Journey map-based tips procedure patient visit experience optimization system

Abstract

The present application relates to the field of medical information technology, in particular to a TIPS operation patient medical experience optimization system based on a journey map, which comprises an intraoperative parameter acquisition module, a parallel node identification module, a cross-department collaborative mapping module, a journey path reconstruction module and an experience optimization decision module. In the present application, the portal vein pressure is continuously collected and the change rate is calculated, and an event record is quickly generated when the pressure suddenly changes, realizing real-time identification and tracking of intraoperative risk points, avoiding potential complications being ignored, splitting the operation node into different parallel links, efficiently connecting the angiography and stent treatment in time, shortening the waiting and switching gap in the process, establishing a collaborative relationship through the comparison of signal time difference in the cross-department link, making the operation cooperation more closely, optimizing the decision based on path reconstruction, making the patient medical process more smooth, reducing repeated operation and invalid waiting, thereby improving the operation efficiency and overall diagnosis and treatment experience.

Description

Technical Field

[0001] This invention relates to the field of medical information technology, and in particular to a TIPS patient medical experience optimization system based on journey maps. Background Technology

[0002] The field of medical information technology refers to the digital support and management of the entire medical service process through information technology. This field covers core aspects such as electronic medical record systems, clinical information systems, image archiving and communication systems, medical data interaction standards, remote consultation platforms, and patient follow-up and management systems. Its overall goal is to achieve unified collection, standardized storage, real-time transmission, and collaborative application of medical data, thereby supporting the hospital's clinical diagnosis and treatment, scientific research analysis, and patient management.

[0003] Among them, the traditional TIPS patient medical experience optimization system refers to a system that provides information processing for patients undergoing transjugular intrahepatic portosystemic shunt surgery, covering aspects such as medical record recording, follow-up reminders, and examination appointments throughout their medical process. This type of system uses electronic medical record archiving, examination time push based on past medical history, and time slot allocation for subsequent imaging examinations and laboratory tests through the hospital's appointment registration platform to provide auxiliary support for the patient's medical experience.

[0004] Current technologies for TIPS patient management rely solely on pre-existing medical histories and pre-set timeframes for scheduling examinations and follow-ups, lacking real-time intraoperative monitoring methods. This results in the failure to promptly capture sudden changes in key physiological indicators. In complex procedures, a single workflow design cannot support the simultaneous execution of different tasks, easily leading to sequential waiting and reduced overall efficiency. Cross-departmental collaboration relies solely on appointment booking platforms for time allocation, lacking dynamic matching of operational signals and action sequences, causing resource waste and uneven treatment pace, ultimately negatively impacting the patient's medical experience. Summary of the Invention

[0005] To address the shortcomings of existing technologies in TIPS patient management, which rely solely on past medical history and pre-set timeframes for scheduling examinations and follow-ups, lacking real-time intraoperative monitoring and failing to promptly capture sudden changes in key physiological indicators; the inability of a single workflow design to support the simultaneous execution of different tasks in complex procedures, leading to sequential waiting and reduced overall efficiency; and the reliance on appointment booking platforms for time allocation in interdepartmental collaboration, lacking dynamic matching of operational signals and actions, resulting in resource waste and uneven treatment rhythms, thus negatively impacting the patient experience, this invention provides a TIPS patient experience optimization system based on a journey map. The technical solution is as follows:

[0006] On the one hand, a TIPS patient medical experience optimization system based on journey maps is provided, which includes: The intraoperative parameter acquisition module acquires portal vein pressure values ​​through hemodynamic monitoring equipment, calculates the rate of change of portal vein pressure values ​​over a continuous time period, and marks a pressure mutation event when the rate of change exceeds a preset threshold, generates an intraoperative key event record table, and transmits it to the parallel node identification module. The parallel node identification module uses a Bayesian classifier to identify parallel operation requirements based on the intraoperative key event record table. It splits a single surgical node into two parallel branches: stent operation and image verification, generates a journey node split map, and transmits it to the cross-departmental collaborative mapping module. The cross-departmental collaborative mapping module calls the journey node split map to establish the collaborative relationship between the radiology department and the interventional department. By comparing the time difference between the contrast agent injection completion signal and the stent release initiation signal, a cross-departmental collaborative journey map is generated and transmitted to the journey path reconstruction module. The journey path reconstruction module redesigns the patient's medical process based on the cross-departmental collaborative journey map. Combining patient waiting and operation time, it uses a Markov decision process to optimize the transfer path between nodes, generates an optimized journey map, and passes it to the experience optimization decision module.

[0007] As a further aspect of the present invention, the intraoperative key event record table includes event type, event time, and event duration; the journey node split graph includes operation branches, verification branches, and branch relationships; the cross-departmental collaborative journey graph includes collaborating departments, signal differences, and collaboration strength; and the optimized journey map includes a transition probability matrix, path sequence, and value function.

[0008] As a further aspect of the present invention, the intraoperative parameter acquisition module includes: The data stream receiving submodule acquires portal vein pressure value sequence data through hemodynamic monitoring equipment, arranges the sequence by timestamp, compares the pressure fluctuation amplitude of adjacent points with the sampling interval, removes abnormal points and integrates them into a continuous time period to obtain portal vein pressure sequence values. The rate of change calculation submodule calls the portal vein pressure sequence value, calculates the ratio of the pressure difference between adjacent time points to the time interval, obtains the rate value of the time interval and arranges it in chronological order, performs scale correction on the rate fluctuation range, and generates a portal vein pressure rate of change sequence. The event marker generation submodule calls the portal vein pressure change rate sequence, compares the rate value with the preset pressure mutation threshold point by point, sets the event marker when the rate value exceeds the threshold, and combines the marker with the timestamp to obtain the intraoperative key event record table.

[0009] As a further aspect of the present invention, the portal vein pressure sequence value refers to the set of portal vein pressure data points collected by a hemodynamic monitoring device and arranged in time stamp order; The portal vein pressure change rate sequence refers to the set of data obtained by dividing the pressure difference between adjacent data points in the portal vein pressure sequence by the corresponding time interval and arranged in chronological order. The preset pressure mutation threshold refers to a pre-defined numerical limit that is compared with data points in the portal vein pressure change rate sequence.

