Method for improving chemotherapy efficiency of oncology department

Abstract

The invention discloses a method for improving chemotherapy efficiency of a tumor department. The method comprises the following steps that a whole disease course management system is utilized, after a chemotherapy regimen of a patient is determined, individual case management and the patient confirm follow-up return time, within effective time before the return time, the system will inform a user to complete specified biochemical inspection, the patient uploads an effective inspection list to the system, and the system automatically judges whether the condition of next chemotherapy is met or not; submitting examinations and medical records of the patients meeting the conditions to doctors of the daytime chemotherapy center, and making prescriptions by the doctors according to results; the system informs the patient of exact hospital return date and time and appointment a bed, the patient directly checks in the daytime chemotherapy center according to appointment prompts in the system after arriving at the hospital, the daytime chemotherapy center confirms that elements are complete, medicines are directly dispensed, infusion is arranged, and the patient can directly leave the hospital after infusion is finished. The method has the beneficial effects that the preparation time of a patient before chemotherapy is shortened; the extra expenditure of outpatient service is reduced; the bed utilization efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for improving the efficiency of chemotherapy in oncology departments. Background Technology

[0002] Based on feedback from the oncology department regarding current chemotherapy practices, the three main shortcomings (pain points) of the current technology are as follows: a. For patients, chemotherapy requires periodic hospital visits, and each visit, excluding treatment, takes a long time, averaging 2.5 days.

[0003] b. For hospitals, periodic chemotherapy consumes a large amount of bed and blood test resources.

[0004] c. For doctors, cyclical chemotherapy takes up outpatient time and appointment slots. Summary of the Invention

[0005] The present invention aims to overcome the aforementioned shortcomings in the prior art and provides a method for improving the efficiency of chemotherapy in oncology departments by shortening the chemotherapy cycle and increasing bed utilization.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method to improve the efficiency of chemotherapy in oncology departments includes the following steps: (1) System identification stage: After the patient’s chemotherapy regimen is determined using the whole course management system, the case management confirms the subsequent return time with the patient. Within the valid time before the return time, the system will notify the user to complete the specified biochemical test. The patient uploads the valid test report to the system, and the system automatically determines whether the conditions for the next chemotherapy are met. If the conditions are met, the patient’s examination and medical record will be submitted to the doctor of the day chemotherapy center, and the doctor will prescribe a pre-prescription based on the results. (2) Intelligent bed scheduling stage: The system informs the patient of the exact date and time of return to the hospital and reserves a bed. After arriving at the hospital, the patient goes directly to the day chemotherapy center according to the reservation prompt in the system. The day chemotherapy center confirms that all the elements are complete, directly issues the medicine, and arranges the infusion. After the infusion is completed, the patient can leave the hospital directly.

[0007] Through the technical solution design of this method, the average preparation time for patients before chemotherapy has been reduced from 2.5 days to 40 minutes; the patient's attending physician does not need to contact the patient during the periodic chemotherapy, reducing additional outpatient expenses; and the hospital can increase the utilization efficiency of its beds by 30%.

[0008] Preferably, in step (1), the specific process of automatic system identification is as follows: (11) The patient uploads a picture of the medical record / examination report. The picture is recognized and converted into text and the valid data is extracted. Based on the extracted valid data, the rule engine is used to intelligently determine the chemotherapy conditions. (12) After the chemotherapy conditions are met, the doctor's processing form is sent out, and the doctor processes and issues a pre-prescription.

[0009] As a preferred option, in step (11), specifically: image preprocessing is performed on the original images uploaded by the patient, text recognition is performed using medical-specific OCR technology, structured data extraction is performed on the patient's uploaded medical records / examination reports, data verification is performed on the extracted structured data, the verified structured data is stored in the medical database, and key fields are extracted simultaneously: patient basic information, test indicators, imaging examination results, and previous treatment records; the Drools rule engine is used to implement multi-cancer rule template configuration, where the rules are divided into hard constraints that must be met and soft constraints that optimize the target, supporting the adjustment of judgment conditions according to different treatment stages, and the rule weights can be dynamically configured to adapt to clinical guideline updates.

[0010] As a preferred option, in step (12), the processing order includes the patient's basic information, test results, and system judgment suggestions. The doctor can adjust or confirm the chemotherapy plan by viewing the system judgment basis, thereby generating a pre-prescription with one click.

