Workplace psychological screening method based on artificial intelligence

Abstract

The invention discloses a workplace psychological screening method based on artificial intelligence, and relates to the technical field of psychological assessment, and the method comprises the steps of S1, data collection, S2, data preprocessing, S3, model construction and training, S4, real-time screening, S5, regular psychological assessment, and S6, risk early warning. The method integrates questionnaire, daily behavior and physiological data, compared with a traditional single questionnaire survey or interview method, complex relations in the data can be mined more accurately, so that the accuracy of psychological screening is improved, model parameters are evaluated and optimized by minimizing a mean square error loss function and multiple indexes, the predictive ability of the model is further enhanced, and the accuracy of psychological screening is improved. The screening result is more objective and reliable, and the problems that a traditional method is greatly influenced by subjective factors and low in accuracy are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of psychological assessment, and particularly relates to a workplace psychological screening method based on artificial intelligence. Background Art

[0002] In the modern workplace environment, employees are faced with increasingly complex and diverse work stressors, such as high-intensity work tasks, fierce workplace competition, complex interpersonal relationships, and rapidly changing work requirements. These stress factors have a significant impact on the mental health of employees, which may in turn lead to a series of problems such as decreased work efficiency, job burnout, and increased turnover rate, posing challenges to the stable development and sustainable competitiveness of enterprises.

[0003] Traditional workplace psychological screening methods mainly rely on subjective assessment tools, such as questionnaires and interviews. The results of questionnaires are often affected by employees' subjective factors, such as social desirability bias, emotional state fluctuations, and differences in understanding questionnaire questions, resulting in a significant reduction in the accuracy and reliability of data. Interview assessments require a large amount of time and human resources, and it is difficult to ensure the objectivity and consistency of the assessments. In response to the above problems, the following solutions are proposed. Summary of the Invention

[0004] The purpose of the present invention is to provide a workplace psychological screening method based on artificial intelligence, which can improve the accuracy of psychological screening through multi-dimensional data collection, and solves the problems that existing methods mainly rely on subjective assessment tools, resulting in low accuracy and reliability of data and consuming a large amount of time and human resources.

[0005] To solve the above technical problems, the present invention is realized through the following technical solutions:

[0006] The present invention is a workplace psychological screening method based on artificial intelligence, and the screening method includes:

[0007] Step S1, data collection: Design and distribute a workplace psychological questionnaire, and at the same time collect employees' daily behavior data and physiological data;

[0008] Step S2, data preprocessing: Clean the data to remove outliers, perform normalization processing on questionnaire scores, comprehensive behavior indicators, and physiological health indicators, and then extract feature vectors;

[0009] Step S3, model construction and training: Construct a model, and optimize and train the model;

[0010] Step S4, real-time screening: Preprocess the real-time data of employees and input it into the model, output the screening results, and compare with the set threshold;

[0011] Step S5, Regular Psychological Assessment: Summarize the data of employees monthly, obtain the average result through model evaluation, record it in the mental health file, and use it to track and analyze the changing trend of their mental state in the long term;

[0012] Step S6, Risk Warning: Classify the psychological risks into three levels: low, medium, and high according to the screening and evaluation results, and take corresponding intervention measures for different levels;

[0013] Preferably, the specific steps of the above-mentioned Step S1, Data Collection, include the following steps:

[0014] Step S11, Questionnaire Test: Design a psychological questionnaire covering various aspects such as workplace stress, job satisfaction, interpersonal relationships, and job burnout. The questionnaire contains a total of n questions, and the score of each question is m i , where i = 1, 2,..., n, and the answer score of the employee is recorded as x i ;

[0015] Calculate the questionnaire score through the questionnaire data. The score formula is:

[0016] Step S12, Collection of Daily Behavior Data: Collect the daily behavior data of employees at work: the number of lateness L and the number of early departures E, the average duration T of completing tasks and the specified duration T 0 , the number of work mistakes M and the total number of tasks N, the number of times P of participating in team projects and the number of times C of actively communicating in the team in the team collaboration situation;

[0017] Calculate the comprehensive behavior index through the behavior data:

[0018]

[0019] In the formula, α, β, γ, and δ are weight coefficients;

[0020] Step S13, Physiological Data Monitoring: Use wearable devices or health monitoring instruments to collect the physiological data of employees: the average heart rate HR, the lower limit of the normal heart rate range is HR min , and the upper limit of the heart rate is HR max ; Systolic blood pressure SBP, diastolic blood pressure DBP, the lower limit of the normal systolic blood pressure range is SBP min , and the upper limit is SBP max , the lower limit of the diastolic blood pressure is DBP min , and the upper limit is DBP max ; Sleep quality score HR, with a full score of 10 points;

[0021] Calculate the physiological health index:

[0022]

[0023] Where θ, μ, ν, and ω are weight coefficients.

