Nursing teaching feedback analysis system based on big data

Abstract

The invention relates to the technical field of teaching feedback analysis, in particular to a nursing teaching feedback analysis system based on big data, which comprises a node registration module, a skill fitting module, a semantic networking module, a variable correction module and a path matching module. According to the method, the continuity and boundary definition of node time sequence arrangement are optimized, the aging expression ability of skill behavior records in the nursing teaching process is enhanced, skill burden indexes are generated in combination with skill node frequency and deviation states, quantitative recognition and sorting screening of teaching difficulties are achieved, and a label connection structure is constructed, so that the teaching efficiency is improved. The structural organization degree and the node density distribution accuracy between skill labels are improved, the fluctuation trend of score variables in a skill sequence is combined, the evaluation participation proportion is adjusted, the weight configuration of multi-source scores in skill performance evaluation is optimized, the coverage proportion between a behavior node sequence and an index action field is used for constructing a path corresponding relation, and the skill performance evaluation accuracy is improved. And the adaptation accuracy between the teaching task and the evaluation index is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of teaching feedback analysis, and particularly relates to a nursing teaching feedback analysis system based on big data. BACKGROUND

[0002] The technical field of teaching feedback analysis includes collecting, sorting, analyzing and managing feedback information in the process of teaching activities, aiming to obtain data interaction information between teachers and students, teaching resources and learning behaviors through systematic means, and to classify, model and behavior mine the data, form a feedback system for education evaluation, teaching optimization and learning situation monitoring, including teaching process data collection mechanism, feedback data structure modeling method, teaching behavior pattern recognition algorithm, teaching decision support model, teaching evaluation tool integration method, collecting teaching activity data by relying on an education information platform, and combining database technology data analysis algorithm and visualization tools to structure the collected results and provide real-time feedback for teachers to provide teaching adjustment basis or for education management departments to provide management evaluation data basis, which is an important support direction for current education management and intelligent teaching system construction, wherein the nursing teaching feedback analysis system based on big data is applied to the nursing professional teaching scene, based on structured nursing teaching process data, and is used for collecting, modeling, analyzing and feedback processing. For the problems of feedback mechanism lag, teaching behavior unquantifiable, and strong subjectivity of teaching effect evaluation in the teaching process of nursing profession, a structure composed of a data collection unit, a data preprocessing unit, a feedback modeling unit and an index analysis unit is constructed, course interaction record unit, questionnaire collection unit and clinical training scoring unit deployed on the teaching terminal are used to obtain student skill operation process index, teacher evaluation content and course interaction record data, through the time stamp-based serialization preprocessing method, multi-source data alignment and abnormal data cleaning are completed, and the task node coding mechanism is used to classify, archive and structureally represent different teaching behaviors, through the nursing skill point frequency statistical method and the evaluation index structure mapping rule, the quantitative presentation and node-level analysis of teaching behavior feedback are completed, which is a system design for teaching feedback analysis based on data sequence integration method, coding classification rule and index mapping relationship.

[0003] The traditional nursing teaching feedback analysis technology only collects the time stamp information of the event itself when processing the node time information, does not analyze the sequence and boundary overlap problems between nodes, is easy to cause inaccurate time sequence expression, mainly depends on the frequency or record label in the skill behavior expression without forming a systematic target deviation judgment mechanism, only classifies and archives in the skill association without extracting the semantic proximity structure between the skill pairs, causes insufficient correlation identification between labels, does not perform fluctuation interference degree evaluation on the multi-source data in the scoring information processing process, cannot effectively distinguish the scoring quality, and only based on task matching in the corresponding relationship between the behavior nodes and the index fields in the path analysis, ignores the corresponding structure between the behavior execution sequence and the evaluation coverage, and is easy to cause teaching path identification deviation and incomplete index adaptation problems. SUMMARY

[0004] The purpose of the present application is to solve the problems existing in the prior art, and the nursing teaching feedback analysis system based on big data is proposed.

[0005] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows: the nursing teaching feedback analysis system based on big data comprises: The node registration module calls the nursing practical training task operation record, analyzes the start and end time and completion sequence of the operation event, constructs the operation node sequence arranged according to the course requirements, judges the time interval change trend between nodes, adjusts the boundary range of the overlapping nodes, rearranges the node sequence, and generates time track reconstruction information; The skill fitting module uses the time track reconstruction information to calculate the execution frequency of the teaching behavior node, analyzes the difference between the current state of the corresponding skill point and the set target, generates the skill burden index in combination with the frequency and deviation degree, sorts and selects the teaching difficulties of the nodes, and generates the skill task adaptation data; The semantic network building module extracts the stage number and skill classification field of the node according to the skill task adaptation data, judges the consistency of the adjacent node label field, identifies the semantic proximity degree between the skill pairs, constructs the node connection structure and assigns the connection weight, and generates the label node density information; The variable correction module analyzes the sequence change of each scoring variable on a plurality of skill nodes based on the label node density information, judges the interference intensity of the scoring fluctuation, adjusts the participation proportion of the scoring variable in the node evaluation, and generates the scoring source credibility mapping information; The path matching module analyzes the skill behavior node sequence based on the scoring source credibility mapping information, extracts the operation number sequence, obtains the action field of the index structure, judges the coverage range of the order group and the field, calculates the matching node proportion, establishes the path relationship between the behavior and the evaluation, and generates the index path matching information.

[0006] As a further aspect of the present invention, the time trajectory reconstruction information includes node sequence index, start and end time boundaries, and overlapping interval distribution; the skill task adaptation data includes node burden index, behavior achievement deviation, and skill modeling weight coefficient; the tag node density information includes semantic tag fields, node pairing relationships, and connection weight distribution; the scoring source reliable mapping information specifically includes scoring variable stability index, variable interference influence coefficient, and participation intensity correction factor; and the indicator path matching information includes indicator coverage structure, path node lookup table, and matching accuracy data.