[0010] As a further aspect of the present invention, the parallel node identification module includes: The event data acquisition submodule extracts the timestamp and operation identifier of each operation node based on the intraoperative key event record table, rearranges the timestamps in chronological order and fills in the missing values, establishes a correspondence between the rearranged results and the operation identifier, removes duplicate entries, and generates a node time sequence. The node feature classification submodule calls the node time sequence to extract the time difference between adjacent nodes, inputs the time difference and operation identifier combination frequency into the Bayesian classifier, calculates the conditional probability and compares it with the parallel requirement threshold, divides the combinations above the threshold into parallel candidate branches, and generates a node classification vector group. The split structure generation submodule filters support operation and image verification nodes based on the node classification vector group, reassigns the temporal positions of the filtered nodes according to the time difference sequence, analyzes the two independent paths and fits the node connection index table to obtain the journey node split map.

[0011] As a further aspect of the present invention, the node time sequence refers to the sequence of operation nodes extracted from the intraoperative key event record table, arranged in time stamp order, with missing values ​​filled in and duplicate entries removed. The parallel requirement threshold refers to the probability threshold for distinguishing parallel candidate branches, calculated by combining the node time difference and the frequency of operation identifier combinations. The node classification vector group is a set of vectors formed by combining the time difference extracted from the node time sequence with the operation identifier and performing classification calculations.

[0012] As a further aspect of the present invention, the cross-departmental collaborative mapping module includes: The synchronous signal acquisition submodule calls the journey node split map to obtain the contrast agent injection signal sequence and stent release start signal input from the radiology and interventional departments. It compares the two types of signal sequences and detects the start and end markers, removes data that does not conform to the time reference scale, and obtains the synchronous signal sequence set. The time difference calculation submodule, based on the set of synchronization signal sequences, calls the timestamps of the contrast agent injection start point and the stent release start point, compares the two types of timestamps item by item and extracts the offset difference, compares the offset difference with the preset time reference threshold item by item and divides the interval to obtain the cross-sequence time difference set. The collaborative journey generation submodule, for the cross-sequence time difference set, calls the signal segment operation identifiers of the radiology and interventional departments, sorts the difference results and maps them to the journey node split graph, connects the correspondence between operations, and generates a cross-departmental collaborative journey graph.

[0013] As a further aspect of the present invention, the journey path reconstruction module includes: The node status acquisition submodule acquires the patient arrival time and stay duration sequence of multiple nodes in the cross-departmental collaborative journey map, establishes an index relationship between the arrival time sequence and the stay duration, counts the number of patients at each time point and pairs them with the stay duration, and generates a node status matrix. The Markov transition probability submodule calls the patient number and stay duration data in the node state matrix, constructs the state transition relationship based on the patient flow between adjacent nodes, calculates the ratio of patient flow frequency to the total number of patients in the node per unit time, and generates the Markov transition probability matrix. The path generation submodule performs a weighted summation of the cumulative transition probabilities between nodes based on the state transition values ​​in the Markov transition probability matrix, selects the first path according to the cumulative transition probability, and generates an optimized journey map.

[0014] As a further aspect of the present invention, the experience optimization decision module evaluates the patient's experience satisfaction at the journey nodes through the optimized journey map, calculates the ratio of waiting time to operation time at the node, marks the node as an experience bottleneck node when the ratio exceeds a set threshold, and outputs a patient experience optimization plan. The patient experience optimization plan includes bottleneck location, ratio threshold, and adjustment suggestions.

[0015] As a further aspect of the present invention, the experience optimization decision module includes: The journey data acquisition submodule calls the patient's waiting time and operation time data in the optimized journey map, calculates the length of the waiting interval and operation interval respectively, sorts the two types of time data according to the node number, and converts them into a matrix structure to generate node time series data. The node ratio calculation submodule calls the node time series data, calculates the ratio for each node based on the waiting time and operation time, compares the ratio with the node time threshold benchmark, records the node ratio results, and generates a node time ratio sequence. The bottleneck node marking submodule detects whether the ratio is greater than the node duration threshold based on the node duration ratio sequence, stores and marks the node numbers that exceed the threshold with the ratio as experience bottleneck nodes, and performs structured processing on the nodes according to the optimization strategy rules to generate a patient experience optimization plan.

[0016] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: By continuously collecting and calculating the rate of change of portal vein pressure, event records can be quickly generated when pressure changes abruptly, enabling real-time identification and tracking of intraoperative risks and preventing potential complications from being overlooked. Surgical procedures are split into different parallel stages, allowing for efficient time-series integration of angiography and stent placement, reducing waiting and switching gaps in the process. Cross-departmental procedures establish collaborative relationships by comparing the time differences between signals, resulting in closer coordination between operations. Decision optimization based on path reconstruction makes the patient's medical process smoother in terms of node transfers, reducing repetitive operations and unnecessary waiting, thereby improving surgical efficiency and the overall patient experience. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a system schematic diagram of the present invention; Figure 2 This is a schematic diagram of the system framework of the present invention; Figure 3 This is a flowchart of the intraoperative parameter acquisition module in this invention; Figure 4 This is a flowchart of the parallel node identification module in this invention; Figure 5 This is a flowchart of the cross-departmental collaborative mapping module in this invention; Figure 6 This is a flowchart of the journey path reconstruction module in this invention; Figure 7 This is a flowchart of the experience optimization decision module in this invention. Detailed Implementation

[0019] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0020] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0021] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0022] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0023] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0024] This invention provides a TIPS patient medical experience optimization system based on journey maps, such as... Figure 1-2 The diagram shown illustrates a journey-map-based system for optimizing the patient experience during TIPS procedures. This system includes: The intraoperative parameter acquisition module acquires portal vein pressure values ​​through hemodynamic monitoring equipment, calculates the rate of change of portal vein pressure values ​​over a continuous time period, and marks a pressure mutation event when the rate of change exceeds a preset threshold, generates an intraoperative key event record table, and transmits it to the parallel node identification module. The parallel node identification module uses a Bayesian classifier to identify parallel operation requirements based on the intraoperative key event record table. It splits a single surgical node into two parallel branches: stent operation and image verification, generates a journey node split map, and passes it to the cross-departmental collaborative mapping module. The cross-departmental collaborative mapping module calls the journey node split map to establish the collaborative relationship between the radiology department and the interventional department. By comparing the time difference between the contrast agent injection completion signal and the stent release initiation signal, a cross-departmental collaborative journey map is generated and transmitted to the journey path reconstruction module. The journey path reconstruction module redesigns the patient medical process based on the cross-departmental collaborative journey map. Combining patient waiting time and operation time, it uses Markov decision process to optimize the transfer path between nodes, generates an optimized journey map, and passes it to the experience optimization decision module. The experience optimization decision module evaluates patient satisfaction at journey nodes by optimizing the journey map, calculates the ratio of waiting time to operation time at each node, and marks the node as an experience bottleneck when the ratio exceeds a set threshold, and outputs a patient experience optimization plan. The intraoperative critical event record sheet includes event type, event time, and event duration; the journey node split diagram includes operation branches, validation branches, and branch relationships; the cross-departmental collaborative journey diagram includes collaborating departments, signal differences, and collaboration intensity; the optimized journey map includes the transition probability matrix, path sequence, and value function; and the patient experience optimization plan includes bottleneck locations, ratio thresholds, and adjustment suggestions.