[0024] Preferably, in step S2, data preprocessing, the formula for normalization is:

[0025]

[0026] Where Q norrn is the normalized questionnaire score, B norrn is the normalized comprehensive behavior index, and P norrn is the normalized physical health index.

[0027] Preferably, in step S2, data preprocessing, the specific extraction of the feature vector is as follows:

[0028] Combine the normalized questionnaire score, comprehensive behavior index, and physical health index into a feature vector X = [Q norm , B norm , P norm .

[0029] Preferably, step S3, model construction and training specifically include the following steps:

[0030] Step S31: Adopt a hybrid algorithm H-DTNN that combines a decision tree and a neural network. The decision tree is used for preliminary feature partitioning and classification of data, and the neural network is used for further non-linear fitting and accurate prediction;

[0031] Step S32: Divide the preprocessed data into a training set and a test set;

[0032] Step S33: Use the training set to train the H-DTNN model. In the decision tree, select the best splitting feature according to the principle of maximum information gain to construct the decision tree structure. In the neural network, adopt a multi-layer perceptron structure, including an input layer, a hidden layer, and an output layer, and use the backpropagation algorithm for parameter update and optimization. The goal is to minimize the mean square error loss function between the prediction result and the actual mental state label:

[0033]

[0034] Where N is the number of training samples, y i is the actual mental state label, is the mental state value predicted by the model;

[0035] Step S34: Evaluate and optimize the model through the test set, and adjust the parameters of the model according to multiple indicators such as accuracy, recall rate, and F1 value.

[0036] Preferably, the specific steps of step S4, real-time screening, are as follows:

[0037] Step S41: Collect the real-time data of employees and calculate the real-time feature vector X real ;

[0038] Step S42: Input X real into the trained H-DTNN model, and the model outputs the screening result R;

[0039] Step S43: Set the threshold R th , and when R < R th , send out a warning signal.

[0040] Preferably, the specific evaluation method of the step S5, regular psychological evaluation is as follows:

[0041] Comprehensively evaluate the psychological state of employees every month, and input the set of feature vectors {X j} after preprocessing the multiple data collected within one month into the model to obtain the average evaluation result:

[0042]

[0043] where j = 1, 2,..., k, and k is the number of data collections in this month.

[0044] The present invention has the following beneficial effects:

[0045] Through multi-dimensional data collection, the present invention integrates questionnaire, daily behavior and physiological data. Compared with the traditional single questionnaire survey or interview method, it can more accurately mine the complex relationships in the data, thereby improving the accuracy of psychological screening. By minimizing the mean square error loss function and optimizing the model parameters with multiple indicators, the prediction ability of the model is further enhanced, making the screening result more objective and reliable, and overcoming the problems of large subjective influence and low accuracy of the traditional method.

[0046] Of course, it is not necessary for any product implementing the present invention to achieve all the above-mentioned advantages simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 It is a schematic flowchart of a workplace psychological screening method based on artificial intelligence according to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0049] 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 belong to the scope of protection of the present invention.

[0050] Please refer to Figure 1 As shown, the present invention is a workplace psychological screening method based on artificial intelligence. The screening method includes:

[0051] Step S1, data collection: Design and distribute a workplace psychological questionnaire, and at the same time collect employees' daily behavior data and physiological data;

[0052] Step S2, data preprocessing: Clean the data to remove outliers, normalize the questionnaire scores, comprehensive behavior indicators, and physiological health indicators, and then extract feature vectors;

[0053] Step S3, model construction and training: Construct a model, and optimize and train the model;

[0054] Step S4, real-time screening: Preprocess the employees' real-time data and input it into the model, output the screening results, and compare with the set threshold;

[0055] Step S5, regular psychological assessment: Summarize the employees' data monthly, obtain the average result through model evaluation, record it in the mental health file, and use it to long-term track and analyze the change trend of their mental state;

[0056] Step S6, risk warning: Divide the psychological risks into three levels: low, medium, and high according to the screening and evaluation results, and take corresponding intervention measures for different levels;

