Intelligent follow-up visit system and method for daytime tumor patients, medium and equipment

Abstract

The invention relates to the technical field of intelligent follow-up visit, in particular to an intelligent follow-up visit system and method for daytime tumor patients, a medium and equipment, and the system comprises a physiological monitoring module, a drug interaction analysis module, a side effect mapping module, a health risk early warning module and an intelligent inquiry optimization module. According to the method, through continuous monitoring and real-time comparison of the physiological data of the patient and effective archiving of data fluctuation in a key period, instant and accurate reference is provided for clinical decision making, deep drug interaction analysis enables drug adjustment to be more targeted, unsuitable drug combinations are reduced, and therefore potential health risks of the patient are relieved; through side effect mapping, the system can predict and prevent potential side effects, the treatment safety is guaranteed, the risk early warning module enables the system to identify major health risks in the first time, the response speed of emergency medical events is improved, and the system can be used for improving the treatment efficiency by analyzing the interaction history of patients and optimizing the inquiry process. And the medical experience of the patient is more targeted and practical.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent follow-up, and particularly to an intelligent follow-up system, method, medium and device for daytime cancer patients. Background Art

[0002] Intelligent follow-up involves using information technology to track and manage patients' health conditions, especially in the long-term health management after patients are discharged from the hospital. It usually includes automated health monitoring, data collection, analysis and feedback mechanisms, which can update patients' health data in real time and process the data through intelligent algorithms to support medical decision-making and patient management.

[0003] Among them, the intelligent follow-up system for daytime cancer patients is a system for daytime tracking and management of cancer patients. The main purpose is to provide continuous medical support and health monitoring during the non-hospitalization period of patients. By using data analysis and intelligent algorithms, the system can monitor the changes in patients' conditions, regularly update treatment suggestions, and at the same time provide drug management and side effect monitoring. It can communicate with patients through intelligent devices to ensure that patients can also receive appropriate medical attention and guidance in the home environment.

[0004] The prior art cannot achieve in-depth drug interaction analysis, which limits its application effect in complex treatment scenarios, resulting in the inability to effectively prevent drug-induced side effects or health problems. The lack of highly targeted health risk assessment and consultation optimization makes it difficult for patients to obtain treatment suggestions customized according to their condition changes, affecting the timeliness and accuracy of treatment. The prior art fails to provide sufficiently flexible monitoring and analysis tools to cope with individual differences, which is particularly important in the management of daytime cancer patients because patients need continuous and targeted attention during the non-hospitalization period. Summary of the Invention

[0005] The purpose of the present invention is to solve the drawbacks existing in the prior art, and to propose an intelligent follow-up system, method, medium and device for daytime cancer patients.

[0006] To achieve the above purpose, the present invention adopts the following technical scheme: An intelligent follow-up system for daytime cancer patients, the system includes: The physiological monitoring module, based on wearable devices, obtains patients' heart rate, systolic blood pressure and body temperature data, records the situation where the data continuously deviates from the normal range within any time period, and obtains a physiological fluctuation record; The drug interaction analysis module, based on the physiological fluctuation record, obtains patients' daily drug intake information, compares the synergistic effects and reaction types between drugs, and identifies drug combinations associated with serious health disturbances to obtain drug sensitivity identification information; The side effect mapping module analyzes the side effect information reported by the patient according to the drug sensitivity identification information, and compares it with the drug name list to identify drug and symptom combinations with cross-effect risks, and obtains the overlapping side effect analysis result; Based on the overlapping side effect analysis result, the health risk warning module combines the patient's real-time heart rate, blood pressure and body temperature data, analyzes the overlapping area of abnormal peaks within the analysis period, evaluates the health risks brought by the overlapping situation, and obtains the risk fluctuation monitoring data; The intelligent consultation optimization module calls the risk fluctuation monitoring data, analyzes the interaction history of the patient participating in the intelligent consultation interface, and reorders the consultation path according to the matching degree between the keywords and the patient's condition, and obtains the optimized configuration of the consultation path.

[0007] The improvements of the present invention are as follows: the physiological fluctuation record includes the floating range, duration, and key floating period; the drug sensitivity identification information includes the list of sensitive drugs, drug interaction intensity, and health impact rating; the overlapping side effect analysis result is specifically the side effect severity, drug side effect relevance, and drug cross-risk index; the risk fluctuation monitoring data includes the peak overlap frequency, abnormal fluctuation duration period, and risk warning level; the optimized configuration of the consultation path includes the optimized consultation frequency and keyword priority adjustment.

[0008] The improvements of the present invention are as follows: the physiological monitoring module includes: The signal acquisition sub-module, based on the wearable device, records the patient's heart rate, systolic blood pressure and body temperature data in real time, performs time serialization processing on each monitoring parameter, and synchronously records the acquisition timestamp of each parameter to obtain a continuous physiological data sequence; The physiological parameter judgment sub-module sets an abnormal floating standard interval for each physiological monitoring parameter based on the continuous physiological data sequence, conducts data comparison, identifies data points that exceed the standard interval, and classifies continuous abnormal data points according to time intervals to obtain an abnormal data identification set; The physiological fluctuation archiving sub-module calls the abnormal data identification set, analyzes the abnormal data time period, records the start and end times of each abnormal data, and archives the deviation period as a key information segment to obtain the physiological fluctuation record.

[0009] The improvements of the present invention are as follows: the drug interaction analysis module includes: The drug information extraction sub-module obtains the patient's daily drug intake information based on the physiological fluctuation record, analyzes the type and dose changes of each drug by identifying drug changes within the time window, and obtains the drug combination change data for each time window; Based on the drug combination change data, the drug synergy reaction comparison sub-module accesses the drug database, compares the mechanism of action and synergy effect attributes of drugs, classifies and matches the functional categories and metabolic pathways of drugs, and forms synergy reaction data for each drug combination; The interference combination identification sub-module screens out drug combinations with high risks according to the synergy reaction data, analyzes the interaction intensity between the action type of the drug combination and physiological fluctuations, calculates risk interaction data, and judges the risk level of the drug combination to obtain drug sensitivity identification information.

[0010] The present invention is improved in that the side effect mapping module includes: The sensitivity identification sub-module extracts data from the drug name, action area and reaction level based on the drug sensitivity identification information, and determines the sensitivity level of the drug by comparing with the sensitivity grading benchmark to obtain a sensitivity level score; The side effect matching sub-module calls the sensitivity level score, analyzes the side effect information reported by the patient, extracts symptom keywords, performs cross-search with the drug name list, judges whether the side effect manifestation appears in the action area recorded in the drug instruction manual, and records the occurrence times and matching degree to obtain a side effect manifestation matching rate; The overlapping risk assessment sub-module identifies the overlapping risks in the drug action area and symptom combination according to the side effect manifestation matching rate, calculates the cross-action risk value, compares it with the risk benchmark, and identifies the risk combinations exceeding the benchmark to obtain an overlapping side effect analysis result.