[0025] Specifically, such as Figure 2 , 3 As shown, the intraoperative parameter acquisition module includes: The data stream receiving submodule acquires portal vein pressure value sequence data through hemodynamic monitoring equipment, arranges the sequence by timestamp, compares the pressure fluctuation amplitude of adjacent points with the sampling interval, removes abnormal points and integrates them into a continuous time period to obtain portal vein pressure sequence values. Portal vein pressure sequence data acquired through hemodynamic monitoring equipment, specifically during TIPS surgery, via a pressure sensor placed within the patient's portal vein. The pressure signal is continuously acquired at a sampling frequency of Hertz. The acquired raw dataset contains timestamps and corresponding pressure readings. Initially, the data points may be out of order or contain outliers due to patient movement or device signal interference. For example, a set of data may be acquired containing... The initial pressure data for each monitoring point is shown in Table 1. First, this sequence of data, containing timestamps and pressure values, is sorted in ascending order based on the timestamp field. The sorted sequence is (10:00:01, ...). mmHg), (10:00:02, mmHg), (10:00:03, mmHg), (10:00:04, mmHg), (10:00:06, mmHg), (10:00:07, mmHg), (10:00:08, mmHg), (10:00:09, mmHg), (10:00:10, mmHg), (10:00:15, Next, outlier removal is performed on the sorted sequence. This process involves a dual assessment of the pressure fluctuation amplitude between adjacent data points and the sampling interval. A pressure mutation threshold is set at [value missing]. Pressure fluctuation within a second interval does not exceed mmHg, with the maximum sampling time interval threshold set to Seconds, from the second data point (10:00:02, Starting from (mmHg), calculate its distance from the previous point (10:00:01). The absolute value of the pressure difference (mmHg) is mmHg, time interval is All seconds were within the threshold range, and were therefore considered normal data. Processing continued to the third point (10:00:03). mmHg), which is different from the previous point (10:00:02, The absolute value of the pressure difference (mmHg) is mmHg, exceeding The pressure mutation threshold of mmHg was used, so this point was identified as an outlier and removed. Then, the point (10:00:02, (mmHg) is the new baseline, and the next point that was not eliminated (10:00:04) is used as the reference. The pressure difference was compared with that of mmHg, and the pressure difference was... mmHg, time interval is All seconds were within the threshold, so processing continued until point (10:00:04) was found. mmHg) and point (10:00:06, The time interval between mmHg) is seconds, not exceeding The maximum sampling interval threshold per second, while the point (10:00:10, mmHg) and point (10:00:15, The time interval between mmHg) is If the time exceeds this threshold, it indicates that data acquisition has been interrupted. Therefore, the data from (10:00:01) to (10:00:10) will be integrated into a continuous time period, and the data point at (10:00:15) will be used as the starting point of the next continuous time period. Similarly, when processing up to the point (10:00:08), When (mmHg), it is different from the previous point (10:00:07, The absolute value of the pressure difference (mmHg) is mmHg, which also exceeded The threshold of mmHg was used, so this point was also removed. After the point-by-point processing of the above steps, the data from multiple consecutive time periods were finally integrated to obtain the portal vein pressure sequence value.

[0026] Table 1: Initial pressure gauges at monitoring points 10:00:02 25.3 10:00:01 25.1 10:00:04 25.4 10:00:08 12.2 10:00:03 38.6 10:00:07 25.5 10:00:06 25.6 10:00:10 25.9 10:00:15 26.1 10:00:09 25.8 As shown in Table 1, it lists a set of raw portal vein pressure monitoring records containing 10 data points obtained from the monitoring device, showing the disordered state and potential outliers in the early stage of data acquisition.

[0027] The rate of change calculation submodule calls the portal vein pressure sequence value, calculates the ratio of the pressure difference between adjacent time points to the time interval, obtains the rate value of the time interval and arranges it in chronological order, performs scale correction on the rate fluctuation range, and generates a portal vein pressure rate of change sequence. The portal vein pressure sequence value is retrieved. This sequence is a data set that has been processed by removing outliers and integrating continuous data. For example, a continuous time period data sequence is obtained: [(10:00:01, (10:00:02) (10:00:04) (10:00:06) (10:00:07) (10:00:09) ), (10:00:10, The units are seconds and mmHg, respectively. Then, based on this sequence, the rate of pressure change between adjacent time points is calculated. Specifically, the first time point (10:00:01) is taken. ) and the second time point (10:00:02, The calculated pressure difference is... mmHg, time interval is The ratio of the two times is the rate value for that time interval. mmHg / s, and associate this rate value with the timestamp of the end time point 10:00:02. Then, take the second time point (10:00:02). ) and the third time point (10:00:04, The calculated pressure difference is... mmHg, time interval is seconds, the rate value is obtained mmHg / s, and associated with the timestamp 10:00:04, and so on, to complete the calculation of adjacent data points, thereby obtaining a set of raw rate values ​​arranged in chronological order, for example: [(10:00:02, (10:00:04) (10:00:06) (10:00:07) (10:00:09) ), (10:00:10, The units are mmHg / s. Next, scale correction is performed on this set of original rate values. The correction process does not directly apply a specific algorithm, but rather smooths the data through a sliding window, with the window size set to [value missing]. That is, for each rate value, the arithmetic mean of the value itself, its preceding value, and its following value is calculated as the corrected rate value. For the first and last values ​​in the sequence, only the average of the value itself and its nearest neighbor is taken as the rate value. For example, mmHg / s (timestamp 10:00:07), its preceding value is mmHg / s, the second value The corrected value is mmHg / s. mmHg / s, and for the first rate value mmHg / s, its correction value is mmHg / s, by performing this operation on the entire rate sequence, a portal vein pressure change rate sequence is generated.