[0007] As a further aspect of the present invention, the node registration module includes: The operation sequence generation submodule obtains the operation records of nursing training tasks, arranges each operation event according to the course requirements, identifies the start time, end time and completion order of each operation, constructs the operation node sequence according to the event order, and generates the node order arrangement result; The time span determination submodule determines the trend of the interval change of the start and end times of adjacent nodes based on the node sequence arrangement result, extracts the time span of each group of continuous operations, analyzes the change of the time span distribution, and obtains the time span distribution characteristic value. The sequence boundary adjustment submodule calls the time span distribution feature value, analyzes the overlapping node pairs in the node start and end time intervals, determines the range of overlapping intervals, adjusts the boundary start and end times of overlapping nodes, rearranges the timeline order of operation nodes, and establishes time trajectory reconstruction information.

[0008] As a further aspect of the present invention, the specific execution process of adjusting the start and end times of the boundary of overlapping nodes is as follows: detect the time difference between the end time of consecutive operation nodes and the start time of adjacent nodes, calculate the overlap ratio of each node pair in the time series, determine the situation of overlap between nodes based on the time overlap ratio, and in the node pairs with overlap, determine the time division point between the end time of the previous node and the start time of the next node according to the start time order, redefine the start and end boundary range of the nodes, and obtain the time boundary adjustment result. The process for obtaining the overlap determination threshold is as follows: based on the normalized difference data of the node execution time, calculate the rate of change of the normalized difference value between consecutive nodes, extract the minimum stable interval of the rate of change distribution, use the mean of the rate of change corresponding to the target stable interval as the overlap determination threshold, and use the overlap determination threshold for overlap identification and determination when adjusting the time boundary.

[0009] As a further aspect of the present invention, the skill fitting module includes: The node frequency acquisition submodule uses the time trajectory reconstruction information to obtain the operation records of teaching behavior nodes in nursing training tasks, extracts the time marker of the node in each training session according to the node number, counts the number of times each node appears in the continuous training records, calculates the average occurrence frequency of the target node within a unit task, and generates behavior node occurrence frequency information. The behavior offset generation submodule calls the frequency information of the behavior nodes, matches the actual completion time of each node in a single task according to the operation number, calculates the average completion time in multiple tasks, obtains the completion time offset sequence, and calculates the skill burden index. The skill index summarization submodule sorts each behavioral node according to the skill burden index, filters out teaching difficulties, determines the importance weight of each node in skill modeling, maps it to the skill modeling path, establishes a skill task adaptation hierarchy, and obtains skill task adaptation data.

[0010] As a further aspect of the present invention, the semantic network construction module includes: The operation tag extraction submodule obtains the operation tag content corresponding to the continuous teaching behavior node based on the skill task adaptation data, extracts the stage number field and skill classification field in the node, judges the consistency of the stage number in the tag field of adjacent nodes, and generates a stage number consistency cluster value. The semantic analysis calculation submodule extracts the skill classification field content between each node pair based on the stage number consistency cluster value, calculates the ratio of the number of intersections of skill classification fields in the node pair to the total number of fields, calculates the semantic proximity value, and generates the node semantic pairing strength value. The connection structure generation submodule calls the node semantic pairing strength value, constructs the connection structure between nodes and assigns connection strength weights, counts the proportion of skill pairs that establish connections in each matching level interval in all tasks, and obtains tag node density information.

[0011] As a further aspect of the present invention, the variable correction module includes: The scoring sequence analysis submodule obtains the label node density information, analyzes the continuous records of multiple scoring sources in each teaching behavior node, calculates the continuous performance sequence changes of each type of scoring variable in multiple skill nodes, analyzes the stability and distribution of scoring fluctuation changes, and generates the fluctuation characteristic coefficient of scoring variables. The fluctuation interference discrimination submodule compares the degree of change of the current rating variable with the average state of similar variables based on the fluctuation characteristic coefficient of the rating variable, determines the interference intensity of the rating fluctuation on the current skill performance result, and obtains the interference influence factor of the rating variable. The variable proportion adjustment submodule determines the weight adjustment range of each scoring variable in the node evaluation based on the interference influence factor of the scoring variable, adjusts the participation ratio of the variable in the node evaluation, and obtains the reliable mapping information of the scoring source.

[0012] As a further aspect of the present invention, the path matching module includes: The node sequence extraction submodule obtains the reliable mapping information of the scoring source, analyzes the arrangement order of skill behavior nodes in the teaching process, extracts the sequence of operation numbers executed continuously in each group of teaching tasks, and establishes the node sequence extraction results. The coverage determination submodule extracts the action coverage field in the indicator structure based on the node sequence extraction results, determines the coverage range of each group of operation number sequences and the corresponding indicator fields, analyzes the matching relationship between each node in the operation sequence and the indicator fields, and obtains the matching node coverage ratio. The path relationship generation submodule compares the coverage of multiple operation number sequences based on the matching node coverage ratio, determines the path distribution between each behavior sequence group and the evaluation index, establishes the path correspondence between behavior and evaluation, and generates index path matching information.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by optimizing the continuity and boundary clarity of the temporal arrangement of nodes, the timely expressive ability of skill behavior records in nursing teaching is enhanced. Skill burden indicators are generated by combining skill node frequency and deviation status, enabling quantitative identification and sorting of teaching difficulties. By constructing a label connection structure, the degree of structured organization between skill labels and the accuracy of node density distribution are improved. By combining the fluctuation trend of scoring variables in the skill sequence, the evaluation participation ratio is adjusted, and the weight configuration of multi-source scoring in skill performance assessment is optimized. The path correspondence is constructed by using the order of behavioral nodes and the coverage ratio between indicator action fields, thereby improving the accuracy of the fit between teaching tasks and evaluation indicators. Attached Figure Description

