Civil recommendation algorithm according to matching degree

Abstract

The invention discloses a caregiver recommendation algorithm according to a matching degree, and relates to the technical field of nursing. The method comprises the following steps of patient information collection: collecting basic information of a patient, obtaining geographical location information of the patient, determining a city and a detailed address of the patient, guiding the patient to describe nursing requirements in detail, and inputting information of a caregiver through a user side information collection module; and the caregiver fills in own professional skill information which covers basic nursing skills, professional nursing specialists and skills in the special nursing field, and uploads related certificate qualification certificates. Through deep analysis and processing of the multi-dimensional demand information of the patient and the comprehensive information of the nursing staff, the intelligent matching model is used to accurately calculate the matching degree, the nursing staff highly meeting the demand of the patient can be screened for the patient, the pertinence and effectiveness of the nursing service are greatly improved, and the problems of poor nursing effect and the like caused by mismatching are avoided.

Description

Technical Field

[0001] The present invention relates to the field of nursing technology, and in particular to a caregiver recommendation algorithm based on matching degree. Background Art

[0002] With the aging of the population and the growing demand for medical care, the nursing service industry is facing huge challenges. The traditional caregiver dispatching system mainly relies on manual scheduling, which is not only inefficient but also prone to errors. In recent years, with the development of information technology, although some intelligent caregiver dispatching systems have begun to appear, for example, a caregiver intelligent dispatching system based on GIS (geographic information system), which can automatically match the nearest caregiver according to the patient's location information, but this system still has some problems. First, it does not take into account the skills and experience of the caregiver, which may lead to mismatches. Secondly, it does not achieve real-time updates. When the status of the caregiver changes, the system may not be able to adjust in time. Finally, it does not provide a user feedback mechanism, and users cannot evaluate the service quality. Therefore, the existing caregiver intelligent dispatching system needs further improvement and upgrading.

[0003] To this end, we provide a caregiver recommendation algorithm based on matching degree to solve the above problems. Summary of the invention

[0004] The purpose of the present invention is to provide a caregiver recommendation algorithm based on matching degree, which solves the problem that the skills and experience of caregivers in the prior art do not match customer needs and have no user feedback mechanism through the cooperation of an intelligent matching model and a real-time feedback system.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] The present invention is a caregiver recommendation algorithm based on matching degree, comprising the following steps:

[0007] Step a: Patient information collection: Through the user-side information collection module, collect the patient's basic information, obtain the patient's geographical location information, determine the city and detailed address, and guide the patient to describe the nursing needs in detail;

[0008] Step b: Caregiver information entry: With the help of the management end information entry module, the caregiver fills in his or her professional skills information, covering basic nursing skills, professional nursing expertise and skills in special nursing fields, and uploads relevant certificates and qualifications, records the caregiver's geographical location and service scope, and clarifies the serviceable area in the city and surrounding areas;

[0009] Step c: Data cleaning and conversion: The data preprocessing submodule in the matching algorithm core module starts the data cleaning unit to remove invalid characters, duplicate information, and incorrectly formatted data from the patient and caregiver information. The data conversion unit converts the text information (nursing skill description, nursing demand description) into a vector form that can be processed by the algorithm, and unifies the date and time formats for subsequent calculations;

[0010] Step d: Constructing a matching model: The weighted scoring unit of the matching model construction submodule sets weights for the factors of the professional skill matching, regional adaptability, time availability, and past service evaluation of the caregiver, and constructs a comprehensive scoring formula. The weights can be adjusted according to actual conditions. The collaborative filtering model unit in the machine learning model integration unit analyzes the matching patterns of similar patients and caregivers, and recommends caregivers who have been selected by similar patients and have good evaluations by calculating the similarity between patients (based on the similarity of nursing needs). The content recommendation model unit matches the patient demand description text and the caregiver skill specialty text based on text semantic analysis, and calculates the semantic similarity score. Other machine learning model (deep learning model) units can be used to process matching calculations in complex scenarios, and perform deep feature extraction and matching prediction based on multiple factors;