[0028] The event marker generation submodule calls the portal vein pressure change rate sequence, compares the rate value with the preset pressure mutation threshold point by point, sets the event marker when the rate value exceeds the threshold, and combines the marker with the timestamp to obtain the intraoperative key event record table. The portal vein pressure rate of change sequence is retrieved. This sequence is a smoothed rate data stream that has been calculated and scaled corrected. For example, the following sequence can be obtained: [(10:00:02, (10:00:04) (10:00:06) (10:00:07) (10:00:09) ), (10:00:10, The units are all mmHg / s, and a section of data showing drastic fluctuations during subsequent monitoring is added: [(10:15:21, (10:15:22) (10:15:23) (10:18:04) (10:18:05) (10:18:06) Next, a preset threshold needs to be set to determine whether a sudden change in pressure has occurred. This pressure change threshold is not arbitrarily set, but rather refers to the hemodynamic characteristics of TIPS surgical risk events in clinical practice. When the portal vein pressure changes rapidly within a short period of time (e.g., The absolute value of the change within seconds exceeds A pressure spike in mmHg indicates critical events such as acute stent thrombosis, severe stenosis, or vascular rupture. Therefore, the pressure spike threshold is set... Set as Based on this, the absolute value of each rate value in the rate of change sequence is compared with... Perform point-by-point comparison; the absolute value of the rate value is lower than The range of mmHg / s was determined to be the steady period, and the absolute value of the rate was within... mmHg / s to The interval between mmHg / s was defined as the observation period, while the absolute value exceeding [a certain value] was [defined as the observation period]. The interval of mmHg / s is determined to be the abrupt change period, based on the rate value. Taking mmHg / s as an example, its absolute value is It is greater than the set threshold. The system detects a pressure surge of mmHg / s, therefore determining that a positive pressure change event has occurred. The system immediately sets an event flag, defined as "rapid pressure increase," and combines this flag with the corresponding timestamp 10:15:22 to form a record (10:15:22, "rapid pressure increase"). Similarly, when processing a rate value... When the absolute value is mmHg / s, it is Also greater than the threshold If the pressure drops to mmHg / s, the system determines that a negative pressure abrupt change has occurred, marking it as "sharp pressure drop," and combining it with the timestamp 10:18:05 to form a record (10:18:05, "sharp pressure drop"). For absolute values ​​not exceeding [a certain value], the system will then determine that a negative pressure abrupt change has occurred, marking it as "sharp pressure drop," and combining it with the timestamp 10:18:05 to form a record (10:18:05, "sharp pressure drop"). The remaining rate points in mmHg / s, for example mmHg / s or For mmHg / s, the system does not set any event markers, but combines the generated markers with timestamps to obtain the intraoperative key event record table.

[0029] Specifically, such as Figure 2 , 4 As shown, the parallel node identification module includes: The event data acquisition submodule extracts the timestamp and operation identifier of each operation node based on the intraoperative key event record table, rearranges the timestamps in chronological order and fills in the missing values, establishes a correspondence between the rearranged results and the operation identifiers, removes duplicate entries, and generates a node time sequence. Based on the intraoperative critical event record table, which contains pressure surge events, and combined with other manually recorded or automatically generated standard operation nodes in the surgical procedure, as shown in Table 2, a set of original records is first extracted one by one from each operation node, along with its corresponding operation identifier, to obtain an initial set of key-value pairs. For example, (10:15:22, "Pressure surges sharply"), (10:20:05, "Hepatic vein puncture"), (10:18:05, "Pressure drops sharply"), (10:22:15, "Shunting pathway established"), (10:25:40, "Stent placement"), (10:25:45, "Imaging verification"), (10:22 :15, "Establishing a shunt"), then, the items in the set are sorted in ascending order according to the timestamp field. The sorted sequence is (10:15:22, "Pressure rises sharply"), (10:18:05, "Pressure drops sharply"), (10:20:05, "Puncture of hepatic vein"), (10:22:15, "Establishing a shunt"), (10:22:15, "Establishing a shunt"), (10:25:40, "Placement of stent"), (10:25:45, "Image verification"). During the sorting process, missing values ​​are filled in. This process checks whether there are two logically consecutive timestamps that are separated by a timestamp interval exceeding the preset maximum reasonable interval (e.g., If an operation (within minutes) exists, insert a node marked "Unknown Operation" at that position, with its timestamp set to the previous node's timestamp plus half the maximum reasonable interval. For example, if the next record after "Establish Branch Channel" is... The surgery was "over" minutes later, which was the middle of the two periods. An "Unknown Operation" node is inserted at the minute mark. Then, the rearranged results are re-established to correspond one-to-one with the operation identifiers, and the operation of removing duplicate entries is performed. This operation compares two adjacent entries. If the timestamp and operation identifier are exactly the same, the latter entry is deleted. In the above sequence, there are two identical records (10:22:15, "Establish a branch channel"), so one of them is deleted. Finally, a sequence without duplicates and arranged in chronological order is obtained, generating the node time sequence.

[0030] Table 2: Combined Record of Surgical Procedures and Events 10:20:05 Hepatic vein puncture 10:25:40 Placement bracket 10:15:22 The pressure rose sharply 10:22:15 Establish a diversion channel 10:25:45 Image verification 10:18:05 The pressure dropped sharply. 10:22:15 Establish a diversion channel As shown in Table 2, this table lists the mixed event records collected during a surgical procedure, including timestamps and the corresponding specific operation or key event names. The data was not sorted by time in the initial state and there were duplicates.