[0011] Preferably, in step (2), the specific procedure for reserving a bed is as follows: (21) Pre-prescription notification: The patient is notified through multiple channels such as APP / SMS, and the details of the pre-prescription are displayed, as well as the precautions for chemotherapy are explained; (22) Collection of bed reservation intentions: A reservation confirmation dialog box pops up on the patient's end. Selecting "Yes" will take you to the request supplement interface, which supports modifying / canceling the reservation; (23) Additional requests: Request types include time preferences, special bed requirements, accompanying person requirements, and other personalized requirements; (24) OptaPlanner Intelligent Bed Scheduling: Planning entities and variables, including bed information, time slot information, and patient information; setting hard constraints and soft constraints, among which hard constraints include non-overlapping time slots for the same bed, special cases requiring specific equipment beds, and matching of medical and nursing resources, and soft constraints include trying to meet patients' time preferences, arranging patients with the same type of cancer in a concentrated manner, and reducing bed vacancy time; the optimization algorithm uses OptaPlanner's Construction Heuristics + Local Search combination to calculate the constraint conflict rate, provide the Top 5 bed scheduling schemes, and support real-time replanning to cope with changes; (25) Patient confirmation and bed reservation: The system pushes the bed allocation plan, the patient selects the preferred plan, the system locks the bed, generates the reservation voucher, and updates the system bed status synchronously.

[0012] Preferably, in step (24), the bed planning process for intelligent bed arrangement is as follows: (241) Collect the chessboard grids of the beds, including the grid information and grid codes, and obtain the set of chessboard cells of the beds C; (242) Collect time slot information. Each day is divided into three time slots: morning, afternoon, and evening. The time slot ID set T is obtained by cyclically collecting the time slots. (243) Collect bed information, which includes location information and additional attributes, and obtain the bed ID set B by looping through the bed information; (244) Calculate the mapping M of each bed chessboard cell, where M: T → C satisfies a strict one-to-one mapping, the time slot and the bed chessboard cell are bidirectionally reversible, and each bed chessboard cell can only belong to one time slot at the same time; (245) Assemble the chessboard and chess pieces, wherein the chessboard information includes bed information, time slot information, patient information, and whether the patient is a fixed patient; the chess piece information consists of patient information, which is divided into chess pieces that have been booked and confirmed by the patient and chess pieces that are to be planned. (246) For changes in the planning pieces, hard constraints: ensure that only one bed will be arranged for a patient at the same time, calculate the unique key value K1 of patient ID / time slot ID, and calculate the set of multiple patients appearing in the same time slot; (247) Soft constraints: Calculate the grid of the patient bed arrangement using patient information and time slot information, and determine whether the grid of the patient bed arrangement is continuous; (248) After the chess pieces are changed, determine whether the bed table is available. Step (246) is a hard constraint. If it is not satisfied, re-enter step (245) for planning. (249) Obtain the top 5 bed list for selection, process the bed list data and resolve conflicts, output the top 5 bed list containing conflict information, save the top 5 copies of the bed list, and assemble the conflict deduction items and scores of the top 5 copies of each bed list together and return them.

[0013] The beneficial effects of this invention are: the average preparation time for patients before chemotherapy is reduced from 2.5 days to 40 minutes; the patient's attending physician does not need to contact the patient during the periodic chemotherapy, reducing additional outpatient expenses; and the hospital increases the utilization efficiency of its beds by 30%. Attached Figure Description

[0014] Figure 1 This is a flowchart of the bed planning process in this invention. Detailed Implementation

[0015] The present invention will now be further described with reference to the accompanying drawings and specific embodiments.