[0014] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the node registration module of the present invention; Figure 3 This is a flowchart of the skill fitting module of the present invention; Figure 4 This is a flowchart of the semantic network building module of the present invention; Figure 5 This is a flowchart of the variable correction module of the present invention; Figure 6 This is a flowchart of the path matching module of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0016] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0017] Please see Figure 1 The nursing teaching feedback analysis system based on big data includes: The node registration module calls the nursing training task operation record, analyzes the start and end time and completion order of the operation events, constructs the operation node sequence arranged according to the course requirements, judges the changing trend of the time interval between nodes, adjusts the boundary range of overlapping nodes, rearranges the node order, and generates time trajectory reconstruction information. The skill fitting module uses time trajectory reconstruction information to calculate the execution frequency of teaching behavior nodes, analyze the difference between the current state of the corresponding skill point and the set goal, generate skill burden index by combining frequency and deviation, sort the nodes and filter teaching difficulties, and generate skill task adaptation data. The semantic network construction module extracts the stage number and skill classification field of the node based on the skill task adaptation data, judges the consistency of the label field of adjacent nodes, identifies the semantic proximity between skill pairs, constructs the node connection structure and assigns connection weights, and generates label node density information. The variable correction module analyzes the sequence changes of each rating variable across multiple skill nodes based on the tag node density information, determines the interference intensity of rating fluctuations, adjusts the participation ratio of rating variables in node evaluation, and generates reliable mapping information of rating sources. The path matching module analyzes the sequence of skill behavior nodes based on the reliable mapping information of the scoring source, extracts the operation number sequence, obtains the action field of the indicator structure, determines the coverage of the sequence group and the field, calculates the proportion of matching nodes, establishes the path relationship between behavior and evaluation, and generates indicator path matching information.

[0018] The time trajectory reconstruction information includes node sequence index, start and end time boundaries, and overlapping interval distribution. The skill task adaptation data includes node burden index, behavior achievement deviation, and skill modeling weight coefficient. The tag node density information includes semantic tag fields, node pairing relationships, and connection weight distribution. The scoring source reliable mapping information specifically includes scoring variable stability index, variable interference influence coefficient, and participation intensity correction factor. The indicator path matching information includes indicator coverage structure, path node comparison table, and matching accuracy data.

[0019] Please see Figure 2 The node registration module includes: The operation sequence generation submodule obtains the operation records of nursing training tasks, arranges each operation event according to the course requirements, identifies the start time, end time and completion order of each operation, constructs the operation node sequence according to the event order, and generates the node order arrangement result; The obtained "Intravenous Infusion" nursing training task operation record contains a series of operation events performed by trainee S01 in task T01, such as {Event ID: E01, Operation: 'Check Doctor's Order', Start Time: 10:00:05, End Time: 10:00:25, Sequence: 1}, {Event ID: E02, Operation: 'Wash Hands', Start Time: 10:00:30, End Time: 10:00:55, Sequence: 2}, {Event ID: E03, Operation: 'Select Vessel', Start Time: 10:01:00, End Time: 10:01:40, Sequence: 3}, etc., defined according to the course requirements. The standard operating sequence {'verify medical orders', 'wash hands', 'select vein', 'disinfect skin', 'puncture', 'fix needle'} arranges these events, identifies and extracts the start timestamp, end timestamp, and completion sequence number of each operation event, and then connects them sequentially according to the completion sequence number to construct the initial operation node sequence, for example, N1('verify medical orders', 10:00:05-10:00:25)->N2('wash hands', 10:00:30-10:00:55)->N3('select vein', 10:01:00-10:01:40), generating the node sequence arrangement result.

[0020] The time span determination submodule determines the trend of the interval between the start and end times of adjacent nodes based on the node order arrangement results, extracts the time span of each group of continuous operations, analyzes the changes in the time span distribution, and obtains the time span distribution characteristic value. Based on the node sequence, for example, N1(10:00:05-10:00:25)->N2(10:00:30-10:00:55)->N3(10:01:00-10:01:40), first determine the trend of the interval between the start and end times of adjacent nodes. Calculate that the interval between the end time of node N1 and the start time of N2 is 5 seconds, and the interval between the end time of N2 and the start time of N3 is also 5 seconds. This indicates that the intervals in this sequence are constant and the trend is stable. Next, extract the time span of each group of consecutive operations and define "hand preparation". The operation group is {N2('wash hands'), Nx('put on gloves')}, and its total time span is from the start time of N2 to the end time of Nx. Then, the time span distribution of the "hand preparation" operation group is analyzed in different trainees or multiple training sessions of the same trainee. For example, the span of trainee S01 is 65 seconds, S02 is 70 seconds, and the second time S01 is 63 seconds. These changes are recorded, and finally, the statistical characteristics of these span value sets are calculated. For example, the mean is 66 seconds and the standard deviation is 3.6 seconds. The standard deviation of 3.6 is the obtained time span distribution characteristic value.