[0011] The present invention is improved in that the health risk warning module includes: Based on the overlapping side effect analysis result, the abnormal peak extraction sub-module combines the real-time heart rate, blood pressure and body temperature data of the patient, respectively detects the fluctuation trend of each type of parameter within the analysis period, extracts the abnormal peak points in the fluctuation curve that meet the physiological extreme value standard, and screens out the interference fluctuations whose duration does not meet the peak duration benchmark to obtain an abnormal physiological peak set; Based on the abnormal physiological peak set, the coincidence risk calculation sub-module obtains the total overlap risk coefficient according to the distribution on the time axis, and divides the corresponding warning levels according to its interval to obtain a risk level interval; Based on the risk level interval, the risk fluctuation marking sub-module screens out the fluctuation sections that maintain the same or increasing level interval in multiple consecutive periods, judges whether the level interval fluctuation exceeds the risk priority marking benchmark, and marks the abnormal fluctuations as high-priority monitoring objects to obtain risk fluctuation monitoring data.

[0012] The present invention is improved in that the intelligent consultation optimization module includes: The keyword extraction submodule calls the risk fluctuation monitoring data, analyzes the patient's interaction history in the intelligent consultation interface, obtains the interactive sentence text data, extracts the keyword groups in the consultation sentences and records their occurrence frequencies, and then sorts the keywords according to the frequencies to establish a high-frequency keyword set; The matching degree calculation submodule calculates the matching degree between the keywords and the patient's condition based on the high-frequency keyword set and the patient's current health data, performs keyword screening, removes keywords that do not meet the conditions, and obtains keyword screening weights; The path rearrangement submodule re-evaluates the relevance to each node in the consultation path according to the keyword screening weights, adjusts the weights of the original path, re-orders the consultation path, and updates the key interaction points of the consultation to obtain an optimized configuration of the consultation path.

[0013] A method for intelligent follow-up of daytime cancer patients, which is performed based on the above-mentioned intelligent follow-up system for daytime cancer patients, comprises the following steps: S1: Based on the wearable device, the patient's heart rate, systolic blood pressure and body temperature data are obtained, and the abnormal floating standard interval is set for each monitoring parameter. The real-time collected physiological data is continuously compared with the standard interval, and the continuous deviation from the normal range is recorded. The deviation period is archived as a key information segment to obtain a record of physiological fluctuations; S2: Based on the physiological fluctuation records, obtain the patient's daily drug intake information, analyze the newly added or adjusted drug combinations within the time window, retrieve the drug database, compare the synergistic effects and reaction types between drugs, identify drug combinations associated with serious health interference, and obtain drug sensitivity identification information; S3: Analyze the side effect information self-reported by the patient according to the drug sensitivity identification information, compare the drug name list, map the side effect content with the drug action area, identify the drug and symptom combination with the risk of cross-action, and obtain the overlapping side effect analysis results; S4: Based on the overlapping side effect analysis results, combined with the patient's real-time heart rate, blood pressure and body temperature data, analyze the overlapping areas of abnormal peaks within the cycle, evaluate the health risks brought by the overlap, mark the continuous abnormal fluctuations as high-priority monitoring objects, and obtain risk fluctuation monitoring data; S5: Call the risk fluctuation monitoring data, analyze the patient's interaction history in the intelligent consultation interface, extract the consultation keywords that appear frequently in the interaction history, reorder the consultation path according to the degree of match between the keywords and the patient's condition, and obtain the consultation path optimization configuration.

[0014] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements an intelligent follow-up system for daytime tumor patients.

[0015] A computer device includes a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, a smart follow-up system for daytime cancer patients is implemented.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, by continuously monitoring and real-time comparing the physiological data of patients, the data fluctuations in key periods are effectively archived, providing immediate and accurate references for clinical decision-making. The in-depth drug interaction analysis makes drug adjustment more targeted, reducing inappropriate drug combinations, thereby reducing the potential health risks of patients. Through side effect mapping, the system can predict and prevent potential side effects, ensuring the safety of treatment. The risk warning module enables the system to identify major health risks in the first time, improving the response speed of emergency medical events. And by analyzing the interaction history of patients, the consultation process is optimized, making the medical experience of patients more targeted and practical. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 FIG. is a module diagram of a smart follow-up system for daytime cancer patients proposed by the present invention; Figure 2 FIG. is a flowchart of the physiological monitoring module in the present invention; Figure 3 FIG. is a flowchart of the drug interaction analysis module in the present invention; Figure 4 FIG. is a flowchart of the side effect mapping module in the present invention; Figure 5 FIG. is a flowchart of the health risk warning module in the present invention; Figure 6 FIG. is a flowchart of the smart consultation optimization module in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present 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 only used to explain the present invention, and are not used to limit the present invention.

[0019] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as limiting the present invention. In addition, in the description of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0020] Embodiment Please refer to Figure 1 , the present invention provides a technical solution: an intelligent follow-up system for daytime cancer patients includes: The physiological monitoring module, based on wearable devices, acquires the patient's heart rate, systolic blood pressure, and body temperature data, sets an abnormal fluctuation standard interval for each monitoring parameter, continuously compares the real-time collected physiological data with the standard interval, records the situation where the data continuously deviates from the normal range within any time period, archives the deviation period as a key information segment, and obtains a physiological fluctuation record; The drug interaction analysis module, based on the physiological fluctuation record, acquires the patient's daily drug intake information, analyzes the newly added or adjusted drug combinations within the time window, retrieves the drug database, compares the synergistic effects and reaction types between drugs, and identifies the drug combinations associated with serious health disturbances to obtain drug sensitivity identification information; The side effect mapping module, according to the drug sensitivity identification information, analyzes the side effect information reported by the patient, refers to the drug name list, maps the side effect content and the drug action area, and identifies the drug and symptom combinations with the risk of cross-action to obtain an overlapping side effect analysis result; The health risk warning module, based on the overlapping side effect analysis result, combines the patient's real-time heart rate, blood pressure, and body temperature data, analyzes the overlapping area of abnormal wave peaks within the period, evaluates the health risks brought by the overlapping situation, marks the continuously abnormal fluctuations as high-priority monitoring objects, and obtains risk fluctuation monitoring data; The intelligent consultation optimization module calls the risk fluctuation monitoring data, analyzes the interaction history of the patient participating in the intelligent consultation interface, extracts the frequently occurring consultation keywords in the interaction history, reorders the consultation path according to the matching degree between the keywords and the patient's condition, and updates the key interaction points of the consultation to obtain an optimized configuration of the consultation path.