[0031] The node feature classification submodule calls the node time sequence to extract the time difference between adjacent nodes, inputs the frequency of the time difference and operation identifier combination into the Bayesian classifier, calculates the conditional probability and compares it with the parallel requirement threshold, divides the combinations above the threshold into parallel candidate branches, and generates a node classification vector group. The node time sequence is called, which is [(10:15:22, "Sharp Pressure Increase"), (10:18:05, "Sharp Pressure Decrease"), (10:20:05, "Hepatic Vein Puncture"), (10:22:15, "Shunting Channel Establishment"), (10:25:40, "Stent Placement"), (10:25:45, "Imaging Verification")]. First, adjacent nodes are extracted from this sequence, and the time difference between them is calculated. For example, the timestamp of the "Stent Placement" node is 10:25:40, and the timestamp of the "Imaging Verification" node is 10:25:45. The time difference between them is... seconds, this time difference The second, combined with the pair of operation identifiers ("stent placement", "image verification"), constitutes a feature unit. This feature unit is then input into a classification process trained on raw surgical data. Specifically, it is derived from a dataset containing... For example, by analyzing the required probabilities in a database of previous TIPS surgical records, it was found that the combination of "stent placement" followed immediately by "image verification" occurred [number missing]. Next time, in this In this case, the time difference is less than There are cases in seconds. The total number of operations in the database that were marked as "parallelizable" by experts was [number] times. The combination of "placement of support" and "image verification" accounted for the majority of the cases. Next, based on statistical data, calculate the current combination ("stent placement", "image verification") when the time difference is less than... The conditional probability that a combination is "parallelizable" under the condition of a time difference of less than a second is calculated as follows: The number of times in seconds divided by the time difference of the combination is less than "Total number of seconds", that is Then, this calculation result is compared with a preset parallel demand threshold, which is set based on sorting the conditional probabilities calculated from the operation combinations in the original data, and taking the top one. The probability value is used as a candidate parallel operation, and its boundary point probability value is... The threshold is set to The calculated conditional probability With threshold Compare, Below Therefore, the combination of "stent placement" and "image verification" is not classified as a parallel candidate branch under the current data. If there is another combination of operations ("balloon dilation" and "pressure measurement"), the time difference is... The conditional probability calculated in the same way for seconds is: ,because Above the threshold The combination (“balloon dilation”, “pressure measurement”) is divided into parallel candidate branches, and the identified parallel candidate branches are aggregated to generate a node classification vector group.

[0032] The split structure generation submodule filters the support operation and image verification nodes based on the node classification vector group, reassigns the temporal positions of the filtered nodes according to the time difference sequence, analyzes the two independent paths and fits the node connection index table to obtain the journey node split map. Based on the node classification vector group, assuming it contains an identified parallel candidate branch: ("Stabilization", "Image Verification"), firstly, combinations containing the two specific types of nodes "Stabilization" and "Image Verification" are selected from this vector group. The combination ("Stabilization", "Image Verification") is successfully selected. Next, the two selected nodes are reassigned in the logical flow according to the time difference sequence between them. The original time sequence is a linear sequence: ... → (10:25:40, "Stabilization") → (10:25:45, "Stabilization") (Image Verification) →… Now, adjust the positions of these two nodes to start in parallel, that is, both start execution at 10:25:40, forming two independent paths. Path 1 is the "Stabilization Operation" path, containing the node "Stabilization Placement", and Path 2 is the "Image Verification" path, containing the node "Image Verification". Subsequently, analyze these two independent paths and construct a node connection index table to describe the new process structure. The construction process of this table is to assign a unique ID to each node in the process and clarify its predecessor and successor nodes. For example, let the ID of "Establishing a Branch Channel" be… The ID for "Surgery Completed" is The ID for "placement bracket" is The ID for "Image Verification" is In the old index table, The successor is , The precursor is The successor is , The precursor is The successor is In the new index table, The subsequent modification was as follows and , The predecessor was modified to The subsequent modifications were as follows: , The front-wheel drive was also modified to The subsequent modifications were as follows: ,and The front end was changed to and By updating the connection relationships of associated nodes in this way, a complete node connection index table is finally fitted, resulting in a journey node split graph.

[0033] Specifically, such as Figure 2 , 5 As shown, the cross-departmental collaborative mapping module includes: The synchronous signal acquisition submodule calls the journey node split map to obtain the contrast agent injection signal sequence and stent release start signal input from the radiology and interventional departments. It compares the two types of signal sequences and detects the start and end markers, and removes data that does not conform to the time reference scale to obtain the synchronous signal sequence set. The system invokes the journey node split graph, which logically defines the "stent placement" and "image verification" nodes as operable in parallel. Based on this, the system acquires real-time operation signals from different sources: a digital signal sequence labeled "stent release" from the interventional radiology department's surgical control console, and a signal sequence labeled "contrast agent injection" from the radiology department's digital subtraction angiography (DSA) equipment, as shown in Table 3. The acquired raw signals include timestamps, the source department, and the specific operation signal. For example, the "stent release_start" signal issued by the interventional radiology department at 10:25:40.5 and the "contrast agent injection_start" signal issued by the radiology department at 10:25:41.2. Next, these two signal sequences are compared; the core of this comparison is to detect their... The system uses paired start and end markers. A complete operation is defined as a combination of a "start" signal and an "end" signal. For example, the system identifies "stent release_start" at 10:25:40.5 and "stent release_end" at 10:25:48.1, constituting a complete stent release event. The system then compares the detected event with a preset time reference scale. This scale sets a reasonable time window based on the stages of the "stent placement" and "image verification" nodes in the journey node split diagram. For example, from 10:25:00 to 10:28:00, the system finds a "contrast agent injection_start" signal with a timestamp of 10:35:10 in the radiology signal sequence. Because of this timestamp... If the data exceeds the upper limit of the preset time window [10:25:00, 10:28:00], it is determined that the data point is invalid data due to clock asynchrony or recording error, and it is removed from the sequence. After this filtering and matching, the signals within the valid time window and with complete start and end identifiers are retained to obtain the synchronization signal sequence set.

[0034] Table 3: Cross-departmental Original Operation Signal Table 10:25:40.5 Interventional Department Stent Release - Initiation 10:25:41.2 Radiology Department Contrast agent injection - initiation 10:25:48.1 Interventional Department Stent release - End 10:35:10 Radiology Department Contrast agent injection - initiation 10:25:49.3 Radiology Department Contrast agent injection - End As shown in Table 3, this table records the original operation signals directly collected from the interventional radiology and radiology equipment, including timestamps, signal sources, and specific start or end operation identifiers, showing the out-of-order and invalid timestamps that existed before data synchronization.