[0021] The sequence boundary adjustment submodule calls the time span distribution feature value, analyzes the overlapping node pairs in the node start and end time intervals, determines the range of overlapping intervals, adjusts the boundary start and end times of overlapping nodes, rearranges the timeline order of operation nodes, and establishes time trajectory reconstruction information. By calling the time span distribution feature value, the node sequence of trainee S02 in task T02 was analyzed. It was found that the start and end times of N4 ('Disinfecting skin', 10:02:00-10:02:20) and N5 ('Preparing adhesive tape', 10:02:15-10:02:30) overlapped. The overlapping interval was determined to be from 10:02:15 to 10:02:20, a total of 5 seconds. The specific adjustment process was as follows: First, it was detected that in the continuous operation nodes, the start time of node N5 (10:02:15) was earlier than the end time of the adjacent preceding node N4 (10:02:20), and the time difference was calculated to be -5 seconds. Then, the node pair in the time sequence was calculated. The overlap ratio is calculated as follows: N4 duration is 20 seconds, N5 duration is 15 seconds, overlap duration is 5 seconds, and overlap ratio is 5 / (20+15-5)=16.7%. Then, the overlap ratio is used to determine if there is overlap between nodes. The overlap threshold is obtained through a specific process. This process involves first normalizing the execution time of all operation nodes in 500 nursing training records to obtain normalized difference data. For example, for the "disinfect skin" operation, the average duration is 18 seconds, and one execution lasted 20 seconds, resulting in a normalized value of 20 / 18=1.11. Another consecutive node, "prepare adhesive tape," has an average duration of 16 seconds, and one execution... After 15 seconds, the normalized value is 15 / 16 = 0.9375, and the normalized difference value is |1.11 - 0.9375| = 0.1725. Then, the rate of change of the normalized difference values ​​for 50 consecutive node pairs is calculated. For example, if the previous difference value is 0.15, the rate of change is (0.1725 - 0.15) / 0.15 = 15%. The interval between -5% and 5% for 10 consecutive rate of change values ​​in the rate of change sequence is extracted as the minimum stable interval. The arithmetic mean of the 10 rates of change within this interval is calculated. For example, if the average is 1.8%, this average of 1.8% is used as the overlap threshold. 0.8% is used for overlap identification during time boundary adjustment. Since 16.7% is greater than 1.8%, it is determined that N4 and N5 overlap. In the overlapping node pair N4 and N5, the time division point between the end time of the previous node N4 and the start time of the next node N5 is determined according to the start time order. This division point is set at the middle point of the overlapping interval, i.e., 10:02:17.5. The end time of N4 is redefined as 10:02:17 and the start time of N5 is redefined as 10:02:18, and the time boundary adjustment result is obtained. Finally, the timeline order of the operation nodes is rearranged based on the adjusted node boundary to establish time trajectory reconstruction information.

[0022] Please see Figure 3 The skill fitting module includes: The node frequency acquisition submodule uses time trajectory reconstruction information to obtain the operation records of teaching behavior nodes in nursing training tasks. It extracts the time marker of the node in each training session based on the node number, counts the number of times each node appears in the continuous training records, calculates the average occurrence frequency of the target node within a unit task, and generates behavior node occurrence frequency information. By reconstructing information from time trajectories, the operation records of the teaching behavior node N5 ('puncture') in 100 "intravenous infusion" nursing training tasks are obtained. Based on the unique number N5 of this node, the time marker of the node's appearance is extracted from the operation log of each training task. For example, it appears once in task T01, twice in T02 (retrieved after the first failure), and once in T03. The total number of times node N5 appears in all 100 consecutive training records is 115. Then, the average occurrence frequency of the target node within a unit task is calculated, i.e., 115 times / 100 tasks = 1.15 times / task, generating behavior node occurrence frequency information.

[0023] The behavior offset generation submodule calls the frequency information of behavior nodes, matches the actual completion time of each node in a single task based on the operation number, calculates the average completion time across multiple tasks, and obtains the completion time offset sequence using the formula: ; Calculate the skill burden level index; in, For the first The skill burden index for each behavioral node For the first The normalized occurrence frequency of a behavior node indicates the frequency of that node's occurrence across all training tasks. With maximum frequency The ratio, For the first The cumulative frequency of each behavior node across all tasks is obtained by counting the number of times the node's operation record appears in each round of tasks. The maximum cumulative frequency among all behavior nodes is obtained by iterating through the frequency set of all nodes and extracting the maximum value. For the first The behavior node in the first Completion time of this task Compared with its average completion time The ratio, For the first The behavior node in the first The actual completion time of this task is obtained by comparing the start and end times of the operation at that node in the task operation log. For the first The average completion time of each behavior node across all tasks, through analyzing the node's performance... All in this task The values ​​are obtained by averaging. For the first The mastery target for each skill point corresponding to each behavioral node is determined by the course skills assessment standards. For the first The actual value of the current skill mastery level of each behavioral node is extracted by the weighted value of the node's score in the trusted mapping information of the scoring source. This is the sample count index, representing the number of times a behavior node is collected in the training records. It is obtained by counting the total number of records for the same node in all training tasks. The node index number represents the position number of the nursing teaching behavior node in the task sequence, which is determined by the node number table in the course structure. The training record index number represents the node. In the The order of occurrence of each training task is determined by the time sequence identifier recorded in the training log. The frequency information of the behavior node calling node N5 ('Piercing') is used to match its actual completion time in a single task T01 based on the operation number N5. Its start time is 10:03:00, end time is 10:03:12, and completion time is... Given 12 seconds, calculate its duration. Average completion time for this task For example, if the time is 10 seconds, obtain its completion time offset sequence across all tasks, such as {+2 seconds, -1 second, +3 seconds, ...}, and then use the formula... Calculate the skill burden index, among which, For node index numbering, here , representing node N5 ('puncture'), This is the index number for the training record, indicating that node 5 is in the [index number]. Records from this practical training task. For the first The normalized occurrence frequency of each behavior node is obtained by using the node... Total frequency of occurrence in all tasks Divide by the highest frequency of occurrence among all nodes For example, the total frequency of occurrence of node N5 ('puncture'). The frequency was 115 times, while the maximum cumulative frequency of occurrence among all nodes was N2 ('wash hands'). Next, then , For the first The node at the th Completion time of this task Compared with its average completion time The ratio, for example, in the first task, In the second mission In the third mission The summation part in the formula This calculation represents the average deviation of the completion time from the average time for all tasks. For example, for the first three tasks, this value would be... Assuming the overall average value after 100 tasks is 0.18, For the first The mastery target for each skill point corresponding to each node is set in the course syllabus, using a 10-point scale, with N5 ('piercing') as an example. Set to 9.0. For the first The actual value of the current skill mastery level of each node is extracted from the trusted mapping information of the scoring sources, which yields a weighted average score for that node from all trusted scoring sources. It is 7.5. For behavior nodes Total number of recorded tasks, here The calculation logic of this formula lies in the fact that the numerator quantifies the complexity and uncertainty of the operation by multiplying the frequency of node occurrence by the instability of operation time, while the denominator quantifies the skill gap by the difference between the mastery level target and the actual value. Dividing the whole yields a comprehensive burden index. The skill burden index reflects the comprehensive load intensity of teaching behavior nodes on teaching resource input, learning effectiveness feedback, and student cognitive pressure during practical training. The larger the value of this index, the more difficult or bottleneck the teaching point or skill bottleneck is. The advantage of this formula is that it comprehensively considers the frequency of operations. Time stability and skill mastery gap The three dimensions make the identification of teaching difficulties more comprehensive and accurate, avoiding the one-sidedness of single-dimensional assessment. Substituting the values ​​of each parameter into the formula yields the result. ; The result of 0.429 is a quantitative indicator of skill burden, which will be compared with the indicator values ​​of other nodes for subsequent sorting and filtering.