[0021] The physiological fluctuation record includes the fluctuation amplitude, duration, and key fluctuation period. The drug sensitivity identification information includes a list of sensitive drugs, drug interaction intensity, and health impact rating. The overlapping side effect analysis result specifically includes side effect severity, drug side effect relevance, and drug cross-risk index. The risk fluctuation monitoring data includes the wave peak overlapping frequency, abnormal fluctuation duration, and risk warning level. The optimized configuration of the consultation path includes the optimized consultation frequency and keyword priority adjustment.

[0022] Please refer to Figure 2 , the physiological monitoring module includes: The signal acquisition sub-module, based on wearable devices, records the patient's heart rate, systolic blood pressure, and body temperature data in real time, performs time serialization processing on each monitoring parameter to ensure the consistency of the time stamps of the data, and synchronously records the acquisition time stamp of each parameter to obtain a continuous physiological data sequence; The intelligent bracelet continuously worn by the patient is used to collect heart rate, systolic blood pressure and body temperature data. The photoelectric sensor inside the device continuously monitors the pulse rhythm of the wrist and extracts the number of beats per minute as the heart rate data. The pressure sensor measures the pressure change of the blood vessel during cardiac systole to generate systolic blood pressure data. At the same time, the thermistor element senses the skin surface temperature as the body temperature data. Each parameter takes 1 second as the sampling period, and a set of complete data is obtained per second. The time service is synchronously called to generate a time stamp, giving each set of data an accurate acquisition time mark. For example, at 07:00:00, a set of data is obtained with a heart rate of 78 beats per minute, a systolic blood pressure of 118 mmHg, and a body temperature of 36.5 °C. The time stamp corresponding to this set of data is "07:00:00". Subsequently, new data is obtained at 07:00:01, and so on, forming a data chain with continuous time tags. When it is recognized that the time difference between two data acquisitions is greater than 1 second, such as there is no sampling record between 07:01:10 and 07:01:12, it indicates that there is a data missing at 07:01:11. At this time, the original data at 07:01:10 and 07:01:12 will be retrieved, and the numerical averages of the heart rate, blood pressure and body temperature will be calculated respectively, and the average value will be filled into the gap at 07:01:11. In this way, the integrity and consistency of the time series are maintained. The filled data will be merged with the original data and cached in the local data pool of the device uniformly. It is set that every 60 seconds is a caching period, a complete 1-minute data packet is constructed, and it is uploaded to the server in batches to obtain a continuous physiological data sequence with a unified time stamp.

[0023] Based on the continuous physiological data sequence, the physiological parameter judgment sub-module sets an abnormal floating standard interval for each physiological monitoring parameter, conducts data comparison, identifies the data points that exceed the standard interval, and classifies the continuous abnormal data points according to the time interval to obtain the abnormal data identification set; Judgment is made item by item on the continuously collected physiological data sequence. First, the heart rate, systolic blood pressure, and body temperature data recorded per second are read, and then they are compared and judged one by one according to the set standard intervals. The set interval for heart rate is 60 - 100 beats per minute, the set interval for systolic blood pressure is 90 - 140 mmHg, and the set interval for body temperature is 36.1 - 37.2 °C. When processing each second, the heart rate data at that moment is judged against 60 and 100. If it is lower than 60 or higher than 100, it is an abnormal point. If the blood pressure data is less than 90 or exceeds 140, it is recorded as abnormal. If the body temperature is lower than 36.1 or higher than 37.2, it is classified as an abnormal body temperature point. For example, at a certain moment, the patient's heart rate is 104 beats per minute, systolic blood pressure is 136 mmHg, and body temperature is 37.6 °C. Then, both the heart rate and body temperature at that moment are abnormal. All abnormal points are classified and recorded according to the parameter type, forming three abnormal point sets. Subsequently, for each type of set, it is checked whether there is a continuous abnormal point segment in chronological order. The judgment criterion is more than three consecutive abnormal points, and the time interval between any two points does not exceed 5 seconds. If this condition is met, it is determined as an abnormal segment. For example, between 08:12:00 and 08:12:08, the heart rate recorded values per second are 102, 103, 105, 107, 104, 106, 108, 110, 109 beats per minute, all exceeding 100, which is identified as a continuous abnormal segment. The start time 08:12:00 and end time 08:12:08 are recorded, and an abnormal identification item is constructed, including the abnormal type (heart rate), abnormal start time, and end time. All such identification items form the abnormal data identification set.

[0024] The physiological fluctuation archiving sub-module calls the abnormal data identification set, analyzes the abnormal data time period, records the start and end times of each abnormal data, archives the deviated time period as a key information segment, and obtains the physiological fluctuation record; Read the parameter types of each record item and the start and end times of the occurrence of the exception, locate the complete physiological parameter sequence in the corresponding time period in the original data, extract the set of data values collected per second in this time period, and at the same time calculate the duration of this time period. For example, the duration between 08:12:00 and 08:12:08 is 9 seconds. Determine this section as a continuous physiological fluctuation segment. During the archiving process, classify and process according to the parameter types, and establish corresponding data structures for heart rate fluctuations, systolic blood pressure fluctuations, and body temperature fluctuations respectively. The structure includes: parameter category, start time, end time, duration, and the set of all sampling values in this time period. For example, a certain blood pressure fluctuation ranges from 09:10:02 to 09:10:12, lasting for 11 seconds in total, and the sampling values are 142, 145, 147, 143, 141, 139, 138, 142, 144, 145, 143 mmHg respectively. Each second corresponds to a time tag. This set of data is encapsulated into a structure and added with a parameter label of "systolic blood pressure". Subsequently, store all archived items according to the patient's unique identification number and file them according to the date. For example, the data on April 22, 2025 is archived to the "20250422" directory, create data partitions for heart rate, blood pressure, and body temperature respectively, compress and upload this type of data to the server storage area, and synchronously generate index records, including information such as patient ID, date, number of exceptions, and total duration, for subsequent query and analysis, and complete the archiving operation of physiological fluctuation records.