[0016] A method to improve the efficiency of chemotherapy in oncology departments includes the following steps: (1) System identification stage: After the patient’s chemotherapy regimen is determined using the whole course management system, the case management confirms the subsequent return time with the patient. Within the valid time before the return time, the system will notify the user to complete the specified blood routine and blood biochemistry tests. The patient can complete the tests at a nearby hospital. The patient uploads the valid test report to the system, and the system automatically determines whether the conditions for the next chemotherapy are met. If the conditions are met, the patient’s examination and medical records will be submitted to the doctor of the day chemotherapy center, and the doctor will prescribe a pre-prescription based on the results. The specific process of automatic system identification is as follows: (11) The patient uploads a picture of the medical record / examination report. The picture is recognized and converted into text and the valid data is extracted. Based on the extracted valid data, the rule engine is used to intelligently determine the chemotherapy conditions. Specifically, the process involves: preprocessing the original images uploaded by patients; using medical-grade OCR technology for text recognition; extracting structured data from the uploaded medical records / examination reports; validating the extracted structured data; storing the validated structured data in a medical database; and simultaneously extracting key fields: patient basic information, laboratory indicators, imaging examination results, and previous treatment records. The Drools rule engine is used to configure rule templates for multiple cancer types. Rules are divided into hard constraints that must be met and soft constraints that optimize for specific goals. The system supports adjusting judgment conditions based on different treatment stages, and rule weights can be dynamically configured to adapt to updates in clinical guidelines.

[0017] (i) User uploads medical records / examination reports Technical Implementation: Medical-specific OCR technology is used to recognize uploaded medical documents; Baidu's Medical Laboratory Report Recognition API can extract structured data such as name, hospital name, examination items, results, and reference range. It has been specifically optimized to address common issues such as tilting and bending, resulting in high recognition accuracy and an average recognition time of less than 2 seconds per image.

[0018] (ii) Image recognition to text conversion and extraction of valid data Data processing flow: Original image → Image preprocessing (denoising / correction) → OCR text recognition → Structured data extraction → Data verification → Storage in medical database.

[0019] Key field extraction: Patient basic information (name, age, medical record number); Laboratory indicators (complete blood count, liver and kidney function tests, etc.); Imaging examination results; Previous treatment records.

[0020] (iii) Rule engine intelligently judges chemotherapy conditions Rule engine architecture: Multiple cancer type rule templates can be configured using rule engines such as Drools; The rules are divided into hard constraints (which must be satisfied) and soft constraints (optimization objectives); The National Cancer Center has implemented a rule-based real-time quality control system.

[0021] Examples of common chemotherapy condition rules:

[0022] Dynamic rule management: Support adjusting judgment criteria according to different treatment stages; Rule weights can be dynamically configured to adapt to updates in clinical guidelines.

[0023] (12) Once the chemotherapy conditions are met, a doctor's processing form is sent out, and the doctor processes and issues a pre-prescription; Specifically, the processing order includes the patient's basic information, test results, and system judgment suggestions. Doctors can adjust or confirm the chemotherapy plan by reviewing the system's judgment criteria, thereby generating a pre-prescription with one click.

[0024] (iv) Once the conditions are met, send the doctor's processing order. System Integration: Interacts with the HIS system via the HL7 / FHIR standard; The processing form includes: basic patient information, test results, and system judgment suggestions; It supports mobile push notifications, allowing doctors to process requests anytime.

[0025] (v) The doctor treats and issues a pre-prescription. Doctor workstation functions: Check the system's judgment criteria; Adjust or confirm the chemotherapy regimen; Generate pre-prescriptions with one click; Special cases may trigger multidisciplinary team (MDT) consultations.

[0026] (2) Intelligent bed scheduling stage: The system informs the patient of the exact date and time of return to the hospital and reserves a bed. After arriving at the hospital, the patient goes directly to the day chemotherapy center according to the reservation prompt in the system. The day chemotherapy center confirms that all the elements are complete, directly issues the medicine, and arranges the infusion. After the infusion is completed, the patient can leave the hospital directly. The specific process for booking a bed is as follows: (21) Prescription notification Patient-side functions: Multi-channel notification via APP / SMS; Prescription details displayed; Precautions for chemotherapy.

[0027] (22) Collection of bed reservation intentions Interaction flow: A confirmation dialog box will pop up. Selecting "Yes" will take you to the request supplementation interface; You can modify or cancel your appointment.

[0028] (23) Additional requests Types of requests that can be collected: Time preference (morning / afternoon); Special bed requirements (such as by the window); Accompanying personnel requirements; Other personalized needs.

[0029] (24) OptaPlanner intelligent bed arrangement Planning entities and variables: Bed availability information; Time slot information; Is the patient information fixed? Patient information (planning variables, values ​​that change during planning).