[0024] The skill index summarization submodule sorts each behavioral node according to the skill burden index, filters out teaching difficulties, determines the importance weight of each node in skill modeling, maps it to the skill modeling path, establishes a skill task adaptation hierarchy, and obtains skill task adaptation data. Based on the calculated skill burden indexes for each behavioral node, such as N1 ('verifying medical orders') being 0.152, N4 ('disinfecting skin') being 0.315, and N5 ('puncture') being 0.429, all nodes are sorted from highest to lowest for this index value. The nodes with the highest index values ​​are selected as teaching difficulties. For example, if there are 50 nodes, the top 5 nodes are selected. N5 ('puncture') is identified as a difficulty due to its high value of 0.429. Subsequently, the importance weight of each node in the skill modeling is determined. The weight value is proportional to the skill burden index value. For example, if the total weight is set to 1, the weight of N5 is... This weight is then mapped to the skill model path that is subsequently constructed, establishing a skill task adaptation hierarchy structure containing {difficult nodes, importance weights}, and finally obtaining skill task adaptation data.

[0025] Please see Figure 4 The semantic network building module includes: The operation tag extraction submodule obtains the operation tag content corresponding to the continuous teaching behavior nodes based on the skill task adaptation data, extracts the stage number field and skill classification field in the node, judges the consistency of the stage number in the tag field of adjacent nodes, and generates a stage number consistency cluster value. Based on the skill task adaptation data, the operation label content corresponding to the consecutive teaching behavior nodes N4 ('Disinfect Skin') and N5 ('Puncture') is obtained. The label of N4 is {Stage Number: P2, Skill Category: ['Aseptic Operation', 'Skin Care']}, and the label of N5 is {Stage Number: P2, Skill Category: ['Aseptic Operation', 'Puncture Technique', 'Patient Communication']}. First, the stage number field of the two nodes is extracted, which is P2. Then, the skill category fields are extracted, which are ['Aseptic Operation', 'Skin Care'] and ['Aseptic Operation', 'Puncture Technique', 'Patient Communication'], respectively. Next, the consistency of the stage number in the label field of adjacent nodes is judged. Since the stage number of the two nodes is P2, it is determined to be consistent, and a stage number consistency cluster value is generated. For this node pair, its cluster value is 1.

[0026] The semantic analysis calculation submodule extracts the skill classification field content between each node pair based on the stage number consistency cluster value, and calculates the ratio of the number of intersections of skill classification fields in the node pair to the total number of fields, using the formula: ; Calculate semantic proximity values ​​and generate node semantic pairing strength values; in, The normalized value for the number of intersections of skill classification fields is obtained by statistically analyzing the ratio of the number of common skill classification fields between each pair of teaching behavior nodes to the maximum possible number of fields. This is the normalized value for node phase number consistency. It is obtained by determining the consistency relationship between nodes and phase numbers, converting the results to 0 and 1 values, and then normalizing the values. The normalized value for the co-occurrence frequency of skill pairs in the same teaching task is obtained by counting the number of times each pair of nodes co-occurs in the task sequence and normalizing it with the maximum co-occurrence frequency. The normalized value for the total number of skill category fields participating in semantic pairing calculation is obtained by extracting the joint set of skill fields from the nodes and then normalizing it. This represents the semantic proximity value between node pairs, reflecting the overall proximity strength of the paired nodes in terms of semantic structure. Based on the consistent cluster value of the stage number for node pair (N4, N5) being 1, the skill classification field content of each node is extracted. N4 is ['Aseptic Technique', 'Skin Care'], and N5 is ['Aseptic Technique', 'Puncture Technique', 'Patient Communication']. The number of intersections of the skill classification fields in the node pair is calculated. The intersection is {'Aseptic Technique'}, and the number is 1. The total number of fields is 2+3=5. Then, the formula is used... The semantic proximity score is calculated, which refers to the semantic coupling strength between two teaching behavior nodes in the operation label field dimension during nursing training. This score integrates the degree of intersection of label fields between nodes, the consistency of stage numbers, and the frequency of co-occurrence between nodes in the actual task flow. It is a numerical matching index that integrates semantic and temporal features. To normalize the number of intersections of skill category fields, first calculate the ratio of the intersection number 1 to the number of skill field unions {'Aseptic Technique', 'Skin Care', 'Puncture Technique', 'Patient Communication'} of the node pair, i.e., 1 / 4 = 0.25. Then normalize this value with the largest ratio calculated among all node pairs (e.g., 0.8) to obtain the result. , This is the normalized value for node stage number consistency. Since the stage numbers are consistent, its value is 1; otherwise, it is 0. After normalization, it remains 1. To normalize the number of times skill pairs co-occur in the same teaching task, we count that N4 and N5 co-occur 95 times in 100 tasks, while the maximum co-occurrence count among all node pairs is 98 times. , The normalized value of the total number of skill category fields participating in the semantic pairing calculation, i.e., the number of skill field unions of this node pair (4), is normalized by the maximum number of unions among all node pairs (e.g., 6). , The formula for semantic proximity between node pairs calculates by weighting the similarity of the tag fields with their co-occurrence frequency in the numerator, while the denominator acts as a moderating term to balance the impact of differences in field similarity and co-occurrence frequency. The advantage of this formula lies in incorporating the number of co-occurrences. The square root operation effectively enhances the connection strength of node pairs that are frequently executed in actual operation, while avoiding weight imbalance caused by excessive co-occurrence frequency. This makes the construction of semantic relations closer to the actual operation process. Substituting the parameter values ​​into the formula yields the result. ; The result of 0.980 indicates a very strong semantic association between N4 and N5. This value will be used as the weight of their connecting edges in the skill network graph to generate the node semantic pairing strength value.