[0025] Please refer to Figure 3 , the drug interaction analysis module includes: The drug information extraction sub-module obtains the patient's daily drug intake information based on the physiological fluctuation records, analyzes the type and dose changes of each drug by identifying the drug changes within the time window, and obtains the drug combination change data for each time window; First, extract the start and end times recorded in the physiological fluctuation record as the boundaries of the time window, and search for all drug intake records of the patient within the corresponding time period based on these boundaries. Read the medication details corresponding to each time point of the patient from the database, including information such as drug name, dosage form, unit dose, daily usage frequency, and administration route. Subsequently, compare the drug data within two consecutive time windows one by one, extract newly added drugs, discontinued drugs, and dosage adjustments, classify and judge the change types of each drug in adjacent windows. When there is a new addition, record its first appearance time and dosage. For example, if the patient starts using azathioprine at 08:00 on April 20, 2025, 2 times a day, 50 mg each time, then mark this drug as a newly added item in this time period. If a drug existed in the previous time period but is missing in the current period, record it as discontinued. For dosage changes, calculate the numerical difference between the current dose and the previous dose. For example, if it is reduced from 100 mg to 75 mg, mark it as a 25 mg dosage reduction. Organize all change items into a drug change data table. Each record in this data table indicates the drug name, dosage value, change category (new addition, discontinuation, adjustment), and the time period label to which it belongs. After completion, organize and package the data table by time period dimension to form the drug combination change data corresponding to the time window.

[0026] The drug synergy reaction comparison sub-module, based on the drug combination change data, accesses the drug database, compares the mechanism of action and synergy effect attributes of drugs, classifies and matches the functional categories and metabolic pathways of drugs, and forms the synergy reaction data for each drug combination; Analyze each change combination data item one by one, extract the list of drug names involved as the initial analysis object. By reading the basic information fields of this type of drug in the drug database, including the mechanism of action field, indication field, metabolic pathway coding field, and pharmacological category identification field, pair any two drugs in the current drug combination pairwise, and extract the content of the mechanism of action field of the paired drugs for item-by-item literal matching judgment. When there are exactly the same or partially overlapping keywords in the mechanism of action fields of the two drugs, it is determined that there is a synergy relationship. Then, compare according to the metabolic pathway coding of each pair of drugs. When the two codes are exactly the same, mark it as a metabolic pathway intersection item. Subsequently, compare the pharmacological categories of the two drugs. If they both belong to the immunosuppressive category, mark it as a functional same-category item. Through the above three comparisons, assign a synergy weight value to each pair of combinations. If one of the three is satisfied, the weight value is recorded as 1. If two are satisfied, the weight is 2. If all three are satisfied, the value is assigned 3. Perform the above comparison process for all drug combinations within each time window, and finally form a synergy reaction data list in units of drug pairs. Each row of this list lists the paired drug names, the content of the matching fields, the metabolic pathway compliance situation, the pharmacological category consistency, and the final synergy reaction score. All paired results are sorted and output according to the score to form the synergy reaction data.

[0027] The interference combination recognition sub-module screens drug combinations with high risks based on the synergistic reaction data, analyzes the interaction intensity between the action types of the drug combinations and the physiological fluctuations, and uses the formula: ; Calculate the risk interaction data , which is used to quantify the degree of health risks generated between drug combinations, and judge the risk levels of the drug combinations to obtain drug sensitivity identification information. Among them, represents the synergistic reaction data of the th drug combination, indicating the synergistic effect and potential reaction types of this drug combination, represents the weight factor of the th drug combination. The factor is based on the severity of the drug combination in historical health interventions and is used to adjust the contribution of each drug combination to the total risk assessment, represents the difference in reaction type data of the th drug combination, reflecting the degree of difference in reaction types among different drug combinations, is the total number of drug combinations, is the number of reaction types; First, screen the drug combinations marked as "high synergy" or "reaction frequency greater than 5 times" as the risk screening target group, and combine the changes in parameters such as heart rate and blood pressure in the corresponding time period in the physiological fluctuation data. Extract the maximum fluctuation amplitude within 6 hours before and after the drug combination intake to construct an interaction response index group, calculate the quantitative connection degree between each drug combination and the physiological response it triggers, further analyze the interaction intensity between the action type of the drug combination and the physiological response data. At the same time, based on the identification of risk combinations, quantitatively integrate their synergistic reaction data, calculate the risk interaction data using the formula. Let the total number of drug combinations be , the number of reaction types be , and select 3 groups of drug combination data for calculation, as follows: The synergistic reaction data of the first group of drug combinations , the weight factor , and the value of this factor is based on the fact that this combination caused 2 moderate adverse reactions in previous cases, with each assignment of 0.55; The synergistic reaction data of the second group of drug combinations , the weight factor , and this combination involved drugs that caused 1 severe reaction and 1 mild reaction, with assignments of 0.8 and 0.5 respectively; The synergistic reaction data of the third group of drug combinations , the weight factor , and this combination only involves frequent synergy but has no clear interference record, with a medium weight assignment; The difference in reaction type data is: , , , and its quantification method is based on the calculation of structural similarity of different reaction types, normalized to between 0 and 1, where the greater the type difference, the closer the value is to 1.

[0028] The first step (molecular part): ; The second step (denominator part): ; ; Calculate : ; The result is ≈0.7484. This value reflects that under the comprehensive influence of the current drug combination on synergistic reaction, historical weight and type difference, the intensity of its causing physiological interference has approached the high-risk boundary. If the value above 0.7 in the defined reference value range is the significant interference interval, then the current result needs to be marked as "attention level" and drug sensitivity identification information is assigned.

[0029] Please refer to Figure 4 , and the side effect mapping module includes: The sensitivity identification sub-module extracts data from the drug name, action area and reaction level based on the drug sensitivity identification information, and determines the sensitivity level of the drug by comparing with the sensitivity grading benchmark to obtain the sensitivity level score; First, read each drug sensitivity record, extract the drug name field, the human anatomical region field on which the drug acts, and the initial grade identification field of the degree of response to the drug. After separating and organizing the fields into a structured data table in sequence, compare them by calling the set sensitivity grading benchmark. In this benchmark table, the common response grade division criteria corresponding to each action region are set, and the sensitivity grade is divided into five grade intervals. Among them, grade 1 represents extremely low sensitivity, grade 2 represents low sensitivity, grade 3 represents medium sensitivity, grade 4 represents high sensitivity, and grade 5 represents extremely high sensitivity. The score intervals corresponding to each grade are set as 1-20, 21-40, 41-60, 61-80, and 81-100 in sequence. By calling the response grade value of each drug and matching it with the grading benchmark value of its anatomical region, if a drug record acts on the liver and its response grade is "moderate", according to the sensitivity reference interval 41-60 corresponding to the moderate degree in the liver region, assign a sensitive grade score of 50 to this drug. If another drug acts on the central nervous system and is marked as "high", the matching value is 70. After completing the sensitive grade matching of all drug records one by one, organize the results into a sensitive grade score set indexed by drug name, and each record indicates the drug name, action region, matching grade name, and corresponding sensitive score.