[0035] The time difference calculation submodule, based on the set of synchronization signal sequences, calls the timestamps of the contrast agent injection start point and the stent release start point, compares the two types of timestamps item by item and extracts the offset difference, compares the offset difference with the preset time base threshold item by item and divides the interval to obtain the cross-sequence time difference set. Based on the synchronization signal sequence set, this set contains valid operational events that have been checked and paired, such as a pair of key start timestamps: the contrast agent injection start timestamp. The timestamp is 10:25:41.2, and the timestamp is the start point of the stent release. The timestamps are 10:25:40.5. The system first retrieves these two timestamps, compares them item by item, and extracts the offset difference between them. The calculation process is as follows: minus ,Right now Seconds, this positive value indicates that the contrast agent injection occurred later than the stent release. Seconds later, the calculated offset difference is compared item by item with a preset time reference threshold. This threshold is set based on best practices for the coordinated operation of contrast agent and stent placement in the clinical TIPS surgical guidelines. The purpose is to ensure that the contrast agent can optimally fill the target vessel when the stent is fully deployed. The threshold is specifically divided into three intervals: the delay interval (offset difference greater than 1 second). (seconds), indicating that the contrast agent was injected too late, and the synchronization interval (offset difference within) Instant (between seconds), indicating ideal injection timing, and an advance interval (offset difference less than 1 second). (seconds), indicating that the contrast agent was injected too early, will cause the offset difference calculated in the previous step to be adjusted. Comparing seconds with intervals, because This difference is assigned to the "synchronization interval". If, in another operation, the calculated offset difference is... seconds, because it is greater than The time difference between the two time intervals is divided into "delay intervals" and this calculation and division process is repeated for each pair of matching operations (such as a balloon dilation and a pressure measurement) in the set of synchronous signal sequences to obtain a set of cross-sequence time differences.

[0036] The collaborative journey generation submodule, for cross-sequence time difference sets, calls the signal segment operation identifiers of the radiology and interventional departments, sorts the difference results and maps them to the journey node split graph, connects the correspondence between operations, and generates a cross-departmental collaborative journey graph. For cross-sequence time difference sets, this set includes, for example: "(contrast agent injection - stent release, "(seconds, synchronization interval)" and "(contrast agent injection - stent release, For data entries like "seconds, delay interval," the operation identifiers of the radiology and interventional radiology signal segments associated with the difference results are first called, namely "contrast agent injection" and "stent release." Then, the difference results are sorted according to their occurrence time during the procedure and mapped one by one to the corresponding parallel nodes in the journey node split graph. Specifically, in the original split graph, the "stent placement" and "image verification" nodes are displayed side by side. Now, the graph is refined based on the difference results. The system represents the difference in seconds by drawing a directed edge from the "Place Support" node to the "Image Verification" node in the graph, and labels this edge with "...". "seconds" indicates that the "image verification" operation started after the "stent placement" operation began. Starts in seconds, for the other one The difference in seconds is recorded and marked with "" at the corresponding position on the graph. The time difference is represented by a number of seconds, and its corresponding "delay interval" is identified by color or label (e.g., a red label for "delay"). In this way, the calculated time difference is represented by directed edges with time delay and status labels, connecting multiple operation nodes that were originally parallel on the graph. This visually connects the actual correspondence between operations in different departments, generating a cross-departmental collaborative journey graph.

[0037] Specifically, such as Figure 2 , 6 As shown, the journey path reconstruction module includes: The node status acquisition submodule acquires the patient arrival time and stay duration sequence of multiple nodes in the cross-departmental collaborative journey map, establishes an index relationship between the arrival time sequence and the stay duration, counts the number of patients at each time point and pairs them with the stay duration, and generates a node status matrix. Obtain a cross-departmental collaborative journey map, which defines several key operational nodes in the TIPS surgical procedure, such as... "Hepatic vein puncture" "Establish diversion channels" "Placement bracket" "Image verification" and "End surgery," retrieve from the original surgical database. For example, the complete log of a TIPS procedure is used to extract the specific arrival and departure times of each patient at each node, forming a multi-node sequence of patient arrival times and stay durations, as shown in Table 4. Next, a one-to-one index relationship is established between the arrival time sequence of each node and the stay duration calculated based on the arrival and departure times (stay duration = departure time - arrival time). For example, for a patient... At the node "Establish a diversion channel", its arrival time is The duration of stay is Seconds, forming a record ( , , , s), then, in order to statistically analyze the concurrency situation, the entire surgical timeline was divided into Divide the data into minutes to form a series of time points, for example... =10:20, =10:21, =10:22, etc., the system iterates through each time point, counts the number of patients in multiple nodes within that minute, and pairs the average stay time of patients in that node within that time point. For example, at 10:22, the system detected the patient. and All are at nodes Therefore, the number of patients at that time point was At the same time, calculate the two patients' conditions. The total duration of stay were respectively Seconds and seconds, its average value is Seconds, finally, the statistical data of the nodes at each point in time are integrated, where the rows of the matrix represent the nodes. to Columns represent time points The elements in the matrix are a pair (number of patients, average length of stay), for example, the matrix in ( , The value at position () , s), generate the node state matrix.

[0038] Table 4: Patient Milestone Time Record Table P01 N1 10:18:30 10:20:05 P01 N2 10:20:05 10:22:15 P01 N3 10:25:40 10:25:48 P02 N1 10:19:10 10:21:00 P02 N2 10:21:00 10:23:25 P03 N1 10:22:00 10:24:00 As shown in Table 4, this table lists the arrival and departure times of some patients at different surgical nodes, which are the basic data for calculating the duration of stay and the concurrent status of nodes.

[0039] The Markov transition probability submodule calls the patient number and stay duration data in the node state matrix, constructs the state transition relationship based on the patient flow between adjacent nodes, calculates the ratio of patient flow frequency to the total number of patients in the node per unit time, and generates the Markov transition probability matrix. The system retrieves patient count and dwell time data from the node state matrix and combines it with the original patient path information. First, it constructs transition relationships between states based on the actual flow of patients between adjacent nodes, with each state representing an operation node. This is done by traversing... For example, surgical logs, statistics from a single node. Flowing directly to another node The frequency of patients, for example, systematic statistics found that after completing "Hepatic vein puncture" procedure Among the patients, there are The name went directly into "Establish diversion channels" The puncture failed and the patient needs to return and repeat the procedure. Operation, therefore, from arrive The transfer frequency is ,from arrive The transfer frequency is ,from The total frequency of departures is Next, the ratio of the frequency of patient movement per unit time (referring to the entire observation period) to the total number of patients leaving the node is calculated to obtain the state transition probability. , its arrival transition probability The calculation process is as follows: arrive transfer frequency Divided from Total number of outflowing patients The result was obtained. Similarly, For parallel nodes "Placement bracket" and "Image verification", assuming in after, All patients who successfully established shunt channels entered both procedures simultaneously, and then merged into the next stage upon completion. "End the surgery" , , and from Departure All patients were transferred to ,therefore The system repeats this calculation process for the transitions between nodes, and finally fills the probability value into a... In the matrix ( (where the total number of nodes is ), where the th Line 1 The elements of the column represent the nodes. Transfer to node The probability is used to generate the Markov transition probability matrix.