[0027] The connection structure generation submodule calls the node semantic pairing strength value, constructs the connection structure between nodes and assigns connection strength weights, counts the proportion of skill pairs that establish connections in each matching level interval in all tasks, and obtains the tag node density information. The semantic pairing strength value of node pair (N4, N5) is 0.980, along with the pairing strength values ​​of all other node pairs. A connection threshold of 0.6 is set. When the pairing strength value of a node pair is greater than 0.6, a connection edge is constructed between the two nodes, and its weight is the strength value. Therefore, a connection is established between N4 and N5 with a weight of 0.980. After traversing all node pairs to complete the edge construction, the percentage of all skill pairs with established connections in all 100 tasks whose semantic pairing strength values ​​fall within different matching level intervals is counted. For example, there are 15 pairs in the interval [0.6, 0.7), accounting for 10%; 45 pairs in the interval [0.7, 0.8), accounting for 30%; 60 pairs in the interval [0.8, 0.9), accounting for 40%; and 30 pairs in the interval [0.9, 1.0], accounting for 20%. Finally, the tag node density information composed of these interval percentages is obtained.

[0028] Please see Figure 5 The variable correction module includes: The rating sequence analysis submodule obtains the label node density information, analyzes the continuous records of multiple rating sources in each teaching behavior node, calculates the continuous performance sequence changes of each type of rating variable in multiple skill nodes, analyzes the stability and distribution of rating fluctuation changes, and generates the rating variable fluctuation characteristic coefficient. Obtain the label node density information, and for the teaching behavior node N5 ('puncture'), analyze the scoring records from two scoring sources (Teacher A, Equipment B) in 5 consecutive training tasks, as shown in Table 1; Table 1. Example of the scoring sequence for node N5 Practical training task number Teacher A score Device B score T01 8.5 8.2 T02 7.0 7.3 T03 8.0 8.1 T04 9.0 8.8 T05 7.5 7.9 As shown in Table 1, the continuous performance sequence changes of each type of rating variable at node N5 are calculated. For teacher A, the rating sequence is {8.5, 7.0, 8.0, 9.0, 7.5}, and for device B, the rating sequence is {8.2, 7.3, 8.1, 8.8, 7.9}. Then, the stability of the rating fluctuation changes of each sequence is analyzed by calculating its standard deviation. The standard deviation of the teacher A rating sequence is 0.791, and the standard deviation of the device B rating sequence is 0.559. The distribution state is recorded. For example, the mean values ​​of the two sets of ratings are 8.0 and 8.06, respectively. Finally, these two standard deviations are used as the characteristic coefficients of the rating variable fluctuation.

[0029] The fluctuation interference discrimination submodule compares the degree of change of the current rating variable with the average state of similar variables based on the fluctuation characteristic coefficient of the rating variable, judges the interference intensity of the rating fluctuation on the current skill performance result, and obtains the interference influence factor of the rating variable. Based on teacher A's fluctuation characteristic coefficient of 0.791 and device B's of 0.559, the degree of change in the current rating variable is compared with the average performance of similar variables. First, the average fluctuation characteristic coefficient of all teachers' ratings at node N5 is calculated. For example, if 10 teachers participate in the rating, the mean of their fluctuation characteristic coefficients is 0.650. Teacher A's coefficient of 0.791 is compared with this mean of 0.650, and it is found that it is higher than the mean. Next, the interference strength of the rating fluctuation on the current skill performance result is determined, and a strong interference factor is defined. Interference range: When the coefficient exceeds the mean [0%, 10%), it is considered low interference; [10%, 25%), it is considered medium interference; and above 25%, it is considered strong interference. Teacher A's coefficient is 21.7% higher than the mean (0.791-0.650) / 0.650=21.7%, which is considered medium interference. This level of interference is quantified into an interference impact factor. For example, the low interference factor is 1.0, the medium interference factor is 1.217, and the strong interference factor is 1.5. Therefore, the interference impact factor of Teacher A's rating variable is 1.217.