[0030] Core constraints: Hard constraints: The time spent in the same bed does not overlap; Special cases require specific equipment and beds; Matching of medical and nursing resources (such as qualifications of chemotherapy nurses).

[0031] Soft constraints: Try to meet the patient's time preferences; Patients with the same type of cancer are grouped together; Reduce the time that beds are vacant.

[0032] Optimization algorithm: Use OptaPlanner's Construction Heuristics + Local Search combination; Calculate constraint conflict rate and provide Top 5 bed arrangement schemes; Supports real-time replanning to respond to changes.

[0033] The optimization algorithm is explained in detail below: During initialization, a heuristic algorithm is used, where each piece is placed individually to obtain the optimal score. Subsequent iterations employ simulated annealing combined with the tabu algorithm. The principle of simulated annealing is to use a probabilistic algorithm to move pieces after each batch of moves, following initialization. Higher temperatures result in greater randomness, while lower temperatures approximate a hill-climbing algorithm (ratio algorithm) to explore more possibilities. The tabu algorithm aims to find a better solution quickly and prevent getting trapped in local optima while failing to find the global optimum.

[0034] Conflict score: The score after moving a batch of cells each time; the score is the sum of the scores for each constraint.

[0035] The score is calculated as a penalty point, which is a negative score. The closer the total is to 0, the better the solution is considered.

[0036] For example, if there's a no-schedule rule in the current business, prohibiting Zhang San from scheduling a bed on October 1, 2025. If Zhang San is scheduled to sleep on October 1, 2025 during the planning stage, 1 point will be deducted. In this case, the score for this constraint is -1 point.

[0037] like Figure 1 As shown, the intelligent bed-spacing process is as follows: (241) Collect the grid cells of the bed chessboard and obtain the set of bed chessboard cells C. Chessboard grid information: Bed information, time slot, patient information H, whether it is fixed (already arranged); Chessboard grid encoding: Composed of chessboard grid information.

[0038] (242) Collect time slot information and obtain the time slot ID set T Retrieves a single time slot using a loop time slot; Each day is divided into three time slots: morning, afternoon, and evening. For example, if you need to schedule beds for the next two days, the time slots would be t1, t2, t3, t4, t5, and t6.

[0039] (243) Collect bed information and obtain the bed ID set B. Retrieve individual bed information from circulating bed availability; Bed information includes location information (specifically, which building, floor, room, and bed number) and additional attributes (whether it is a VIP room, a single room, etc.).

[0040] The purpose of this set of bed IDs B is: to be used for chessboard tiling (each chessboard cell can be considered a unique value of bed ID + time slot ID); and to be used for business (there are constraints on beds in various constraints, and specific bed information can be obtained through the bed ID for subsequent logical calculations of bed business).

[0041] (244) Calculate the mapping M of each bed chessboard grid. Where M: T → C Variable definition: T: Set of time slot IDs (discrete time intervals) C: Set of chessboard cells (two-dimensional coordinate space) Constraints: Strict one-to-one mapping; Bidirectional reversible query between time slots and bed chessboard cells; No duplicate assignment (each cell can only belong to one time slot at the same time).

[0042] This step mainly involves tiling the chessboard, which is done by tiling it using the Cartesian product of (bed ID + time slot ID) to assemble the chessboard.

[0043] (245) Assemble the chessboard and chess pieces The chessboard information includes bed information, time slot information, whether it is fixed (whether patient information is fixed), and patient information (planning variables, values ​​that change during planning). The chess piece information consists of patient information, divided into pieces that have already been placed (those that have been booked and confirmed by the patient) and pieces that are yet to be planned.

[0044] (246) Changes in the chess pieces of the plan Hard constraint: Ensure that a patient will only be assigned to one bed at a time. Calculate the unique key value K1 of patient ID / time slot ID and calculate the set of multiple patients appearing at the same time (same time slot). The purpose of this restriction is to prevent the same patient from being assigned to two beds at the same time. For example, Zhang San cannot be assigned to bed 101-1 and bed 102-2 at the same time on October 1, 2025.

[0045] Calculation method: The planned grid is grouped by (time slot id + patient id) K1. If there is an element in the group that is greater than 1, a penalty point is required.

[0046] (247) Soft constraints Patient planning time slot quantity limit: For example, if patient Zhang San needs to wait for a bed for one day, he will be in a maximum of 3 cubicles; By using patient information and time slot information, it is possible to calculate how many bed grids the patient is assigned to.