[0040] The path generation submodule performs a weighted summation of the cumulative transition probabilities between nodes based on the state transition values ​​in the Markov transition probability matrix, selects the first path according to the cumulative transition probability, and generates an optimized journey map. Based on the state transition values ​​in the Markov transition probability matrix, which quantifies the probability of a patient transferring between different surgical nodes, the matrix shows, for example, the matrix displays... , , And the probability of a non-ideal path. (Assuming that after the diversion channel is established, there is) (The probability of needing to re-puncture) is calculated by the system through cumulative transition probabilities of the paths between nodes. This calculation is not a direct summation, but rather a multiplication of the transition probabilities of each step on the path to obtain the total probability of the path occurring. For example, an ideal path... for Its cumulative transfer probability is Another path that involves rework for Its cumulative transfer probability is Subsequently, the system sorts the calculated cumulative probabilities of the paths, arranging the probability values ​​from high to low. ,therefore The sorting is higher than The system selects the path that ranks first, i.e., the path with the highest cumulative transfer probability. (Parallel) As the most frequent and efficient main process experienced by patients, it generates an optimized journey map.

[0041] Specifically, such as Figure 2 , 7 As shown, the experience optimization decision module includes: The journey data acquisition submodule calls the patient's waiting time and operation time data in the optimized journey map, calculates the length of the waiting interval and operation interval respectively, sorts the two types of time data according to the node number, and converts them into a matrix structure to generate node time series data. Call the optimized journey map defined "Hepatic vein puncture" "Establish diversion channels" "Placement bracket" "Image verification" and At points such as "End of surgery", the system starts from... For example, extract the waiting time and operation time data for each patient between nodes from the original surgical record. The waiting time is calculated by subtracting the departure time of the previous node from the arrival time of the next node. For example, for patients... leave The time is ,arrive The time is , its in The waiting time is The operation time is calculated as the departure time minus the arrival time within the same node, for example, for a patient. exist The duration of stay is minus ,Right now Seconds, this seconds is The system's operation time The waiting time and operation time data for each node in the example sample are summarized, and their average values ​​are calculated to obtain the average waiting interval length and average operation interval length for each node. For example, after calculation, The average waiting time is seconds, average operation time is Second, The average waiting time is seconds, average operation time is Seconds, then, these two types of average duration data are sorted according to node number. Sort the data in the specified order and transform the sorted result into a... The matrix structure is shown in Table 5, where the first column is the average waiting time, the second column is the average operation time, and the row index corresponds to the node number, generating node time series data.

[0042] Table 5: Average Duration Data for Nodes N1 180 90 N2 120 135 N3 45 60 N4 40 65 N5 0 15 As shown in Table 5, this table summarizes the average waiting time and average operation time of key nodes in the optimized journey map through statistical calculations, providing a quantitative time data basis for subsequent bottleneck analysis.

[0043] The node ratio calculation submodule calls the node time series data, calculates the ratio for each node based on the waiting time and operation time, compares the ratio with the node time threshold benchmark, records the node ratio results, and generates a node time ratio sequence. The system retrieves node time-series data, which is a matrix of average waiting time and average operation time for each node. Based on this matrix, the system calculates the ratio for each node by dividing the node's average waiting time by its average operation time. For example, its ratio is calculated as follows: Seconds divided by Seconds to get the result For nodes The ratio is calculated as follows: Divide by seconds In seconds, the result was obtained in approximately Next, the calculated ratio for each node is compared with a preset node duration threshold, which is set based on the original... For example, the surgery with the shortest total time Example (i.e., the most efficient) ) The sample was analyzed, and this was calculated. Example: The ratio of waiting time to operation time of nodes in the sample, sorted in ascending order, and taking the first value. Using the percentile value as a benchmark, this value is calculated. This value is set as the node duration threshold benchmark, and the system will use the node... ratio With threshold Compare, Greater than , will node ratio With threshold Compare, Less than The system sequentially calculates and compares the ratios of the nodes, records the ratio result for each node, and finally obtains an array corresponding to the order of the nodes, i.e. ( ), generate a sequence of node duration ratios.

[0044] The bottleneck node marking submodule detects whether the ratio is greater than the node duration threshold based on the node duration ratio sequence, stores and marks the node number that exceeds the threshold with the ratio as the experience bottleneck node, and performs structured processing on the node according to the optimization strategy rules to generate a patient experience optimization plan. Based on the node duration ratio sequence ( Each sequence is checked to see if the ratio is greater than the node duration threshold set in the previous steps. The detection revealed that the first element Greater than Therefore, the node corresponding to this value is numbered. and ratio Perform paired storage to form records ( , ), and for nodes Add a "bottleneck node" marker, where the ratios of all other nodes are less than [a certain value]. Nodes marked as bottlenecks are not initially identified. Instead, based on a pre-defined optimization strategy rule base, the nodes are processed in a structured manner. This rule base includes the following rules: If the waiting time of a bottleneck node is significantly higher than its operation time (i.e., the ratio is greater than...), then... If this occurs, the "preceding process check" rule is triggered, and the system immediately retrieves the node. The pre-operative log of "hepatic vein puncture," which includes data from the patient's entry into the operating room to the preparation stage before surgery, reveals the average time spent in this stage. The waiting time is consumed in the "preoperative imaging data retrieval and confirmation" step. Based on this analysis, the system has structured this specific step and proposed an operational adjustment suggestion: to move the "preoperative imaging data retrieval and confirmation" step to the waiting area before the patient enters the operating room, instead of occupying the main operation time on the operating table. This node marker, problem location, and structured adjustment suggestion are integrated to generate a patient experience optimization plan.