[0030] The side effect matching sub-module calls the sensitive grade score, analyzes the side effect information reported by the patient, extracts the symptom keywords, cross-searches with the drug name list, judges whether the side effect manifestation appears in the action region recorded in the drug instruction manual, and records the occurrence times and matching degree to obtain the side effect manifestation matching rate; Read the text information of side effect feedback spontaneously reported by patients within a certain period of time, perform word segmentation and stop word removal operations on the text, extract high-frequency keywords such as typical symptom words like "headache", "nausea", "rash", "diarrhea", etc., and convert them into an indexable format to construct a list of side effect keywords. Subsequently, call the current drug list, obtain the instruction manual information of each drug from the database, focus on extracting the paragraphs describing the common side effects and their occurrence sites, and establish a dictionary structure of the description area according to the drug names. Then, conduct a double cross-search on the extracted side effect keywords and the content of each drug's description area. The first round of search is to directly match whether the keyword appears in the instruction manual, and the second round is for fuzzy matching or synonym search. For example, "drowsiness" can be regarded as a match with "somnolence". In the matching process, each successful hit is counted as one occurrence, and the matching times of each keyword and each drug will be accumulated. The matching degree is calculated as the ratio of the number of hits to the total number of words in the side effects described in the drug instruction manual. If both "headache" and "insomnia" in the patient feedback terms appear in the side effect paragraph of drug A's instruction manual, and there are a total of 5 side effect items for drug A, then the matching degree is 2 / 5 = 0.4. Perform this operation for each drug respectively, record the corresponding matching degree percentage and classify and output it. Display the sensitive level score and the side effect keyword matching degree side by side. The output structure includes the patient ID, drug name, sensitive score, number of matching keywords, and matching rate percentage to obtain the side effect manifestation matching rate.

[0031] The overlapping risk assessment sub-module identifies the overlapping risks in the drug action areas and symptom combinations based on the side effect manifestation matching rate, using the formula: ; Calculate the cross-action risk value to evaluate the risk level of overlapping side effects between different drugs, compare it with the risk benchmark, identify the risk combinations exceeding the benchmark, and obtain the analysis result of overlapping side effects. Among them, is the sensitive level score of the th drug, determined according to the sensitivity identification information of the drug, and measures the sensitivity of the side effects produced by the drug on the patient. is the side effect manifestation matching rate of the th drug, indicating the matching degree of the side effects caused by this drug and the known drug action areas. is the symptom coverage of the th drug, measuring the coverage range of the symptoms reported by the patient and the side effects caused by this drug. is the number of overlaps of the drug action areas of the th drug, calculating the number of overlaps that occur between different drug action areas, reflecting the risk of interaction or enhanced side effects between different drugs. is the total risk factor, which is the sum of the risk assessments of all drug combinations. is the number of drug combinations; Call the side effect manifestation matching rate obtained from the patient's self-reported side effect information, and the sensitivity level score obtained from the previous identification, and correspond them one by one with the drug action area information. Extract the sensitivity score, side effect matching degree and related symptom coverage range of each group of drugs in the cross item item by item. At the same time, calculate the number of overlapping drug action areas for each group of drugs. For example, for drugs A and B, the sensitivity level scores are 0.85 and 0.75 respectively, the side effect matching rates are 0.80 and 0.60 respectively, the symptom coverage degrees are 0.50 and 0.70 respectively, and the number of overlapping action areas are 2 and 3. Calculate the cross risk item corresponding to each combination item. After substituting specific parameters, the calculation process for three groups of drug combinations is as follows: For the first group (g = 1): , , , ; ; The total value is ; For the second group (g = 2): , , , ; ; The total value is ; For the third group (g = 3): , , , ; ; The total value is ; Sum the above three items. The numerator is: Let the total risk factor Then the cross-action risk value is calculated as follows: ; If this value exceeds the overlapping risk benchmark value of 0.65, it means that the selected drug combination has an overlapping side effect combination with a relatively high risk level, which should be included in the overlapping side effect analysis results and submitted for identification by the clinical warning system or further drug screening process.

[0032] Please refer to Figure 5 , the health risk warning module includes: Based on the results of the overlapping side effect analysis and combined with the patient's real-time heart rate, blood pressure and body temperature data, the abnormal peak extraction sub-module detects the fluctuation trends of each type of parameter within the analysis period respectively, extracts the abnormal peak points that meet the physiological extreme value criteria in the fluctuation curve, and screens out the interference fluctuations whose duration does not meet the peak duration benchmark, obtaining the set of abnormal physiological peaks; First, determine the boundaries of the analysis period, which is usually set to 24 hours, and the start and end times are determined by the latest timestamp in the physiological fluctuation record. Subsequently, call the real-time acquisition data sequences of the patient's heart rate, systolic blood pressure and body temperature within this period, and construct continuous fluctuation curves for these three types of data respectively. Each curve forms a two-dimensional sequence according to the time horizontal axis and the corresponding physiological value vertical axis. Taking each sampling point as a node, compare the numerical height of the point within a certain number of time units before and after respectively. If the current point value is higher than the two adjacent sampling points and the numerical change amplitude exceeds the physiological extreme value benchmark, this point is initially identified as an abnormal peak point. The physiological extreme value benchmark is configured through default settings or expert-set values. For example, the heart rate benchmark is 120 beats per minute, the systolic blood pressure benchmark is 160 mmHg, and the body temperature benchmark is 38.0 °C. If it exceeds this threshold and forms a local maximum value, it is marked as a candidate abnormal peak. Subsequently, enter the interference fluctuation screening link, analyze the duration of each abnormal point one by one. The duration is calculated as the total length of the time slices continuously exceeding the threshold. If an abnormal point only lasts for 3 seconds in the upper and lower time segments and does not reach the set minimum duration benchmark time of 5 seconds, it is determined as an interference fluctuation and excluded. If it lasts for 5 seconds or more, this point is retained as a formal abnormal peak. For example, a patient's body temperature continuously records values of 38.3 °C, 38.4 °C, 38.2 °C, 38.5 °C, 38.6 °C, 38.1 °C from 09:00:10 to 09:00:16, that is, record 09:00:13 as the peak point. Since the duration of this fluctuation is 7 seconds and meets the benchmark, finally, all the compliant abnormal peak points retained through judgment are uniformly sorted into the set of abnormal physiological peaks and marked and recorded according to the parameter types.