[0030] The variable proportion adjustment submodule determines the weight adjustment range of each scoring variable in the node evaluation based on the interference and influence factors of the scoring variables, adjusts the participation ratio of the variable in the node evaluation, and obtains the reliable mapping information of the scoring source. Based on Teacher A's rating variable interference factor of 1.217, the weight adjustment range in the N5 node evaluation is determined. Initially, all rating sources (e.g., 10 teachers and 1 device) have equal weights, all 1 / 11. The weight adjustment range is calculated by dividing the original weight by its interference factor. Therefore, Teacher A's new participation ratio is calculated as (1 / 11) / 1.217. For Device B, its fluctuation characteristic coefficient of 0.559 is lower than the average coefficient of all device ratings of 0.600, indicating low interference. Its interference factor is 1.0, and its participation ratio remains unchanged. This adjustment is performed on all rating sources, and all adjusted participation ratios are normalized to ensure a sum of 1. After this process, Teacher A's weight decreases from the initial 0.091, while the weights of other rating sources increase accordingly. Finally, the new weight set of all rating sources at the N5 node is obtained, which is the reliable mapping information of the rating sources.

[0031] Please see Figure 6 The path matching module includes: The node sequence extraction submodule obtains reliable mapping information of the scoring source, analyzes the arrangement order of skill behavior nodes in the teaching process, extracts the sequence of operation numbers executed continuously in each group of teaching tasks, and establishes node sequence extraction results. Obtain reliable mapping information of the scoring source and analyze the standard arrangement order of skill behavior nodes in the "intravenous infusion" teaching process, such as S={N1,N2,N3,N4,N5,N6}. Then, extract the sequence of operation numbers actually executed consecutively by student S01 in task T01. The actual operation sequence of this student is A1={N1,N2,N4,N3,N5,N6}, where the order of N3 and N4 is reversed. At the same time, extract the operation sequence A2={N1,N2,N4,N5,N5,N6} of another student S02 in task T02, where N5 is executed twice. These actual operation sequences are used as the node sequence extraction results to build the system.

[0032] The coverage determination submodule extracts the action coverage field in the indicator structure based on the node sequence extraction results, determines the coverage range of each group of operation number sequences and the corresponding indicator fields, analyzes the matching relationship between each node in the operation sequence and the indicator fields, and obtains the matching node coverage ratio. Based on the node sequence extraction result A1={N1,N2,N4,N3,N5,N6}, the indicator for "aseptic operation principle" is obtained from the evaluation indicator structure. The action coverage field defined by this indicator is M1={N2,N4,N5}. The coverage interval of the operation number sequence A1 and the indicator field M1 is judged. It is found that A1 contains all the nodes {N2,N4,N5} required by M1. Then, the matching relationship between each node in the operation sequence A1 and the indicator field M1 is analyzed. N2, N4, and N5 in the sequence A1 are completely matched with the nodes in M1. The number of matched nodes is 3. The total length of the sequence A1 is 6. The coverage ratio of the matched nodes in the operation sequence A1 is calculated to be 3 / 6=50%. The same analysis is performed on the sequence A2={N1,N2,N4,N5,N5,N6}. Its coverage ratio is 4 / 6=66.7%, and the coverage ratio of the matched nodes is obtained.

[0033] The path relationship generation submodule compares the coverage of multiple operation number sequences based on the matching node coverage ratio, determines the path distribution between each behavior sequence group and the evaluation index, establishes the path correspondence between behavior and evaluation, and generates index path matching information. Based on the matching node coverage rate of 50% for trainee S01 and 66.7% for trainee S02, the coverage of these two operation number sequences is compared. It is determined that trainee S02's operation sequence shows higher coverage in the evaluation index of "aseptic operation principle". Then, the path distribution between each behavior sequence group, namely A1 and A2, and the evaluation index M1 is determined. The path of A1 is {N2->N4->N5}, while the path of A2 is {N2->N4->N5->N5}. The system records these specific behavior paths and finally establishes a path correspondence between the behavior sequences A1 and A2 and the evaluation index "aseptic operation principle". This relationship includes the specific behavior sequence, coverage rate, and path distribution details, generating index path matching information.

[0034] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A nursing teaching feedback analysis system based on big data, characterized in that, The system includes: The node registration module calls the nursing training task operation record, analyzes the start and end time and completion order of the operation events, constructs the operation node sequence arranged according to the course requirements, judges the changing trend of the time interval between nodes, adjusts the boundary range of overlapping nodes, rearranges the node order, and generates time trajectory reconstruction information. The skill fitting module uses the time trajectory reconstruction information to calculate the execution frequency of teaching behavior nodes, analyze the difference between the current state of the corresponding skill point and the set target, generate a skill burden index by combining frequency and deviation, sort the nodes and filter teaching difficulties, and generate skill task adaptation data. The semantic network construction module extracts the stage number and skill classification field of the node based on the skill task adaptation data, judges the consistency of the label field of adjacent nodes, identifies the semantic proximity between skill pairs, constructs the node connection structure and assigns connection weights, and generates label node density information. Based on the label node density information, the variable correction module analyzes the sequence changes of each rating variable across multiple skill nodes, determines the interference intensity of rating fluctuations, adjusts the participation ratio of rating variables in node evaluation, and generates reliable mapping information of the rating source.

2. The nursing teaching feedback analysis system based on big data according to claim 1, characterized in that, The time trajectory reconstruction information includes node sequence index, start and end time boundaries, and overlapping interval distribution; the skill task adaptation data includes node burden index, behavior achievement deviation, and skill modeling weight coefficient; the tag node density information includes semantic tag fields, node pairing relationships, and connection weight distribution; and the scoring source reliable mapping information specifically includes scoring variable stability index, variable interference influence coefficient, and participation intensity correction factor.

