the correlation parameter is:

[0033] Based on multiple patients, obtain the proportion of patients with the target facial feature when the patients have mental abnormalities;

[0034] In the two-dimensional coordinate system with the mental state parameter as the X-axis and the proportion of the target facial feature as the Y-axis, construct a two-dimensional fluctuation graph of the mental state parameter - the proportion of the target facial feature;

[0035] The mental state parameters are equally divided into multiple intervals according to numerical values, and the area of the region enclosed by the wave line in each interval and the X-axis is calculated respectively. The ratio of this area to the area threshold is the relevant characterization value. , where n represents the nth region enclosed by the wave line in different intervals and the X-axis, and n takes positive integers;

[0036] Through the formula: , the relevant parameter XC is calculated and obtained.

[0037] As a further solution of the present invention: a relevant threshold CY is preset, and the relevant parameter XC is compared and analyzed with the relevant threshold CY;

[0038] If the relevant parameter XC < the relevant threshold CY, it is determined that the target facial feature has nothing to do with mental abnormality;

[0039] If the relevant parameter XC ≥ the relevant threshold CY, it is determined that the target facial feature is related to mental abnormality.

[0040] As a further solution of the present invention: a warning threshold YY is preset, and the warning parameter YJ is compared and analyzed with the warning threshold YY;

[0041] If the warning parameter YJ < the warning threshold YY, it indicates that the patient has no tendency of mental abnormality;

[0042] If the warning parameter YJ ≥ the warning threshold YY, it indicates that the patient has a tendency of mental abnormality, and a mental abnormality alarm is generated until the medical staff turns it off.

[0043] Advantages of the present invention:

[0044] (1) By comprehensively analyzing the patient's behavior, emotion, physiological characterization value, and facial feature, the present invention can more accurately reflect the patient's mental state, providing the possibility for personalized diagnosis, helping doctors formulate more precise treatment plans according to the specific conditions of different patients, and improving the reliability of diagnosis;

[0045] (2) By continuously detecting the patient's facial feature, the system can timely detect the tendency of mental abnormality and generate an alarm when necessary to remind the medical staff to intervene in time, which helps to prevent the deterioration of mental diseases and reduce the occurrence of accidents. Description of the Drawings

[0046] The present invention will be further described below with reference to the drawings.

[0047] Figure 1 is a schematic diagram of the system of the present invention;

[0048] Figure 2It is a flowchart for analyzing whether the mental state of the target patient is abnormal in the present invention;

[0049] Figure 3 It is a flowchart for analyzing whether the mental abnormality of the patient is related to the target facial features in the present invention. Specific embodiments

[0050] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0051] Embodiment 1

[0052] Please refer to Figure 1 、 Figure 2 As shown, the present invention is a mental diagnosis system based on convolutional neural network and facial recognition, including:

[0053] The mental state analysis module is used to analyze whether the mental state of the target patient is abnormal;

[0054] Among them, the mental state analysis module includes: a mental state acquisition unit, a mental state processing unit, and a mental state processing unit;

[0055] The mental state acquisition unit is used to obtain the mental state data of the target patient;

[0056] Among them, the mental state data includes: a behavior representation value, an emotion representation value, and a physiological representation value;

[0057] By observing the behavior of the target patient and rating the behavior of the target patient from 0 to 10, this rating is the behavior representation value, and the behavior representation value is marked as XW;

[0058] By observing the emotion of the target patient and rating the emotion of the target patient from 0 to 10, this rating is the emotion representation value, and the emotion representation value is marked as QX;

[0059] By observing the physiological condition of the target patient and rating the physiological condition of the target patient from 0 to 10, this rating is the physiological representation value, and the physiological representation value is marked as SL;

[0060] The physiological conditions include: insomnia, loss of appetite, hyperactivity, hypoactivity, and mental illness history;

[0061] The mental state processing unit is used to calculate the mental state parameters according to the mental state data;

[0062] Perform data processing on the behavioral representation value XW, the emotional representation value QX, and the physiological representation value SL. Through the formula: , calculate to obtain the mental state parameter JS, where are preset scale factors and are all greater than 0;

[0063] The mental state analysis unit is used to conduct a comparative analysis on the mental state parameter, so as to analyze whether the mental state of the target patient is abnormal;

[0064] The preset mental state threshold is JY, and a comparative analysis is conducted on the mental state parameter JS and the mental state threshold JY;

[0065] If the mental state parameter JS < the mental state threshold JY, it is determined that the mental state of the target patient is normal;

[0066] If the mental state parameter JS ≥ the mental state threshold JY, it is determined that the mental state of the target patient is abnormal.