[0011] Step e: Comprehensive evaluation and ranking: The comprehensive evaluation submodule of the intelligent matching execution module starts the professional skills evaluation unit, and calculates the matching degree score between the professional skills of the caregiver and the patient's needs through text matching and keyword extraction technology. The geographical evaluation unit calculates the actual transportation time cost or distance score according to the geographical location of the patient and the caregiver. The time evaluation unit analyzes the overlap between the caregiver's available service time and the patient's required time to generate a time matching score. The evaluation evaluation unit refers to the caregiver's past service evaluation and converts data such as the praise rate into an evaluation score;

[0012] Step f: dispatch confirmation: the recommendation display submodule of the dispatch confirmation interaction module displays a list of recommended caregivers to the management personnel, including the caregiver's photo, name, professional skill highlights, comprehensive matching score, and service history data (number of past services, average rating);

[0013] Step g: The manager manually reviews and adjusts the recommendation results according to the actual situation (current work arrangements of the caregiver, special service requirements) through the manual review submodule, and finally determines the caregiver to be assigned;

[0014] Step h: Service implementation and data collection: During the service process, the caregiver uses the operation record submodule of the service implementation data collection module to record the time, content, and changes in the patient's physical indicators of key nursing operations. The emergency record submodule records the emergencies (patient's sudden illness, mood swings) and their treatment methods during the service process. The communication record submodule records the content and communication methods (face-to-face communication, telephone communication) between the caregiver and the patient and their family members.

[0015] Step i: Evaluation feedback integration: The score collection submodule of the evaluation feedback integration module collects patients' multi-dimensional scores on caregivers (professional skills, service attitude, and communication skills), and standardizes the score data. The text feedback collection submodule collects patients' text evaluations of caregivers and performs sentiment analysis (to determine whether the evaluation is positive or negative).

[0016] Step j: Machine learning model comparison and update: The model experiment submodule of the machine learning model comparison module runs multiple recommendation system strategies at the same time, including collaborative filtering model, content recommendation model, and deep learning model, and sets different parameter combinations for experiments. The effect evaluation submodule evaluates the performance of different models by calculating the accuracy (the overlap rate between the recommended caregivers and the caregivers that the patients are actually satisfied with), recall rate (the proportion of actually satisfied caregivers recommended), and patient satisfaction improvement.

[0017] Step k: Third-party data fusion and optimization: The data interface submodule of the third-party data fusion module establishes a secure and stable interface with different third-party health databases (hospital medical record system, professional testing agency database) to obtain patient health data. The data parsing submodule parses data in various formats from third-party databases (medical record files in XML format, test reports in CSV format) and extracts key information (disease diagnosis, treatment plan, test indicators). The data fusion submodule integrates third-party data with existing patient demand information in the system and matches caregivers with corresponding nursing experience based on the patient's disease diagnosis. The privacy and security submodule ensures compliance with data privacy regulations during the data fusion process, uses anonymization and data desensitization technology to process sensitive data and prevent data leakage. At the same time, based on the effect and compliance of data fusion, it optimizes the data fusion process and strategy to improve the effectiveness of the matching algorithm in using third-party data.

[0018] The present invention is further configured such that the nursing needs in step a include nursing types (life care, medical care, rehabilitation care), special nursing requirements (special needs caused by illness or religious beliefs, etc.) and preferences for caregivers (gender, age, work experience).

[0019] The present invention is further configured such that in step b, the caregiver enters information to set his or her own working time schedule, including the number of working days per week and the time period available for service each day, and writes a personal profile to introduce work experience and nursing philosophy.

[0020] The present invention is further configured such that the data standardization unit in step c unifies the dimensions of different types of data (caregiver scoring data and distance data) so that each data is comprehensively evaluated on the same scale.

[0021] The present invention is further configured such that in step d, the third-party data access unit establishes an interface with a third-party health database to obtain detailed health data of the patient (medical history, allergy history, recent physical examination indicators), and integrates these data into the matching model to refine the matching criteria.

[0022] The present invention is further configured such that the weight adjustment submodule in step e dynamically adjusts the weight of each evaluation unit according to different nursing scenarios (emergency care, long-term care) and special needs of patients to generate a final matching score, and the caregiver sorting submodule sorts the caregivers in descending order according to the matching scores, generates a recommendation list, and puts the caregivers with the highest matching degree in front.

[0023] The present invention is further configured such that the dispatch feedback submodule in step g feeds back the dispatch result (finally selected caregiver, dispatch time) to the intelligent matching module for algorithm learning and optimization.