3. The nursing teaching feedback analysis system based on big data according to claim 1, characterized in that, The node registration module includes: The operation sequence generation submodule obtains the operation records of nursing training tasks, arranges each operation event according to the course requirements, identifies the start time, end time and completion order of each operation, constructs the operation node sequence according to the event order, and generates the node order arrangement result; The time span determination submodule determines the trend of the interval change of the start and end times of adjacent nodes based on the node sequence arrangement result, extracts the time span of each group of continuous operations, analyzes the change of the time span distribution, and obtains the time span distribution characteristic value. The sequence boundary adjustment submodule calls the time span distribution feature value, analyzes the overlapping node pairs in the node start and end time intervals, determines the range of overlapping intervals, adjusts the boundary start and end times of overlapping nodes, rearranges the timeline order of operation nodes, and establishes time trajectory reconstruction information.

4. The nursing teaching feedback analysis system based on big data according to claim 3, characterized in that, The specific execution process of adjusting the start and end times of overlapping nodes is as follows: detect the time difference between the end time of consecutive operation nodes and the start time of adjacent nodes, calculate the overlap ratio of each node pair in the time series, determine the overlap between nodes based on the time overlap ratio, and in the overlapping node pairs, determine the time division point between the end time of the previous node and the start time of the next node according to the start time order, redefine the start and end boundary range of the nodes, and obtain the time boundary adjustment result. The process for obtaining the overlap determination threshold is as follows: based on the normalized difference data of the node execution time, calculate the rate of change of the normalized difference value between consecutive nodes, extract the minimum stable interval of the rate of change distribution, use the mean of the rate of change corresponding to the target stable interval as the overlap determination threshold, and use the overlap determination threshold for overlap identification and determination when adjusting the time boundary.

5. The nursing teaching feedback analysis system based on big data according to claim 3, characterized in that, The skill fitting module includes: The node frequency acquisition submodule uses the time trajectory reconstruction information to obtain the operation records of teaching behavior nodes in nursing training tasks, extracts the time marker of the node in each training session according to the node number, counts the number of times each node appears in the continuous training records, calculates the average occurrence frequency of the target node within a unit task, and generates behavior node occurrence frequency information. The behavior offset generation submodule calls the frequency information of the behavior nodes, matches the actual completion time of each node in a single task according to the operation number, calculates the average completion time across multiple tasks, and obtains the completion time offset sequence using the formula: ； Calculate the skill burden level index; in, For the first The skill burden index for each behavioral node For the first The normalized occurrence frequency of each behavior node For the first The behavior node in the first The ratio of the completion time of a task to its average completion time. For the first Each behavioral node corresponds to a skill point setting, which represents a mastery target. For the first The actual value of the current skill mastery level of each behavioral node. For sample count index, Assign node index numbers, Assign index numbers to the training records; The skill index summarization submodule sorts each behavioral node according to the skill burden index, filters out teaching difficulties, determines the importance weight of each node in skill modeling, maps it to the skill modeling path, establishes a skill task adaptation hierarchy, and obtains skill task adaptation data.

6. The nursing teaching feedback analysis system based on big data according to claim 5, characterized in that, The semantic network construction module includes: The operation tag extraction submodule obtains the operation tag content corresponding to the continuous teaching behavior node based on the skill task adaptation data, extracts the stage number field and skill classification field in the node, judges the consistency of the stage number in the tag field of adjacent nodes, and generates a stage number consistency cluster value. The semantic analysis calculation submodule extracts the skill classification field content between each node pair based on the stage number consistency cluster value, calculates the ratio of the number of intersections of skill classification fields in the node pair to the total number of fields, calculates the semantic proximity value, and generates the node semantic pairing strength value. The connection structure generation submodule calls the node semantic pairing strength value, constructs the connection structure between nodes and assigns connection strength weights, counts the proportion of skill pairs that establish connections in each matching level interval in all tasks, and obtains tag node density information.

7. The nursing teaching feedback analysis system based on big data according to claim 6, characterized in that, The variable correction module includes: The scoring sequence analysis submodule obtains the label node density information, analyzes the continuous records of multiple scoring sources in each teaching behavior node, calculates the continuous performance sequence changes of each type of scoring variable in multiple skill nodes, analyzes the stability and distribution of scoring fluctuation changes, and generates the fluctuation characteristic coefficient of scoring variables. The fluctuation interference discrimination submodule compares the degree of change of the current rating variable with the average state of similar variables based on the fluctuation characteristic coefficient of the rating variable, determines the interference intensity of the rating fluctuation on the current skill performance result, and obtains the interference influence factor of the rating variable. The variable proportion adjustment submodule determines the weight adjustment range of each scoring variable in the node evaluation based on the interference influence factor of the scoring variable, adjusts the participation ratio of the variable in the node evaluation, and obtains the reliable mapping information of the scoring source.

8. The nursing teaching feedback analysis system based on big data according to claim 1, characterized in that, The system also includes: Based on the reliable mapping information of the scoring source, the path matching module analyzes the sequence of skill behavior nodes, extracts the operation number sequence, obtains the action field of the indicator structure, determines the coverage of the sequence group and the field, calculates the proportion of matching nodes, establishes the path relationship between behavior and evaluation, and generates indicator path matching information. The indicator path matching information includes the indicator coverage structure, path node comparison table, and matching accuracy data.

9. The nursing teaching feedback analysis system based on big data according to claim 8, characterized in that, The path matching module includes: The node sequence extraction submodule obtains the reliable mapping information of the scoring source, analyzes the arrangement order of skill behavior nodes in the teaching process, extracts the sequence of operation numbers executed continuously in each group of teaching tasks, and establishes the node sequence extraction results. The coverage determination submodule extracts the node sequence results, obtains the action coverage field in the indicator structure, determines the coverage interval of each group of operation number sequences and the corresponding indicator field, analyzes the matching relationship between each node in the operation sequence and the indicator field, and obtains the matching node coverage ratio. The path relationship generation submodule compares the coverage of multiple operation number sequences based on the matching node coverage ratio, determines the path distribution between each behavior sequence group and the evaluation index, establishes the path correspondence between behavior and evaluation, and generates index path matching information.

Images