[0067] Embodiment 2

[0068] Based on Embodiment 1, please refer to Figure 1 , Figure 3 As shown, the present invention is a mental diagnosis system based on a convolutional neural network and face recognition, and further includes:

[0069] The face recognition module is used to analyze whether the mental abnormality of the patient is related to the target facial features;

[0070] Among them, the face recognition module includes: a facial feature acquisition unit, a facial feature acquisition unit;

[0071] The facial feature acquisition unit is used to acquire the facial features of the patient and evaluate whether they are valid;

[0072] Capture an image of the patient's face, obtain the degree of the target facial features of the patient and the existence time of the target facial features, and draw a target facial feature fluctuation graph with the X-axis being the existence time of the target facial features and the Y-axis being the degree of the target facial features, and obtain the area enclosed by the fluctuation line and the X-axis, and mark this area as the fluctuation parameter BD;

[0073] The preset fluctuation parameter threshold is BY, and a comparative analysis is conducted on the fluctuation parameter BD and the fluctuation parameter threshold BY;

[0074] If the fluctuation parameter BD < the fluctuation parameter threshold BY, it indicates that the existence time of the target facial features of the patient is short or the degree is low, that is, it is determined that the target facial features are invalid;

[0075] If the fluctuation parameter BD ≥ the fluctuation parameter threshold BY, it indicates that the existence time of the target facial features of the patient is long or the degree is high, that is, it is determined that the target facial features are valid;

[0076] When the facial feature acquisition unit is used to analyze whether the target facial features of a patient are effective, it determines whether the target facial features are related to mental abnormalities;

[0077] Based on multiple patients, when a patient has mental abnormalities, obtain the proportion of patients with the target facial features among all patients;

[0078] In a two-dimensional coordinate system with the mental state parameter as the X-axis and the proportion of the target facial features as the Y-axis, construct a two-dimensional fluctuation graph of the mental state parameter - proportion of the target facial features;

[0079] Divide the mental state parameter into multiple intervals by value, and calculate the area of the region enclosed by each interval wave line and the X-axis respectively. The ratio of this area to the area threshold is the relevant characterization value , where n represents the nth region enclosed by the wave line of different intervals and the X-axis, and n takes positive integers;

[0080] Through the formula: , calculate and obtain the relevant parameter XC;

[0081] Preset the relevant threshold as CY, and compare and analyze the relevant parameter XC with the relevant threshold CY;

[0082] If the relevant parameter XC < the relevant threshold CY, it is determined that the target facial features are not related to mental abnormalities;

[0083] If the relevant parameter XC ≥ the relevant threshold CY, it is determined that the target facial features are related to mental abnormalities.

[0084] Embodiment 3

[0085] Based on Embodiment 1 and Embodiment 2, the present invention is a mental diagnosis system based on a convolutional neural network and facial recognition, and further includes:

[0086] The early warning module is used to detect the facial features of a patient, so as to predict the mental state of the patient;

[0087] When the target facial features are related to mental abnormalities, obtain the fluctuation parameters of the target facial features of the patient , and according to the relevant parameters of the target facial features , calculate and obtain the early warning parameter YJ;

[0088] Through the formula: , calculate and obtain the early warning parameter YJ, where 、 are preset proportional parameters, and are both greater than 0, i represents the ith target facial feature, and i takes positive integers;

[0089] The preset warning threshold is YY, and the warning parameter YJ is compared and analyzed with the warning threshold YY;

[0090] If the warning parameter YJ < the warning threshold YY, it indicates that the patient's mental state has no abnormal tendency;

[0091] If the warning parameter YJ ≥ the warning threshold YY, it indicates that the patient's mental state has an abnormal tendency, and a mental abnormality alarm is generated until the medical staff turns it off.

[0092] The working principle of the present invention:

[0093] The mental state analysis module is used to analyze whether the mental state of the target patient is abnormal;

[0094] Among them, the mental state analysis module includes: a mental state acquisition unit, a mental state processing unit, and a mental state processing unit;

[0095] The mental state acquisition unit is used to obtain the mental state data of the target patient;

[0096] The mental state processing unit is used to calculate the mental state parameter according to the mental state data;

[0097] The mental state analysis unit is used to compare and analyze the mental state parameter, so as to analyze whether the mental state of the target patient is abnormal;

[0098] The face recognition module is used to analyze whether the patient's mental abnormality is related to the target facial features;

[0099] Among them, the face recognition module includes: a facial feature acquisition unit, a facial feature acquisition unit;

[0100] The facial feature acquisition unit is used to collect the patient's facial features and evaluate whether they are effective;

[0101] The facial feature acquisition unit is used to analyze whether the target facial features are related to mental abnormalities when the target facial features of the patient are effective;

[0102] The warning module is used to detect the patient's facial features, so as to predict the patient's mental state;

[0103] When the target facial features are related to mental abnormalities, obtain the fluctuation parameter of the patient's target facial features and calculate the warning parameter YJ according to the relevant parameters of the target facial features ;

[0104] Through the formula: calculate the warning parameter YJ, where 、 is a preset proportional parameter, and both are greater than 0. i represents the i-th target facial feature, and i is a positive integer;

[0105] The preset warning threshold is YY, and the warning parameter YJ is compared and analyzed with the warning threshold YY;

[0106] If the warning parameter YJ ≥ the warning threshold YY, it indicates that the patient has a tendency of mental abnormality, and a mental abnormality alarm is generated.