[0047] Patient bed time continuity constraint: For example, if patient Zhang San needs to wait for a bed for one day, he will definitely hope to be in the same bed consecutively. By combining patient information and time slot information, it is possible to calculate whether the chessboard grid of the patient's assigned bed is continuous.

[0048] The purpose of the continuous logic is twofold: First, to prevent patients from having to queue for a bed on the morning and evening of October 1, 2025, only to miss the afternoon and then queue again on the morning of October 2, 2025, which would be very inconvenient for them. Second, it would be inconvenient for patients to have different bed numbers for the morning, afternoon, and evening of October 1, 2025.

[0049] The first calculation method: Filter the planned cells by patient ID to obtain the cell for this patient's bed assignment. Each cell has a time slot ID. Using the time slot ID, you can retrieve the specific time information from the time slot set, such as the month and day, and whether it's morning, noon, or afternoon. For example, if Zhang San needs to schedule beds for 2 days (6 sessions), the time slot information obtained is: 2025-10-01 morning, 2025-10-02 morning, 2025-10-02 afternoon, 2025-10-02 evening, 2025-10-03 morning, 2025-10-04 morning. The maximum consecutive time slot is 4, so the penalty is 6 - 4 = 2.

[0050] The second calculation method is similar to the first, except that it changes from continuous time to continuous time plus continuous bed availability.

[0051] (248) Determine if the bed table is usable after the chess pieces are moved. Step (246) is a hard constraint. If it is not met, the process will re-enter step (245) for planning.

[0052] (249) Generate a copy of the bed list Obtain a top 5 bed list for selection, process the bed list data and resolve conflicts, output a top 5 bed list including conflict information, save a top 5 copy of the bed list, and assemble the conflict deduction items and scores of the top 5 copies of each bed list and return them together.

[0053] Scores: OptaPlanner provides an interface with BestSolutionConsumer for each score change, allowing us to listen, process the logic, and save the changed data.

[0054] Deductions: Before returning the top 5 solutions, we run the constraints on each solution again, and save the deduction scores and the grid rules of the deduction design and return them.

[0055] (25) Patient confirmation and bed reservation Confirmation process: The system pushes out bed allocation plans (including conflict rate explanations); Patients choose preferred treatment options; The system locks in the bed and generates a reservation voucher; The bed status in the HIS system is updated synchronously.

[0056] The core functional highlight of this method is: 1. Request Collection: Supports flexible configuration of 10+ personalized requests; 2. Intelligent Planning: The OptaPlanner-based algorithm can generate Top 5 solutions; 3. Conflict visualization: Each solution clearly indicates the constraint conflict rate and the number of demands satisfied; 4. Real-time synchronization: Bed availability is updated to the HIS system in real time to avoid double booking.

[0057] Through the technical solution design of this method, the average preparation time for patients before chemotherapy has been reduced from 2.5 days to 40 minutes; the patient's attending physician does not need to contact the patient during the periodic chemotherapy, reducing additional outpatient expenses; and the hospital can increase the utilization efficiency of its beds by 30%.

Claims

1. A method for improving the efficiency of chemotherapy in an oncology department, characterized by, Specifically comprising the following steps: (1) System identification stage: after the patient's chemotherapy plan is determined, the case management and the patient confirm the subsequent return time, within the effective time before the return time, the system will notify the user to complete the designated biochemical test, the patient uploads the effective test sheet to the system, the system automatically judges whether the conditions for the next chemotherapy are met, the patient examination and medical record are submitted to the doctor of the day chemotherapy center, and the doctor issues a pre-prescription according to the results; (2) Intelligent bed arrangement stage: the system informs the patient of the exact return date and time and reserves a bed, and the patient goes to the day chemotherapy center directly according to the reservation prompt in the system, and the day chemotherapy center confirms that the elements are complete, directly issues medicines, arranges infusion, and after the infusion is completed, the patient can directly leave the hospital.

2. The method of claim 1, wherein the step of administering the chemotherapy is performed in a hospital setting. In step (1), the specific process of automatic judgment of the system is as follows: (11) The patient uploads the medical record sheet / picture of the test sheet, the picture recognition converts the text and extracts the effective data, and the effective data is used to judge the chemotherapy condition intelligently by using the rule engine; (12) After meeting the chemotherapy condition, the doctor handles the single and issues a pre-prescription.