[0033] Based on the set of abnormal physiological peaks, the coincidence risk calculation sub-module, according to the distribution on the time axis, uses the formula: ; Obtain the total overlap risk coefficient , which is a comprehensive risk indicator used to measure the peak overlap situation of all the patient's monitoring parameters, and divide the corresponding warning levels according to its interval to obtain the risk level interval. Among them, represents the number of parameter overlaps within the th overlap interval, reflecting the number of abnormal peaks of different physiological parameters appearing within this overlap interval, represents the time length of the th overlap interval, which refers to the total duration covered by this interval, represents the The difference in peak intensity within an overlapping interval is used to measure the variation degree of the peaks of each parameter within this interval. represents the parameter influence coefficient of the th overlapping interval. The coefficient is weighted according to the physiological influence importance of the parameter. represents the total number of overlapping intervals within the current analysis period; First, it is necessary to identify the start and end times of each peak. A periodic window with a time granularity of one minute is used to remap the abnormal peak points in all heart rate, blood pressure, and body temperature data to a unified time coordinate system. Within each periodic window, if it is analyzed that there are peak values of two or more physiological parameters whose time intervals are less than 1 minute and overlap, then they are marked as an overlapping interval, and the number of occurrences of the peaks of each parameter within this interval is calculated as the parameter overlap quantity. For example, if both the heart rate and blood pressure have peaks at 20 seconds and 35 seconds within a certain minute, then the overlap quantity here is 2. If the body temperature also has a peak during this time period, then the overlap quantity is 3. Define this overlap quantity as ; After confirming the overlapping interval, record the start and end times of each interval. The difference between the two is the duration of this interval, denoted as , for example, if the start time of the first overlapping interval is 10:02:05 and the end time is 10:02:11, then the duration of this interval is 6 seconds; Next, it is necessary to calculate the difference between the peak intensities within each overlapping interval. The peak value of each of the heart rate, blood pressure, and body temperature is respectively subjected to intensity normalization processing to remove the unit influence. Then, within the overlapping interval, the normalized values are respectively taken, and the difference between the maximum value and the minimum value is calculated and squared to obtain the peak intensity difference, defined as , for example, if the three normalized peak values within this interval are 1.2, 0.9, and 1.5 respectively, then the maximum difference is 1.5 - 0.9 = 0.6, the square is 0.36, and then multiplied by the peak number normalization value 1.5 within the interval to obtain ; Then, assign an influence level to each parameter. Set the level according to its contribution to the physiological risk. The heart rate is 1.2, the blood pressure is 1.0, and the body temperature is 0.9. The average of the influence levels of all parameters participating in the overlapping interval is used as the parameter influence coefficient of this interval , for example, if there is an overlap between the heart rate and blood pressure in this interval, then , substitute the obtained , , , into the formula. Suppose there are three overlapping intervals, and their parameters are as follows: The 1st interval: , , , ; The 2nd interval: , , , ; Interval 3: , , , ; Perform itemized calculations: ; ; ; Add the three together to obtain the total overlap risk coefficient: ; The obtained total overlap risk coefficient is 3.117. If the preset risk level benchmark intervals are: low-risk interval [0–1.5], medium-risk interval [1.5–2.5], high-risk interval [2.5–4.0], then this value is in the high-risk level interval, indicating that there is a phenomenon of multiple overlapping peaks in the patient during the current monitoring period. This result will be used as the classification basis for subsequent risk level intervals.

[0034] The risk fluctuation marking sub-module filters the fluctuation sections that maintain the same or an increased level interval for multiple consecutive cycles according to the risk level interval, determines whether the level interval fluctuation exceeds the risk priority marking benchmark, and marks the abnormal fluctuations as high-priority monitoring objects to obtain the risk fluctuation monitoring data; First, read the abnormal peak data within multiple analysis cycles. Taking every 24 hours as a cycle, match the peak value of each type of physiological parameter within each cycle with the risk classification benchmark. The set risk levels are divided into five levels: Level 1 is the safe range, Level 2 is the attention range, Level 3 is the warning range, Level 4 is the high warning range, and Level 5 is the extremely high-risk range. The classification criteria are set according to the parameter type. For example, for heart rate, Level 3 is 110 - 120 beats per minute, Level 4 is 121 - 130 beats per minute, and Level 5 is over 130 beats per minute. Analyze the level to which each peak record belongs one by one, and count whether the peak level of each type of parameter does not decrease or shows a continuous upward trend within 3 or more consecutive cycles. For example, if the highest heart rate peaks recorded for a certain patient in three consecutive days are 118, 123, and 131 beats per minute respectively, then it is judged that they are in Level 3, Level 4, and Level 5 respectively, forming an increasing trend. Subsequently, call the risk priority marking benchmark, which is set to trigger a high-priority monitoring mark when the fluctuation level remains stable at Level 4 or Level 5 for three consecutive cycles. If it is judged that the above increasing sequence meets this benchmark, then mark the entire fluctuation paragraph from the starting cycle to the current cycle as a high-priority monitoring section. If the fluctuation level of a certain section is at Level 5 in two cycles and drops to Level 3 in the third cycle, it does not meet the benchmark and is not marked. Record the start time, end time, physiological parameter type, and the corresponding level sequence for all qualified wave bands respectively to generate risk fluctuation monitoring data.

[0035] Please refer to Figure 6 , the intelligent consultation optimization module includes: The keyword extraction sub-module calls the risk fluctuation monitoring data, analyzes the interaction history of the patient participating in the intelligent consultation interface, obtains the interactive statement text data, extracts the keyword groups in the consultation statements and records their occurrence frequencies, then sorts the keywords according to the frequency, and establishes a high-frequency keyword set; Obtain the current fluctuation cycles marked as high priority for the patient and the corresponding physiological parameter labels, then retrieve all the statement interaction records of the patient using the intelligent interrogation system during the corresponding cycles. Compare the timestamp of each statement with the fluctuation monitoring time, and only retain the statements whose interaction occurrence time is within the fluctuation period as the analysis objects. Then, perform language tokenization on each statement, identify the complete phrases and remove the non-content words to construct a list of lemmas. By counting the total number of times each lemma appears in all statements, generate a keyword frequency comparison table. Subsequently, sort all the lemmas in descending order of their appearance frequencies and remove the keywords whose appearance times are lower than the set threshold, usually set to 2 times. That is, if the keyword "chest tightness" appears 3 times in 5 statements, it is retained, and if "sweating" appears only once, it is removed. Then, aggregate and count all the retained keywords according to their semantic categories. For example, "nausea" and "retching" are classified into the digestive system reaction category, record the aggregation frequency and sort again, and finally form a high-frequency keyword set with clear frequency sorting and clear category attribution. Each keyword in the set includes its appearance times, the index of the source statement, and the aggregation category label.