[0107] The setting of the magnitude of the above threshold is for the convenience of comparison. Regarding the magnitude of the threshold, it depends on the amount of sample data and the base quantity set by those skilled in the art for each set of sample data;

[0108] The above formulas are all obtained by collecting a large amount of data for software simulation and selecting a formula close to the true value. The factors in the formula are set by those skilled in the art according to the actual situation; for example, the formula ; Those skilled in the art collect multiple sets of mental state data and set corresponding mental state parameters for each set of mental state data; substitute the set mental state parameters and the collected mental state data into the formula. Any three formulas form a system of linear equations with three variables. Screen the calculated factors and take the average value to obtain , , The values of are 1.27, 1.56, and 1.43 respectively;

[0109] The magnitude of the factor is a specific value obtained by quantifying each parameter for the convenience of subsequent comparison. Regarding the magnitude of the factor, it depends on the amount of mental state data and the mental state parameters initially set by those skilled in the art for each set of mental state data; as long as the proportional relationship between the parameter and the quantified value is not affected, such as the mental state parameter is proportional to the value of the behavioral characterization.

[0110] The above has described a detailed implementation of the present invention, but the content described is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made according to the scope of the present invention application should still fall within the scope covered by the patent of the present invention.

Claims

1. A mental diagnosis system based on convolutional neural network and facial recognition, characterized in that: include: The mental state analysis module is used to analyze whether the target patient's mental state is abnormal, including: a mental state acquisition unit and a mental state processing unit; A mental state acquisition unit is used to acquire mental state data including behavior representation values, emotion representation values, and physiological representation values ​​of a target patient; By observing the target patient's behavior, emotions, and physiological conditions, a score of 0-10 is given, and the behavior representation value mark, emotion representation value mark, and physiological representation value mark are marked respectively; The mental state processing unit is used to calculate and obtain mental state parameters according to the mental state data; Process the behavior representation value, emotion representation value, and physiological representation value, and obtain the mental state parameters through calculation; The mental state analysis unit is used to compare and analyze mental state parameters to analyze whether the target patient's mental state is abnormal: If the mental state parameter is ≥ the mental state threshold, the target patient is judged to be mentally abnormal; The facial recognition module includes: a facial feature acquisition unit and a facial feature analysis unit; The facial feature collection unit is used to collect the facial features of the patient and evaluate whether they are effective; Capturing an image of the patient's face, obtaining the degree of the patient's target facial features and the time the target facial features exist, and drawing a target facial feature fluctuation graph with the X-axis representing the time the target facial features exist and the Y-axis representing the degree of the target facial features, and obtaining the area enclosed by the fluctuation line and the X-axis, and marking the area as a fluctuation parameter; Preset fluctuation parameter thresholds, and conduct comparative analysis between fluctuation parameters and fluctuation parameter thresholds; If the fluctuation parameter is ≥ the fluctuation parameter threshold, it means that the patient's target facial feature exists for a long time or to a high degree, that is, the target facial feature is judged to be valid; otherwise, the target facial feature is judged to be invalid; The facial feature analysis unit is used to analyze whether the target facial features of the patient are related to mental abnormality when the target facial features of the patient are valid; Based on multiple patients, the proportion of target facial features among all patients with mental disorders is obtained; In a two-dimensional coordinate system with the mental state parameter as the X-axis and the target facial feature ratio as the Y-axis, a two-dimensional fluctuation diagram of the mental state parameter-target facial feature ratio is constructed; The mental state parameters are divided into multiple intervals according to the numerical values, and the area of ​​the region enclosed between the wave line and the X-axis of each interval is calculated respectively. The ratio of this area to the area threshold is the relevant representation value; Obtain relevant parameters through calculation; Preset relevant thresholds and compare and analyze relevant parameters with relevant thresholds; If the relevant parameter ≥ the relevant threshold, the target facial feature is judged to be related to mental abnormality; The early warning module is used to detect the facial features of the patient and thus predict the patient's mental state; In the case where the target facial features are related to mental abnormalities, the fluctuation parameters of the target facial features of the patient are obtained, and the warning parameters are calculated based on the relevant parameters of the target facial features; Preset warning thresholds and compare and analyze warning parameters with warning thresholds; If the warning parameter is ≥ the warning threshold, it means that the patient has an abnormal mental tendency and a mental abnormality alarm is generated until the medical staff turns it off.

Images