[0045] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A TIPS patient medical experience optimization system based on journey maps, characterized in that, The system includes: The intraoperative parameter acquisition module acquires portal vein pressure values ​​through hemodynamic monitoring equipment, calculates the rate of change of portal vein pressure values ​​over a continuous time period, and marks a pressure mutation event when the rate of change exceeds a preset threshold, generates an intraoperative key event record table, and transmits it to the parallel node identification module. The parallel node identification module uses a Bayesian classifier to identify parallel operation requirements based on the intraoperative key event record table. It splits a single surgical node into two parallel branches: stent operation and image verification, generates a journey node split map, and transmits it to the cross-departmental collaborative mapping module. The cross-departmental collaborative mapping module calls the journey node split map to establish the collaborative relationship between the radiology department and the interventional department. By comparing the time difference between the contrast agent injection completion signal and the stent release initiation signal, a cross-departmental collaborative journey map is generated and transmitted to the journey path reconstruction module. The journey path reconstruction module redesigns the patient's medical process based on the cross-departmental collaborative journey map. Combining patient waiting and operation time, it uses Markov decision process to optimize the transfer path between nodes, generates an optimized journey map, and transmits it to the experience optimization decision module. The parallel node identification module includes: The event data acquisition submodule extracts the timestamp and operation identifier of each operation node based on the intraoperative key event record table, rearranges the timestamps in chronological order and fills in the missing values, establishes a correspondence between the rearranged results and the operation identifier, removes duplicate entries, and generates a node time sequence. The node feature classification submodule calls the node time sequence to extract the time difference between adjacent nodes, inputs the time difference and operation identifier combination frequency into the Bayesian classifier, calculates the conditional probability and compares it with the parallel requirement threshold, divides the combinations above the threshold into parallel candidate branches, and generates a node classification vector group. The split structure generation submodule filters support operation and image verification nodes according to the node classification vector group, reallocates the temporal positions of the filtered nodes according to the time difference sequence, analyzes two independent paths and fits the node connection index table to obtain the journey node split map. The node time sequence refers to the sequence of operation nodes extracted from the intraoperative key event record table, arranged in timestamp order, with missing values ​​filled in and duplicate entries removed; The parallel requirement threshold refers to the probability threshold for distinguishing parallel candidate branches, calculated by combining the node time difference and the frequency of operation identifier combinations. The node classification vector group is a set of vectors formed by combining the time difference extracted from the node time sequence with the operation identifier and performing classification calculations.

2. The TIPS patient medical experience optimization system based on journey map according to claim 1, characterized in that, The intraoperative critical event record table includes event type, event time, and event duration; the journey node split graph includes operation branches, verification branches, and branch relationships; the cross-departmental collaborative journey graph includes collaborating departments, signal difference, and collaboration strength; and the optimized journey map includes a transition probability matrix, path sequence, and value function.

3. The TIPS patient medical experience optimization system based on journey map according to claim 1, characterized in that, The intraoperative parameter acquisition module includes: The data stream receiving submodule acquires portal vein pressure value sequence data through hemodynamic monitoring equipment, arranges the sequence by timestamp, compares the pressure fluctuation amplitude of adjacent points with the sampling interval, removes abnormal points and integrates them into a continuous time period to obtain portal vein pressure sequence values. The rate of change calculation submodule calls the portal vein pressure sequence value, calculates the ratio of the pressure difference between adjacent time points to the time interval, obtains the rate value of the time interval and arranges it in chronological order, performs scale correction on the rate fluctuation range, and generates a portal vein pressure rate of change sequence. The event marker generation submodule calls the portal vein pressure change rate sequence, compares the rate value with the preset pressure mutation threshold point by point, sets the event marker when the rate value exceeds the threshold, and combines the marker with the timestamp to obtain the intraoperative key event record table.

4. The TIPS patient medical experience optimization system based on journey map according to claim 3, characterized in that, The portal vein pressure sequence value refers to the set of portal vein pressure data points collected by hemodynamic monitoring equipment and arranged in time stamp order; The portal vein pressure change rate sequence refers to the set of data obtained by dividing the pressure difference between adjacent data points in the portal vein pressure sequence by the corresponding time interval and arranged in chronological order. The preset pressure mutation threshold refers to a pre-defined numerical limit that is compared with data points in the portal vein pressure change rate sequence.

5. The TIPS patient medical experience optimization system based on journey map according to claim 1, characterized in that, The cross-departmental collaborative mapping module includes: The synchronous signal acquisition submodule calls the journey node split map to obtain the contrast agent injection signal sequence and stent release start signal input from the radiology and interventional departments. It compares the two types of signal sequences and detects the start and end markers, removes data that does not conform to the time reference scale, and obtains the synchronous signal sequence set. The time difference calculation submodule, based on the set of synchronization signal sequences, calls the timestamps of the contrast agent injection start point and the stent release start point, compares the two types of timestamps item by item and extracts the offset difference, compares the offset difference with the preset time reference threshold item by item and divides the interval to obtain the cross-sequence time difference set. The collaborative journey generation submodule, for the cross-sequence time difference set, calls the signal segment operation identifiers of the radiology and interventional departments, sorts the difference results and maps them to the journey node split graph, connects the correspondence between operations, and generates a cross-departmental collaborative journey graph.

6. The TIPS patient medical experience optimization system based on journey map according to claim 1, characterized in that, The journey path reconstruction module includes: The node status acquisition submodule acquires the patient arrival time and stay duration sequence of multiple nodes in the cross-departmental collaborative journey map, establishes an index relationship between the arrival time sequence and the stay duration, counts the number of patients at each time point and pairs them with the stay duration, and generates a node status matrix. The Markov transition probability submodule calls the patient number and stay duration data in the node state matrix, constructs the state transition relationship based on the patient flow between adjacent nodes, calculates the ratio of patient flow frequency to the total number of patients in the node per unit time, and generates the Markov transition probability matrix. The path generation submodule performs a weighted summation of the cumulative transition probabilities between nodes based on the state transition values ​​in the Markov transition probability matrix, selects the first path according to the cumulative transition probability, and generates an optimized journey map.

7. The TIPS patient medical experience optimization system based on journey map according to claim 1, characterized in that, The experience optimization decision module evaluates the patient's experience satisfaction at the journey nodes through the optimized journey map, calculates the ratio of waiting time to operation time at the node, marks the node as an experience bottleneck node when the ratio exceeds a set threshold, and outputs the patient experience optimization plan. The patient experience optimization plan includes bottleneck location, ratio threshold, and adjustment suggestions.

8. The TIPS patient medical experience optimization system based on journey map according to claim 7, characterized in that, The experience optimization decision module includes: The journey data acquisition submodule calls the patient's waiting time and operation time data in the optimized journey map, calculates the length of the waiting interval and operation interval respectively, sorts the two types of time data according to the node number, and converts them into a matrix structure to generate node time series data. The node ratio calculation submodule calls the node time series data, calculates the ratio for each node based on the waiting time and operation time, compares the ratio with the node time threshold benchmark, records the node ratio results, and generates a node time ratio sequence. The bottleneck node marking submodule detects whether the ratio is greater than the node duration threshold based on the node duration ratio sequence, stores and marks the node numbers that exceed the threshold with the ratio as experience bottleneck nodes, and performs structured processing on the nodes according to the optimization strategy rules to generate a patient experience optimization plan.

Images