[0036] Based on the high-frequency keyword set, the matching degree calculation sub-module combines the patient's current health data to calculate the matching degree between the keywords and the patient's condition, performs keyword screening, removes the keywords that do not meet the conditions, and obtains the keyword screening weights. Extract the latest physiological parameter records, drug usage information, and side effect feedback content from the patient's current health data. Perform a status comparison process for each high-frequency keyword to determine whether the word can find a direct or indirect corresponding item in the current health data. The judgment method is whether the keyword content is literally consistent or logically corresponding to the current abnormal item of the indicator. For example, if the keyword is "headache" and the patient lists "persistent temporal discomfort" in the side effect report, it is determined that the match is established. During the confirmation of the match, set a matching weight for each keyword. Set the basic weight to 10. If the keyword directly appears in the symptom field of the physiological data, the weight is assigned as 10. If it indirectly appears in the drug side effect entries, the weight is assigned as 7. If it is only semantically related to the action area of the drug name, the weight is assigned as 5. When the keyword has no matching item in the current health data, the weight is assigned as 0 and it is recorded as an object to be removed. After completing the status comparison and weight calculation of all keywords, sort out the keyword screening list. The list records the original frequency, matching situation, and the screened weight assigned to each keyword. For example, the original frequency of "palpitation" is 4 times, it matches the abnormal heart rate fluctuation data, and the weight is assigned as 10. The frequency of "nausea" is 2 times, and there is no drug or side effect related item in the current record, so the weight is assigned as 0. Finally, remove the items with a weight of 0 and retain the remaining keywords to form the keyword screening weights.

[0037] The path rearrangement submodule re-evaluates the relevance to each node in the consultation path according to the keyword screening weight, adjusts the weight of the original path, re-orders the consultation path, and updates the key interaction points of the consultation to obtain the optimal configuration of the consultation path; The node structure diagram of the current consultation path is read. The structure diagram is a directed path composed of interactive nodes. Each node represents a consultation question and its branch options. Each filtered keyword is semantically compared with the content description of the path node. The screening weight and the node content relevance score are multiplied to obtain the new weight score of the path node. For example, if the node content is "Have you ever had chest tightness symptoms?", the screening weight of the keyword "chest tightness" is 10, then the weight score of the node is 10. If another node is "Have you ever had nausea or vomiting?", the keyword "nausea" weight is 0, then the score is 0. After completing the above operations for all nodes, the original path is reordered according to the node score, and the nodes with higher scores are displayed first. If the original path is node A→B→C, it is adjusted to C→A→B. The new path order is recorded, and the key interaction point number of each node is synchronously updated. For low-weight nodes in some continuous paths, jump connections are set to high-weight nodes or new keyword-related nodes are added, and the updated structure is reassembled into a consultation path with enhanced keyword relevance.

[0038] An intelligent follow-up method for daytime tumor patients comprises the following steps: S1: Based on the wearable device, the patient's heart rate, systolic blood pressure and body temperature data are obtained, and the abnormal floating standard interval is set for each monitoring parameter. The real-time collected physiological data is continuously compared with the standard interval, and the continuous deviation from the normal range is recorded. The deviation period is archived as a key information segment to obtain a record of physiological fluctuations; S2: Based on physiological fluctuation records, obtain the patient's daily drug intake information, analyze the newly added or adjusted drug combinations within the time window, retrieve the drug database, compare the synergistic effects and reaction types between drugs, identify drug combinations associated with serious health interference, and obtain drug sensitivity identification information; S3: Analyze the side effects self-reported by patients based on the drug sensitivity identification information, compare the drug name list, map the side effects content with the drug action area, identify the drug and symptom combinations with the risk of cross-action, and obtain the overlapping side effects analysis results; S4: Based on the results of overlapping side effect analysis, combined with the patient's real-time heart rate, blood pressure and body temperature data, analyze the overlapping areas of abnormal peaks within the cycle, evaluate the health risks brought by the overlap, mark the continuous abnormal fluctuations as high-priority monitoring objects, and obtain risk fluctuation monitoring data; S5: Call the risk fluctuation monitoring data, analyze the interaction history of the patient participating in the intelligent consultation interface, extract the frequently occurring consultation keywords in the interaction history, and reorder the consultation path according to the matching degree between the keywords and the patient's condition to obtain the optimized configuration of the consultation path.

[0039] A computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the intelligent follow-up system for daytime cancer patients.

[0040] A computer device, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, it implements the intelligent follow-up system for daytime cancer patients.

[0041] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent follow-up system for daytime cancer patients, characterized in that: The system comprises: The physiological monitoring module is based on wearable devices to obtain the patient's heart rate, systolic blood pressure and body temperature data, and record the continuous deviation of the data from the normal range in any time period to obtain a record of physiological fluctuations; The drug interaction analysis module obtains the patient's daily drug intake information based on the physiological fluctuation record, compares the synergistic effects and reaction types between drugs, identifies drug combinations associated with serious health interference, and obtains drug sensitivity identification information; The side effect mapping module analyzes the side effect information self-reported by the patient according to the drug sensitivity identification information, compares it with the drug name list, identifies the drug and symptom combination with the risk of cross-action, and obtains the overlapping side effect analysis result; The health risk warning module analyzes the overlapping areas of abnormal peaks within the cycle based on the overlapping side effect analysis results, combined with the patient's real-time heart rate, blood pressure and body temperature data, evaluates the health risks brought by the overlap, and obtains risk fluctuation monitoring data; The intelligent consultation optimization module calls the risk fluctuation monitoring data, analyzes the patient's interaction history in the intelligent consultation interface, reorders the consultation path according to the degree of match between the keywords and the patient's condition, and obtains the consultation path optimization configuration.

2. The intelligent follow-up system for daytime cancer patients according to claim 1, characterized in that: The physiological fluctuation record includes the fluctuation amplitude, duration, and key fluctuation period; the drug sensitivity identification information includes the sensitive drug list, drug interaction intensity, and health impact rating; the overlapping side effect analysis results are specifically side effect severity, drug side effect correlation, and drug cross-risk index; the risk fluctuation monitoring data includes peak overlapping frequency, abnormal fluctuation duration period, and risk warning level; the consultation path optimization configuration includes optimized consultation frequency and keyword priority adjustment.