[0024] The present invention is further configured such that the data uploading submodule in step h uploads the collected data to the data storage and management module in real time to ensure the timeliness and integrity of the data.

[0025] The present invention is further configured such that the data integration submodule in step i integrates the score and text feedback data, stores them in the data storage and management module, and transmits key feedback information (reasons for negative reviews) to the intelligent matching module so that the algorithm can adjust the matching strategy in time according to the patient's actual experience and improve the quality of future recommendations.

[0026] The present invention is further configured such that the model selection submodule in step j selects the most suitable model combination or parameter setting according to the effect evaluation result, and updates the model configuration of the matching algorithm to continuously optimize the accuracy and efficiency of the matching degree calculation.

[0027] The present invention has the following beneficial effects:

[0028] 1. The present invention, through in-depth analysis and processing of multi-dimensional demand information of patients and comprehensive information of caregivers, uses an intelligent matching model to accurately calculate the degree of matching, and can screen out caregivers who are highly consistent with the needs of patients, greatly improving the pertinence and effectiveness of nursing services, and avoiding problems such as poor nursing effects due to mismatch.

[0029] 2. Based on accurate matching results, the present invention can reasonably allocate caregiver resources, give full play to the professional skills and experience of caregivers, reduce idleness and waste of human resources, and improve the operational efficiency of the entire nursing service system.

[0030] 3. The present invention improves patient satisfaction: Due to the precise matching of caregivers with patient needs and subsequent continuous optimization based on user feedback, patients can receive more considerate and professional nursing services, and their satisfaction with the nursing services will be significantly improved.

[0031] 4. The present invention accumulates a large amount of patient demand data, caregiver service data and matching result data during operation. Through in-depth mining and analysis of these data, a scientific basis can be provided for nursing service decision-making. At the same time, the matching algorithm and service process are continuously optimized according to user feedback and data changes, so that the entire nursing service system can keep pace with the times and adapt to the ever-changing market demand and industry development. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for describing the embodiments are briefly introduced below.

[0033] Figure 1 A flowchart for an algorithm for recommending caregivers by degree of match;

[0034] Figure 2 System diagram of a caregiver recommendation algorithm based on matching degree. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the present invention will be described below in conjunction with the drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Specific embodiment 1

[0037] See also Figure 1-Figure 2 The present invention is a caregiver recommendation algorithm based on matching degree, and the caregiver recommendation algorithm based on matching degree comprises the following steps:

[0038] Step a: Patient information collection: Through the user-side information collection module, collect the patient's basic information, obtain the patient's geographic location information, determine the city and detailed address, and guide the patient to describe the nursing needs in detail. The nursing needs include the type of care (life care, medical care, rehabilitation care), special care requirements (special needs caused by illness or religious beliefs, etc.) and preferences for caregivers (gender, age, work experience);

[0039] Step b: Caregiver information entry: With the help of the management end information entry module, the caregiver fills in his or her own professional skills information, covering basic nursing skills, professional nursing expertise and skills in special nursing fields, and uploads relevant certificates and qualifications, records the caregiver's geographical location and service scope, and clarifies the serviceable area in the city and surrounding areas. The caregiver enters information to set his or her own working time schedule, including the number of working days per week and the service time period per day, and writes a personal profile to introduce work experience and nursing philosophy;

[0040] Step c: Data cleaning and conversion: The data preprocessing submodule in the matching algorithm core module starts the data cleaning unit to remove invalid characters, duplicate information and wrong format data in the patient and caregiver information. The data conversion unit converts the text information (nursing skill description, nursing demand description) into a vector form that can be processed by the algorithm, and unifies the date and time formats for subsequent calculations. The data standardization unit unifies the dimensions of different types of data (nursing staff score data and distance data) so that each data can be comprehensively evaluated on the same scale.