[0057] Step S1, data collection specifically includes the following steps:

[0058] Step S11, questionnaire test: Design a psychological questionnaire covering multiple aspects such as workplace stress, job satisfaction, interpersonal relationships, and job burnout. The questionnaire contains a total of n questions, and the score of each question is m i , where i = 1, 2,..., n, and the answer score of the employee is recorded as x i ;

[0059] Calculate the questionnaire score through the questionnaire data. The score formula is:

[0060] Step S12, collection of daily behavior data: Collect the daily behavior data of employees at work: the number of lateness L and the number of early departures E, the average duration T to complete tasks and the specified duration is T 0, the number of work mistakes M and the total number of tasks N, the number of times P participating in team projects and the number of times C of taking the initiative to communicate in the team collaboration situation;

[0061] Calculate the comprehensive behavior index through behavioral data:

[0062]

[0063] In the formula, α, β, γ, δ are weight coefficients;

[0064] Step S13, Physiological data monitoring: Use wearable devices or health monitoring instruments to collect employees' physiological data: average heart rate HR, the lower limit of the normal range of heart rate is HR min , and the upper limit of heart rate is HR max ; Systolic blood pressure SBP, diastolic blood pressure DBP, the lower limit of the normal range of systolic blood pressure is SBP min , and the upper limit is SBP max , the lower limit of diastolic blood pressure is DBP min , and the upper limit is DBP max ; Sleep quality score HR, with a full score set at 10 points;

[0065] Calculate the physiological health index:

[0066]

[0067] In the formula, θ, μ, ν, ω are weight coefficients.

[0068] In step S2, data preprocessing, the formula for normalization is:

[0069]

[0070] In the formula, Q norrn is the normalized questionnaire score, B norrn is the normalized comprehensive behavior index, and P norrn is the normalized physiological health index.

[0071] In step S2, data preprocessing, the specific extraction of feature vectors is:

[0072] Combine the normalized questionnaire score, comprehensive behavior index, and physiological health index into a feature vector X = [Q norm , B norm , P norm .

[0073] Step S3, Model construction and training specifically include the following steps:

[0074] Step S31: Adopt the hybrid algorithm H-DTNN that combines decision tree and neural network. The decision tree is used for preliminary feature partitioning and classification of data, and the neural network is used for further non-linear fitting and accurate prediction.

[0075] Step S32: Divide the preprocessed data into a training set and a test set.

[0076] Step S33: Use the training set to train the H-DTNN model. In the decision tree, select the best splitting feature according to the principle of maximum information gain to construct the decision tree structure. In the neural network, adopt a multi-layer perceptron structure, including an input layer, a hidden layer, and an output layer, and use the backpropagation algorithm for parameter update and optimization. The goal is to minimize the mean square error loss function between the prediction result and the actual psychological state label:

[0077]

[0078] where N is the number of training samples, y i is the actual psychological state label, is the psychological state value predicted by the model;

[0079] Step S34: Evaluate and optimize the model through the test set, and adjust the model parameters according to multiple indicators such as accuracy, recall rate, and F1 value.

[0080] Step S4. The specific steps for real-time screening are as follows:

[0081] Step S41: Collect the real-time data of employees and calculate the real-time feature vector X real ;

[0082] Step S42: Input X real into the trained H-DTNN model, and the model outputs the screening result R;

[0083] Step S43: Set the threshold R th . When R < R th , send out a warning signal.

[0084] Step S5. The specific evaluation method for regular psychological assessment is as follows:

[0085] Comprehensively evaluate the psychological state of employees every month. Input the set of preprocessed feature vectors {X j} of the multiple data collected within a month into the model to obtain the average evaluation result:

[0086]

[0087] where j = 1, 2,..., k, and k is the number of data collection times in this month.