3. The intelligent follow-up system for daytime cancer patients according to claim 1, characterized in that: The physiological monitoring module comprises: The signal acquisition submodule is based on a wearable device, which records the patient's heart rate, systolic blood pressure and body temperature data in real time, performs time serialization processing on each monitoring parameter, and synchronously records the acquisition timestamp of each parameter to obtain a continuous physiological data sequence; The physiological parameter judgment submodule sets an abnormal floating standard interval for each physiological monitoring parameter based on the continuous physiological data sequence, performs data comparison, identifies data points that exceed the standard interval, and classifies continuous abnormal data points by time interval to obtain an abnormal data identification set; The physiological fluctuation archiving submodule calls the abnormal data identification set, analyzes the abnormal data time period, records the start and end time of each abnormal data, archives the deviation time period as a key information segment, and obtains the physiological fluctuation record.

4. The intelligent follow-up system for daytime cancer patients according to claim 1, characterized in that: The drug interaction analysis module includes: The drug information extraction submodule obtains the patient's daily drug intake information based on the physiological fluctuation record, identifies drug changes within a time window, analyzes the type and dosage changes of each drug, and obtains drug combination change data for each time window; The drug synergistic reaction comparison submodule accesses the drug database based on the drug combination change data, compares the drug action mechanism and synergistic effect attributes, classifies and matches the functional categories and metabolic pathways of the drugs, and forms synergistic reaction data for each drug combination; The interference combination identification submodule screens high-risk drug combinations according to the synergistic reaction data, analyzes the interaction intensity between the action type of the drug combination and the physiological fluctuation, calculates the risk interaction data, determines the risk level of the drug combination, and obtains drug sensitivity identification information.

5. The intelligent follow-up system for daytime cancer patients according to claim 1, characterized in that: The side effect mapping module includes: The sensitivity identification submodule extracts data from the drug name, action area and reaction level based on the drug sensitivity identification information, determines the drug sensitivity level by comparing it with the sensitivity grading benchmark, and obtains the sensitivity level score; The side effect matching submodule calls the sensitivity level score, analyzes the side effect information reported by the patient, extracts symptom keywords, cross-searches with the drug name list, determines whether the side effect manifestation appears in the action area recorded in the drug instructions, and records the number of occurrences and the matching degree to obtain the side effect manifestation matching rate; The overlapping risk assessment submodule identifies the overlapping risks in the drug action area and symptom combination according to the side effect manifestation matching rate, calculates the cross-action risk value, compares it with the risk benchmark, identifies the risk combination that exceeds the benchmark, and obtains the overlapping side effect analysis results.

6. The intelligent follow-up system for daytime cancer patients according to claim 1, characterized in that: The health risk early warning module includes: The abnormal peak extraction submodule is based on the overlapping side effect analysis results and combines the patient's real-time heart rate, blood pressure and body temperature data to detect the fluctuation trend of each type of parameter within the analysis period, extract the abnormal peak points that meet the physiological extreme value standard in the fluctuation curve, and screen out the interference fluctuations whose duration does not meet the peak duration benchmark to obtain the abnormal physiological peak set; The overlap risk calculation submodule obtains the total overlap risk coefficient based on the abnormal physiological peak set and the distribution on the time axis, and divides the corresponding warning level according to its interval to obtain the risk level interval; The risk fluctuation marking submodule screens the fluctuation segments that remain in the same or increased level range for multiple consecutive periods according to the risk level range, determines whether the level range fluctuation exceeds the risk priority marking benchmark, and marks the abnormal fluctuation as a high-priority monitoring object to obtain risk fluctuation monitoring data.

7. The intelligent follow-up system for daytime cancer patients according to claim 1, characterized in that: The intelligent consultation optimization module includes: The keyword extraction submodule calls the risk fluctuation monitoring data, analyzes the patient's interaction history in the intelligent consultation interface, obtains the interactive sentence text data, extracts the keyword groups in the consultation sentences and records their occurrence frequencies, and then sorts the keywords according to the frequencies to establish a high-frequency keyword set; The matching degree calculation submodule calculates the matching degree between the keywords and the patient's condition based on the high-frequency keyword set and the patient's current health data, performs keyword screening, removes keywords that do not meet the conditions, and obtains keyword screening weights; The path rearrangement submodule re-evaluates the relevance to each node in the consultation path according to the keyword screening weights, adjusts the weights of the original path, re-orders the consultation path, and updates the key interaction points of the consultation to obtain an optimized configuration of the consultation path.

8. An intelligent follow-up method for daytime tumor patients, characterized in that: The method is used to implement the intelligent follow-up system for daytime tumor patients according to any one of claims 1 to 7, comprising the following steps: S1: Based on the wearable device, the patient's heart rate, systolic blood pressure and body temperature data are obtained, and the abnormal floating standard interval is set for each monitoring parameter. The real-time collected physiological data is continuously compared with the standard interval, and the continuous deviation from the normal range is recorded. The deviation period is archived as a key information segment to obtain a record of physiological fluctuations; S2: Based on the physiological fluctuation records, obtain the patient's daily drug intake information, analyze the newly added or adjusted drug combinations within the time window, retrieve the drug database, compare the synergistic effects and reaction types between drugs, identify drug combinations associated with serious health interference, and obtain drug sensitivity identification information; S3: Analyze the side effect information self-reported by the patient according to the drug sensitivity identification information, compare the drug name list, map the side effect content with the drug action area, identify the drug and symptom combination with the risk of cross-action, and obtain the overlapping side effect analysis results; S4: Based on the overlapping side effect analysis results, combined with the patient's real-time heart rate, blood pressure and body temperature data, analyze the overlapping areas of abnormal peaks within the cycle, evaluate the health risks brought by the overlap, mark the continuous abnormal fluctuations as high-priority monitoring objects, and obtain risk fluctuation monitoring data; S5: Call the risk fluctuation monitoring data, analyze the patient's interaction history in the intelligent consultation interface, extract the consultation keywords that appear frequently in the interaction history, reorder the consultation path according to the degree of match between the keywords and the patient's condition, and obtain the consultation path optimization configuration.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the intelligent follow-up system for daytime cancer patients according to any one of claims 1 to 7 is implemented.

10. A computer device comprising a memory and a processor, characterized in that: The memory stores a computer program, and when the processor executes the computer program, the intelligent follow-up system for daytime cancer patients according to any one of claims 1 to 7 is implemented.

Images