[0041] Step d: Construct a matching model: The weighted scoring unit of the matching model construction submodule sets weights for the factors of the professional skill matching, regional adaptability, time availability, and past service evaluation of the caregiver, and constructs a comprehensive scoring formula. The weights can be adjusted according to actual conditions. The collaborative filtering model unit in the machine learning model integration unit analyzes the matching pattern between similar patients and caregivers, and recommends caregivers who have been selected by similar patients and have good evaluations by calculating the similarity between patients (based on the similarity of nursing needs). The content recommendation model unit matches the patient demand description text and the caregiver skill specialty text based on text semantic analysis, and calculates the semantic similarity score. Other machine learning model (deep learning model) units can be used to process matching calculations in complex scenarios, and perform deep feature extraction and matching prediction based on multiple factors. The third-party data access unit establishes an interface with a third-party health database to obtain detailed health data of patients (medical history, allergy history, recent physical examination indicators), and integrates these data into the matching model to refine the matching criteria;

[0042] Step e: Comprehensive evaluation and sorting: The comprehensive evaluation submodule of the intelligent matching execution module starts the professional skill evaluation unit, and calculates the matching degree score between the professional skills of the caregiver and the patient's needs through text matching and keyword extraction technology. The regional evaluation unit calculates the actual transportation time cost or distance score according to the geographical location of the patient and the caregiver. The time evaluation unit analyzes the overlap between the caregiver's service time and the patient's demand time to generate a time matching score. The evaluation evaluation unit refers to the caregiver's past service evaluation and converts data such as the praise rate into an evaluation score. The weight adjustment submodule dynamically adjusts the weight of each evaluation unit according to different nursing scenarios (emergency care, long-term care) and special needs of patients to generate the final matching score. The caregiver sorting submodule sorts the caregivers in descending order according to the matching score, generates a recommendation list, and puts the caregiver with the highest matching degree in the front.

[0043] Step f: dispatch confirmation: the recommendation display submodule of the dispatch confirmation interaction module displays a list of recommended caregivers to the management personnel, including the caregiver's photo, name, professional skill highlights, comprehensive matching score, and service history data (number of past services, average rating);

[0044] Step g: The manager manually reviews and adjusts the recommendation results according to the actual situation (current work schedule of the caregiver, special service requirements) through the manual review submodule, and finally determines the caregiver to be assigned. The dispatch feedback submodule feeds back the dispatch result (finally selected caregiver, dispatch time) to the intelligent matching module for algorithm learning and optimization;

[0045] Step h: Service implementation and data collection: During the service process, the caregiver uses the operation record submodule of the service implementation data collection module to record the time, content, and changes in the patient's physical indicators of key nursing operations. The emergency record submodule records the emergencies (sudden illness of the patient, emotional fluctuations) and their treatment methods during the service process. The communication record submodule records the content and communication methods (face-to-face communication, telephone communication) of the caregiver's communication with the patient and his family. The data upload submodule uploads the collected data to the data storage and management module in real time to ensure the timeliness and integrity of the data.

[0046] Step i: Evaluation feedback integration: The score collection submodule of the evaluation feedback integration module collects patients' multi-dimensional scores on caregivers (professional skills, service attitude, and communication skills), and standardizes the score data. The text feedback collection submodule collects patients' text evaluations of caregivers and performs sentiment analysis (determines whether the evaluation is positive or negative). The data integration submodule integrates the score and text feedback data, stores them in the data storage and management module, and passes key feedback information (reasons for negative reviews) to the intelligent matching module so that the algorithm can adjust the matching strategy in time according to the patient's actual experience and improve the quality of future recommendations.

[0047] Step j: Machine learning model comparison and update: The model experiment submodule of the machine learning model comparison module runs multiple recommendation system strategies at the same time, including collaborative filtering model, content recommendation model, and deep learning model, and sets different parameter combinations for experiments. The effect evaluation submodule evaluates the performance of different models by calculating the accuracy (the overlap rate between the recommended caregivers and the caregivers that the patients are actually satisfied with), recall rate (the proportion of recommended caregivers who are actually satisfied with the patients), and patient satisfaction improvement. The model selection submodule selects the most suitable model combination or parameter setting based on the effect evaluation results, and updates the model configuration of the matching algorithm to continuously optimize the accuracy and efficiency of the matching degree calculation;