[0088] A specific application of this embodiment is as follows:

[0089] Step S1, data collection:

[0090] Step S11, design a psychological questionnaire covering various aspects such as workplace stress, job satisfaction, interpersonal relationships, and burnout. The questionnaire contains a total of n questions, and the score of each question is m i (i = 1, 2,..., n), and the score of the employee's answer is recorded as x i ;

[0091] Calculate the questionnaire score:

[0092] Distribute it to employees through an online platform, and require employees to complete the answering within a specified time. The system automatically records the score situation;

[0093] Step S12, collect daily behavior data: Collect the daily behavior data of employees at work, such as attendance records (number of late arrivals L, number of early departures E), work efficiency (average duration T to complete tasks, the specified duration is T 0 ), work error rate (number of errors M, the total number of tasks is N), team collaboration situation (number of times P participating in team projects, number of times C of taking the initiative to communicate in the team), etc.;

[0094] Calculate the comprehensive behavior index:

[0095] In the formula, α, β, γ, δ are weight coefficients;

[0096] Step S13, physiological data monitoring: Use wearable devices or health monitoring instruments to collect the physiological data of employees, such as heart rate (average heart rate HR, the lower limit of the normal range of heart rate is HR min , the upper limit is HR max ), blood pressure (systolic blood pressure SBP, diastolic blood pressure DBP, the lower limit of the normal range of systolic blood pressure is SBP min , the upper limit is SBP max , the lower limit of diastolic blood pressure is DBP min , the upper limit is DBP max ), sleep quality (sleep score HR, full score 10 points), etc.;

[0097] Calculate the physiological health index:

[0098]

[0099] In the formula, θ, μ, ν, ω are weight coefficients;

[0100] Step S2, Data Preprocessing: Clean the collected data to remove outliers (such as questionnaire scores significantly deviating from the normal range, obvious errors in physiological data, etc.);

[0101] Normalize the data, mapping the questionnaire score Q, comprehensive behavior index B, and physiological health index P to the interval [0, 1] respectively. The formula is:

[0102]

[0103] Extract the feature vector: Combine the normalized questionnaire score, comprehensive behavior index, and physiological health index into a feature vector X = [Q norm , B norm , P norm ;

[0104] Step S3, Model Construction and Training:

[0105] Step S31: Adopt the hybrid algorithm H-DTNN that combines decision tree and neural network. The decision tree is used for preliminary feature partitioning and classification of data, and the neural network is used for further non-linear fitting and accurate prediction;

[0106] Step S32: Divide the preprocessed data into a training set (accounting for 80%) and a test set (accounting for 20%);

[0107] Step S33: Use the training set to train the H-DTNN model. In the decision tree part, select the best splitting feature according to the principle of maximum information gain to construct the decision tree structure; in the neural network part, adopt a multi-layer perceptron structure, including an input layer, hidden layers (assuming 2 hidden layers with the number of nodes being 5 and 3 respectively) and an output layer, and use the backpropagation algorithm for parameter update and optimization. The goal is to minimize the mean square error loss function between the prediction result and the actual mental state label:

[0108]

[0109] In the formula, N is the number of training samples, y i is the actual mental state label, is the mental state value predicted by the model;

[0110] Step S34: Evaluate and optimize the model through the test set, and adjust the parameters of the model (such as the depth of the decision tree, the learning rate of the neural network, the number of hidden layer nodes, etc.) according to indicators such as accuracy, recall rate, and F1 value to improve the generalization ability of the model;

[0111] Step S4, Real-time Screening:

[0112] Step S41: Collect the real-time data of employees, and obtain the real-time feature vector X according to the above data collection and preprocessing method real ;

[0113] Step S42: Input X real into the trained H-DTNN model. The model outputs the screening result R, and the value range of R is [0, 1]. The closer R is to 1, the healthier the mental state is, and the closer R is to 0, the more likely there are mental problems;

[0114] Step S43: Set a threshold R th (for example, R th = 0.6, determined according to the actual situation). When R < R th , send out a warning signal;

[0115] Step S5, Regular Psychological Assessment: Conduct a comprehensive assessment of the mental state of employees every month. Input the multiple data collected within a month (the set of preprocessed feature vectors {X j}, j = 1, 2, …, k, where k is the number of data collections in this month) into the model to obtain the average assessment result The assessment result is used as part of the employee's mental health file to provide long-term tracking and analysis of the employee's mental state for the enterprise;

[0116] Step S6, Risk Warning: According to the screening and assessment results, give warnings to employees with mental risks. The warning levels are divided into mild, moderate, and severe. Different levels adopt different intervention measures. For example, mild risks can be intervened by recommending online psychological counseling resources, organizing small team activities, etc.; moderate risks arrange professional psychological counselors for regular communication and counseling; severe risks consider temporarily adjusting the work position or giving leave and other measures.