3. The method of claim 2, wherein the step of administering the chemotherapy is performed in a hospital setting. In step (11), specifically: image preprocessing is performed on the original picture uploaded by the patient, medical special OCR technology is used for text recognition, structured data extraction is performed on the medical record sheet / test sheet uploaded by the patient, data verification is performed on the extracted structured data, the verified structured data is stored in the medical database, and key field extraction is performed simultaneously: patient basic information, test index, image examination result, and past treatment record; the Drools rule engine is used to realize the configuration of multi-cancer rule templates, wherein the rules are divided into hard constraints that must be met and soft constraints that are optimization targets, the judgment conditions can be adjusted according to different treatment stages, the rule weight can be dynamically configured, and the clinical guideline update is adapted.

4. The method of claim 2 or 3, wherein the method is characterized by, In step (12), specifically: the processing single contains patient basic information, test result, and system judgment suggestion, the doctor adjusts or confirms the chemotherapy plan by checking the system judgment basis, thereby generating a pre-prescription by one key.

5. The method of claim 1, wherein the method is characterized by, In step (2), the specific process of reserving a bed is as follows: (21) Pre-prescription notification: the patient is notified through APP / sms multi-channel on the patient side, and the pre-prescription details are displayed, as well as the chemotherapy precautions; (22) Bed reservation intention collection: a reservation confirmation dialog box is popped up on the patient side, selecting "yes" enters the appeal supplement interface, and the reservation can be modified / cancelled; (23) Supplementary additional appeal: appeal types include time preference, special bed requirement, accompanying personnel requirement, and other individualized requirements; (24) OptaPlanner intelligent bed arrangement: planning entities and variables, including bed information, time slot information, and patient information; setting hard and soft constraints, where hard constraints include non-overlapping time on the same bed, special cases requiring specific equipment beds, and matching medical resources, and soft constraints include trying to meet patient time preferences, concentrating patients with the same cancer, and reducing bed vacancy time; the optimization algorithm uses a combination of Construction Heuristics + LocalSearch of OptaPlanner, calculates the constraint conflict rate, provides Top5 bed arrangement schemes, and supports real-time re-planning to respond to changes; (25) Patient confirmation and bed reservation: the system pushes the bed arrangement scheme, the patient selects the preferred scheme, the system locks the bed, generates a reservation voucher, and synchronously updates the system bed status.

6. The method of claim 1, wherein the method is characterized by, In step (24), the bed planning process of intelligent bed arrangement is as follows: (241) Collect bed chessboard cells, including chessboard cell information and chessboard cell code, and obtain a set of bed chessboard cell C; (242) Collect time slot information, divide it into morning, afternoon, and night, three time slots, and obtain a set of time slot ID T by cycling the time slot; (243) Collect bed information, including location information and additional attributes, and obtain a set of bed ID B by cycling the bed information; (244) Calculate the mapping M of each bed chessboard cell, where M: T → C satisfies a strict one-to-one mapping, and the time slot and the bed chessboard cell are bidirectionally reversible, each bed chessboard cell can only belong to one time slot at the same time; (245) Assemble the chessboard and the chess pieces, where the chessboard information includes bed information, time slot information, patient information, and whether it is a fixed patient information; The chess piece information is composed of patient information, divided into chess pieces that have been reserved and determined by patients and chess pieces to be planned; (246) For the chess pieces to be planned, the hard constraint is to ensure that a patient will only be arranged on one bed at the same time, calculate the unique key value K1 of patient ID / time slot ID, and calculate the set of multiple patients appearing in the same time slot; (247) Soft constraint: calculate the bed chessboard cell of the patient arrangement through patient information and time slot information, and calculate whether the bed chessboard cell of the patient arrangement is continuous; (248) Judge whether the bed table is available after the chess piece moves, where step (246) belongs to the hard constraint, if it is not satisfied, re-enter step (245) for planning; (249) Get top5 bed tables for selection, process bed table data and solve conflicts, output top5 bed tables containing conflict information, save top5 copies of bed tables, and return the conflict deduction items and scores of each bed table top5 copy together.

Images