[0048] Step k: Third-party data fusion and optimization: The data interface submodule of the third-party data fusion module establishes a secure and stable interface with different third-party health databases (hospital medical record system, professional testing agency database) to obtain patient health data. The data parsing submodule parses data in various formats from third-party databases (medical record files in XML format, test reports in CSV format) and extracts key information (disease diagnosis, treatment plan, test indicators). The data fusion submodule integrates third-party data with existing patient demand information in the system and matches caregivers with corresponding nursing experience based on the patient's disease diagnosis. The privacy and security submodule ensures compliance with data privacy regulations during the data fusion process, uses anonymization and data desensitization technology to process sensitive data and prevent data leakage. At the same time, based on the effect and compliance of data fusion, it optimizes the data fusion process and strategy to improve the effectiveness of the matching algorithm in using third-party data. Specific embodiment 2

[0050] General home care needs matching

[0051] Data collection and preprocessing phase

[0052] Patient information collection: A family member of a 70-year-old man living in Xuhui District, Shanghai, inputs information through the user-side information collection module. The basic information of the elderly is entered, and the family member indicates that the elderly needs life care due to mobility difficulties, such as assistance with daily living, simple rehabilitation exercises, etc., and prefers female caregivers between 40 and 50 years old, who are patient and have certain elderly care experience.

[0053] Caregiver information entry: Caregiver Xiao Li fills in her own information in the management end information entry module. She has 5 years of experience in elderly care, is good at daily care and basic rehabilitation guidance, holds a professional certificate in elderly care, lives in Minhang District, Shanghai, can provide full-day service from Monday to Friday, and states in her personal profile that she is good at communicating with the elderly.

[0054] Data cleaning and conversion: The data preprocessing submodule in the matching algorithm core module processes the information of both parties. It removes unnecessary punctuation marks in the information input by the family members, converts the description of the elderly care needs into a vector form that the algorithm can recognize, and standardizes the data such as Xiao Li’s work experience time and unifies the dimensions for subsequent calculations.

[0055] Matching calculation phase

[0056] Constructing a matching model: The weighted scoring unit of the matching model construction submodule sets the professional skills matching weight to 0.4, the regional adaptability weight to 0.2, the time availability weight to 0.2, and the past service evaluation weight to 0.2. Xiao Li's professional skills in elderly care are highly matched with the needs of the elderly, scoring 0.35 points (out of 0.4); geographically, Xuhui District and Minhang District are moderately far apart, with convenient transportation, and a regional adaptability score of 0.15 points; time fully meets the needs, scoring 0.2 points; past service evaluation is good, scoring 0.18 points. The comprehensive matching score is 0.88 points.

[0057] Comprehensive evaluation and ranking: The intelligent matching execution module ranks the caregivers according to the above calculation results. Xiao Li ranks high among many candidate caregivers and is included in the recommendation list.

[0058] Order dispatch and feedback stage

[0059] Dispatch confirmation: The dispatch confirmation interaction module displays the recommended list to the manager, who reviews and determines to dispatch the order to Xiao Li. The dispatch feedback submodule feeds back the result to the intelligent matching module.

[0060] Service implementation and data collection: During the service process, Xiao Li uses the operation record submodule to record the time and situation of helping the elderly get up and take a walk every day, uses the emergency record submodule to record the elderly's emotional fluctuations and comfort methods, and uses the communication record submodule to record the content of communication with the elderly and their families. The data upload submodule uploads this data to the data storage and management module in real time.

[0061] Evaluation feedback integration: After the service, the family members give Xiao Li a score of 9 (out of 10) for professional skills, 10 for service attitude, and 9 for communication skills through the score collection submodule, and write "Xiao Li's care is very careful and the elderly are very satisfied" in the text feedback collection submodule. The data integration submodule stores this data and passes key information to the intelligent matching module for algorithm optimization. Specific embodiment three

[0063] Postoperative rehabilitation care needs matching

[0064] Data collection and preprocessing phase

[0065] Patient information collection: Mr. Zhang, a patient who had just undergone heart bypass surgery in a hospital in Chaoyang District, Beijing, inputs information through the user-side information collection module. Mr. Zhang is 55 years old and needs postoperative rehabilitation care, including wound care, cardiac rehabilitation training guidance and other professional medical care services. He hopes that the caregiver has experience in postoperative heart care, preferably male, aged between 30 and 40, and can provide full-time service within one month after surgery.