[0117] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0118] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A workplace psychology screening method based on artificial intelligence, characterized in that: The screening method comprises the following steps: Step S1, data collection: design and distribute workplace psychology questionnaires, and collect employees’ daily behavior data and physiological data; Step S2, data preprocessing: cleaning the data to remove outliers, normalizing the questionnaire scores, comprehensive behavioral indicators and physiological health indicators, and then extracting feature vectors; Step S3: Model construction and training: construct a model, and optimize and train the model; Step S4, real-time screening: pre-process the real-time data of employees and input it into the model, output the screening results, and compare them with the set threshold; Step S5, regular psychological assessment: Summarize the employee data every month, obtain the average result through model assessment, and record it in the mental health file for long-term tracking and analysis of the trend of employee mental status changes; Step S6, risk warning: Divide the psychological risks into three levels: light, moderate and severe according to the screening assessment results, and take different intervention measures for different levels.

2. The method for workplace psychology screening based on artificial intelligence according to claim 1, characterized in that: The step S1, data collection specifically includes the following steps: Step S11, questionnaire test: Design a psychological questionnaire covering workplace stress, job satisfaction, interpersonal relationships, occupational burnout, etc. The questionnaire contains n questions in total, and each question is worth m points. i , where i = 1, 2, ..., n, and the employee's answer score is recorded as x i ; The questionnaire score is calculated based on the questionnaire data. The scoring formula is: Step S12, daily behavior data collection: collect employees' daily behavior data at work: number of late arrivals L and early departures E, average time to complete a task T and the specified time T0, number of work errors M and total number of tasks N, number of team projects participated in in team collaboration P and number of active communication in the team C; Through behavioral data, calculate comprehensive behavioral indicators: In the formula, α, β, γ, and δ are weight coefficients; Step S13, physiological data monitoring: using wearable devices or health monitoring instruments to collect the physiological data of employees: average heart rate HR, the lower limit of the normal range of heart rate is HR min , the upper limit of heart rate is HR max ; Blood pressure systolic pressure SBP, diastolic pressure DBP, the lower limit of normal systolic pressure is SBP min , upper limit is SBP max , the lower limit of diastolic blood pressure is DBP min , upper limit is DBP max ; Sleep quality score HR, full score is 10 points; Calculate physiological health indicators: Where θ, μ, ν, and ω are weight coefficients.

3. The method for workplace psychology screening based on artificial intelligence according to claim 1, characterized in that: In the step S2, data preprocessing, the formula for normalization processing is: In the formula, Q norrn is the normalized questionnaire score, B norrn is the normalized comprehensive behavioral index, P norrn It is the normalized physiological health index.

4. The method for workplace psychology screening based on artificial intelligence according to claim 3, characterized in that: In the step S2, data preprocessing, the feature vector is extracted as follows: The normalized questionnaire scores, comprehensive behavioral indicators and physiological health indicators are combined into a feature vector X = [Q norm ,B norm ,P norm ].

5. The method for workplace psychology screening based on artificial intelligence according to claim 1, characterized in that: The step S3, model building and training, specifically includes the following steps: Step S31: using a hybrid algorithm H-DTNN that integrates decision tree and neural network; Step S32: Divide the preprocessed data into a training set and a test set; Step S33: Use the training set to train the H-DTNN model. In the decision tree, select the best split feature according to the principle of maximum information gain, build a decision tree structure, and in the neural network, use a multi-layer perceptron structure, including an input layer, a hidden layer, and an output layer. Use the back propagation algorithm to update and optimize the parameters. The goal is to minimize the mean square error loss function between the predicted result and the actual psychological state label: In the formula, N is the number of training samples, y i is the actual mental state label, is the psychological state value predicted by the model; Step S34: Evaluate and optimize the model through the test set, and adjust the parameters of the model according to multiple indicators such as accuracy, recall rate, and F1 value.

6. The method for workplace psychology screening based on artificial intelligence according to claim 1, characterized in that: The specific steps of step S4, real-time screening are: Step S41: Collect the real-time data of employees and calculate the real-time feature vector X real ; Step S42: X real Input into the trained H-DTNN model, and the model outputs the screening result R; Step S43: Setting the threshold R th , when R<R th When an emergency occurs, an early warning signal is issued.

7. The method for workplace psychology screening based on artificial intelligence according to claim 1, characterized in that: The specific evaluation method of step S5, regular psychological evaluation is: Conduct a comprehensive assessment of the mental state of employees every month, and collect multiple data within a month, and pre-process the feature vector set {X j }, input into the model, and get the average evaluation result: In the formula, j = 1, 2, ..., k, k is the number of data collection times in that month.

Images