[0066] Caregiver information entry: Caregiver Xiao Wang fills in information in the management end information entry module. He has 3 years of experience in cardiac rehabilitation nursing, is proficient in cardiac postoperative nursing procedures and rehabilitation training methods, holds a professional nursing certificate, lives in Dongcheng District, Beijing, and can provide full-time services within the time required by Mr. Zhang. His personal profile emphasizes his rigorous and serious work attitude.

[0067] Data cleaning and conversion: The data preprocessing submodule of the matching algorithm core module cleans and converts the information. It removes typos in Mr. Zhang’s demand information, converts cardiac rehabilitation nursing needs into algorithm vectors, and standardizes Xiao Wang’s work experience data.

[0068] Matching calculation phase

[0069] Constructing a matching model: The weighted scoring unit of the matching model construction submodule takes into account the professional requirements of postoperative rehabilitation nursing, setting the weight of professional skills matching to 0.5, the weight of regional adaptability to 0.1, the weight of time availability to 0.2, and the weight of past service evaluation to 0.2. Xiao Wang's professional skills in cardiac rehabilitation nursing highly match Mr. Zhang's needs, scoring 0.45 points; geographically, Chaoyang District and Dongcheng District are close, with a regional adaptability score of 0.08 points; time meets the requirements, scoring 0.2 points; past service evaluation is good, scoring 0.18 points. The comprehensive matching score is 0.91 points.

[0070] Comprehensive evaluation and ranking: The intelligent matching execution module ranks the caregivers according to their matching degree, and Xiao Wang ranks at the top and is included in the recommended list.

[0071] Order dispatch and feedback stage

[0072] Dispatch confirmation: The dispatch confirmation interaction module shows the recommended results to the manager, who confirms and dispatches the order to Xiao Wang. The dispatch feedback submodule feeds back the dispatch status to the intelligent matching module.

[0073] Service implementation and data collection: During Xiao Wang's service, the operation record submodule records each wound care and cardiac rehabilitation training data in detail, the emergency record submodule records Mr. Zhang's minor discomfort and treatment measures, and the communication record submodule records the communication with Mr. Zhang and his family about the rehabilitation plan. The data upload submodule uploads the data in a timely manner.

[0074] Evaluation feedback integration: After one month of service, Mr. Zhang gave Xiao Wang a score of 9.5 for professional skills, 9 for service attitude, and 8.5 for communication skills. He also made some suggestions on adjusting the intensity of rehabilitation training in the text feedback. The data integration submodule stores the data and passes the suggestions to the intelligent matching module for subsequent algorithm optimization to better match the postoperative rehabilitation care needs.

[0075] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that technicians in the relevant technical field can better understand and utilize the present invention.

Claims

1. A caregiver recommendation algorithm based on matching degree, characterized by: The following steps are involved: Step a: Patient information collection: Through the user-side information collection module, collect the patient's basic information, obtain the patient's geographical location information, determine the city and detailed address, and guide the patient to describe the nursing needs in detail; Step b: Caregiver information entry: With the help of the management end information entry module, the caregiver fills in his or her professional skills information, covering basic nursing skills, professional nursing expertise and skills in special nursing fields, and uploads relevant certificates and qualifications, records the caregiver's geographical location and service scope, and clarifies the serviceable area in the city and surrounding areas; Step c: Data cleaning and conversion: The data preprocessing submodule in the matching algorithm core module starts the data cleaning unit to remove invalid characters, duplicate information, and incorrectly formatted data from the patient and caregiver information. The data conversion unit converts the text information into a vector form that can be processed by the algorithm, and unifies the date and time formats for subsequent calculations. Step d: Constructing a matching model: The weighted scoring unit of the matching model construction submodule sets weights for the factors of the professional skill matching, regional adaptability, time availability, and past service evaluation of the caregiver, and constructs a comprehensive scoring formula. The weights can be adjusted according to actual conditions. The collaborative filtering model unit in the machine learning model integration unit analyzes the matching patterns of similar patients and caregivers, and recommends caregivers who have been selected by similar patients and have good evaluations by calculating the similarity between patients. The content recommendation model unit matches the patient demand description text and the caregiver skill specialty text based on text semantic analysis, and calculates the semantic similarity score. Other machine learning model units can be used to process matching calculations in complex scenarios, and perform deep feature extraction and matching prediction based on multiple factors; Step e: Comprehensive evaluation and ranking: The comprehensive evaluation submodule of the intelligent matching execution module starts the professional skills evaluation unit, and calculates the matching degree score between the professional skills of the caregiver and the patient's needs through text matching and keyword extraction technology. The geographical evaluation unit calculates the actual transportation time cost or distance score according to the geographical location of the patient and the caregiver. The time evaluation unit analyzes the overlap between the caregiver's available service time and the patient's required time to generate a time matching score. The evaluation evaluation unit refers to the caregiver's past service evaluation and converts data such as the praise rate into an evaluation score; Step f: dispatch confirmation: the recommendation display submodule of the dispatch confirmation interaction module displays a list of recommended caregivers to the management personnel, including information such as the caregiver's photo, name, professional skill highlights, comprehensive matching score, and service history data; Step g: The manager manually reviews and adjusts the recommendation results according to the actual situation through the manual review submodule, and finally determines the caregiver to be assigned; Step h: Service implementation and data collection: During the service process, the caregiver uses the operation record submodule of the service implementation data collection module to record the time, content, and changes in the patient's physical indicators of key nursing operations, and uses the emergency record submodule to record the emergencies and their handling methods during the service process. The communication record submodule records the content and communication methods of the caregiver's communication with the patient and his family members; Step i: Evaluation feedback integration: The score collection submodule of the evaluation feedback integration module collects the multi-dimensional scores of the patients on the caregivers and standardizes the score data. The text feedback collection submodule collects the text evaluation content of the patients on the caregivers and performs sentiment analysis; Step j: Machine learning model comparison and update: The model experiment submodule of the machine learning model comparison module runs multiple recommendation system strategies at the same time, including collaborative filtering model, content recommendation model, and deep learning model, and sets different parameter combinations for experiments. The effect evaluation submodule evaluates the performance of different models by calculating accuracy, recall rate, and patient satisfaction. Step k: Third-party data fusion and optimization: The data interface submodule of the third-party data fusion module establishes a secure and stable interface with different third-party health databases to obtain patient health data. The data parsing submodule parses data in various formats from third-party databases and extracts key information. The data fusion submodule merges third-party data with existing patient demand information in the system and matches caregivers with corresponding nursing experience based on the patient's disease diagnosis. The privacy and security submodule ensures compliance with data privacy regulations during the data fusion process.

2. The caregiver recommendation algorithm according to claim 1, characterized in that: The care needs in step a include care type, special care requirements and preferences for caregivers.

3. The caregiver recommendation algorithm according to claim 1, characterized in that: In step b, the caregiver enters information to set his or her own working time schedule, including the number of working days per week and the time period available for service each day, and writes a personal profile to introduce work experience and nursing philosophy.

4. The caregiver recommendation algorithm according to claim 1, characterized in that: In the step c, the data standardization unit unifies the dimensions of different types of data so that all data can be comprehensively evaluated at the same scale.

5. The caregiver recommendation algorithm according to claim 1, characterized in that: In step d, the third-party data access unit establishes an interface with the third-party health database to obtain detailed health data of the patient and integrates the data into the matching model.

6. The caregiver recommendation algorithm according to claim 1, characterized in that: In step e, the weight adjustment submodule dynamically adjusts the weight of each evaluation unit according to different nursing scenarios and special needs of patients to generate a final matching score. The caregiver sorting submodule sorts the caregivers in descending order according to the matching scores, generates a recommendation list, and puts the caregivers with the highest matching degree at the front.

7. The caregiver recommendation algorithm according to claim 1, characterized in that: In step g, the dispatch feedback submodule feeds back the dispatch results to the intelligent matching module for algorithm learning and optimization.

8. The caregiver recommendation algorithm according to claim 1, characterized in that: In step h, the data uploading submodule uploads the collected data to the data storage and management module in real time.

9. The caregiver recommendation algorithm according to claim 1, characterized in that: In step i, the data integration submodule integrates the rating and text feedback data, stores them in the data storage and management module, and transmits the key feedback information to the intelligent matching module.

10. The caregiver recommendation algorithm according to claim 1, characterized in that: In step j, the model selection submodule selects the most suitable model combination or parameter setting according to the effect evaluation result, and updates the model configuration of the matching algorithm.

Images