Information processing system, information processing method, and program

Abstract

Provided are an information processing system, an information processing method, and a program that reduce the burden on medical workers and improve treatment throughput. This information processing system has a memory that stores a program, and a processor that executes processing in accordance with the program. The processor executes the program to: estimate a change over time of a clinical examination value of a patient during treatment on the basis of a preset parameter; and estimate, on the basis of a variation range of the parameter, a prediction range that is a range in which the change over time of a prediction value, which is the estimated clinical examination value, can stochastically take place.

Description

Information processing systems, information processing methods, and programs

[0001] This disclosure relates to a technology for developing treatment plans based on clinical test values ​​obtained from patients.

[0002] Lymphocytes, including killer T cells and helper T cells, are cells that play a role in the human immune system and are known to possess high radiosensitivity. Among patients who received radiation therapy, those with a small decrease in absolute lymphocyte count (ALC) may have a better prognosis (see, for example, Non-Patent Document 1).

[0003] These research findings suggest that treatment effectiveness can be improved by designing radiation therapy plans to prevent the reduction of ALC. One example of a treatment planning method to prevent the decrease of ALC is disclosed in Patent Document 1. Patent Document 1 discloses that areas affected by radiation exposure to ALC are designated as lymphocyte-related organs at risk (LOARs), and dose constraints are applied to them in the same way as to other organs at risk (OARs).

[0004] Non-Patent Document 2 discloses a mathematical model that predicts the increase or decrease in ALC during the treatment period, using dose indicators of LOAR obtained from the treatment plan as input. The dose indicators include, for example, dose-volume histograms (DVH). Such mathematical models are useful in determining the balance of dose between the target and LOAR while maintaining ALC. Non-Patent Document 2 proposes a method for adjusting the internal parameters of the mathematical model to reproduce the temporal changes of ALC on average, based on measured ALC values ​​obtained from the treatment results of several dozen patients. The main internal parameters include the radiosensitivity of lymphocytes and the recovery rate of lymphocytes per unit time.

[0005] Cho Y, et al. , “Lymphocyte dynamics during and after chemo-radiation correlate to does and outcome in stage III NSCLC patients Radiother Oncol. 2022 Mar;168:1-7

[0006] Jin JY, Mereniuk, et al. , “A framework for modeling radiation induced lymphopenia in radiotherapy”, Radiother Oncol. 2020 Mar; 144:105-113

[0007] U.S. Patent Application Publication No. 2023 / 0094681

[0008] However, there are individual differences in the radiosensitivity of lymphocytes or the rate of lymphocyte recovery per unit time. The mathematical model disclosed in Non-Patent Literature 2 may contain errors in the predicted temporal progression of ALC for individual patients. Healthcare professionals, including physicians, must estimate the error in ALC based on their own judgment and determine the radiation dose.

[0009] For example, if medical professionals become overly concerned with dose constraints on LOAR and overestimate the error in ALC, the radiation dose to the target will be insufficient, resulting in inadequate treatment. Medical professionals must estimate the error in ALC and determine the dose each time they administer radiation therapy. Furthermore, if the effect of a single radiation therapy is insufficient, the number of radiation therapy sessions given to a single patient will increase. This not only increases the burden on medical professionals but also reduces the treatment throughput. This problem is not limited to radiation therapy where ALC is used as a clinical test value, but is common to many treatments that are affected by individual differences.

[0010] One of the purposes included in this disclosure is to provide information processing systems, information processing methods, and programs that reduce the burden on healthcare professionals and improve treatment throughput.

[0011] An information processing system according to one aspect of the present disclosure includes a memory for storing a program and a processor for executing processing according to the program, wherein the processor, by executing the program, estimates the temporal changes in a patient's clinical test values ​​during treatment based on preset parameters, and estimates a predicted range, which is the range that the temporal changes in the estimated clinical test values ​​(predicted values) can take probabilistically, based on the range of variation of the parameters.

[0012] According to one aspect of this disclosure, the predicted range of clinical laboratory values ​​in treatment is estimated based on the range of variation of a parameter that estimates the temporal progression of clinical laboratory values. Healthcare professionals can understand the range of error in clinical laboratory values ​​from the predicted range of clinical laboratory values. This reduces the burden on healthcare professionals and improves treatment throughput.

[0013] This is a schematic diagram showing an example configuration of an adaptive radiotherapy system according to the embodiment. This is a flowchart showing an example of the operation procedure of machine learning by the treatment planning device according to the embodiment. This is a flowchart showing an example of the operation procedure of the treatment planning device in adaptive radiotherapy using the adaptive radiotherapy system according to the embodiment. This is a graph showing an example of the DVH of an arbitrary ROI among multiple ROIs. This is a graph showing an example of the temporal change of ALC displayed on the output device shown in Figure 1. This is a diagram showing an example of an image when the output device shown in Figure 1 displays the previous prediction result and the latest prediction result.

[0014] An embodiment of the information processing system of this embodiment will be described with reference to the drawings. In the embodiment, the description will be given in a case where the information processing system of this embodiment constitutes a part of an adaptive radiotherapy system. Hereafter, particle beams may also be referred to as radiation, and these terms will not be distinguished unless otherwise specified.

[0015] Figure 1 is a schematic diagram showing one example configuration of an adaptive radiotherapy system according to an embodiment. The adaptive radiotherapy system 1 comprises a treatment planning device 2, a radiation irradiation device 3, an examination device 4, and an information processing system 10. The information processing system 10 includes the treatment planning device 2 and a storage device 5. The treatment planning device 2 is connected to the radiation irradiation device 3, the examination device 4, and the storage device 5 via a network 100. The network 100 is a LAN (Local Area Network) or a WAN (Wide Area Network). The network 100 may include the Internet.

[0016] The radiation irradiation device 3 is a device that irradiates the affected area of ​​a patient with a particle beam. The radiation irradiation device 3 is a device that irradiates with a particle beam using, for example, a spot scanning method. The spot scanning method is a method in which multiple spots are virtually created in three dimensions within and around the target area to be irradiated with radiation, such as a tumor in the patient's body, and a narrow beam is irradiated to each spot. A predetermined irradiation dose is set for each of the multiple spots. When the radiation irradiation device 3 irradiates one of the multiple spots with the predetermined dose, it deflects the beam to the next spot and irradiates each spot with the predetermined dose in order. By applying the predetermined dose to all spots, the radiation irradiation device 3 forms the desired dose distribution in the target area.

[0017] In this embodiment, spot scanning particle beam therapy is described as an example, but the adaptive radiotherapy of this embodiment may also be applied to radiotherapy using intensity modulation with X-rays. Examples of such therapies include intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT).

[0018] This section briefly describes the operation of an intensity-modulated X-ray irradiation device, such as IMRT or VMAT. The X-ray irradiation device irradiates the target with X-rays from multiple directions. While the dose distribution formed within the target by X-rays irradiated from multiple directions is non-uniform, the superposition of contributions from all directions provides a uniform dose distribution within the patient's body that matches the three-dimensional shape of the target. In this case, the fluence distribution of the X-rays to be irradiated from each direction is obtained by solving an inverse problem that determines the dose from each direction from the overall required dose. This inverse problem is solved by a computer. In this embodiment, the treatment planning device 2 solves the inverse problem. Furthermore, a multi-leaf collimator may be installed between the X-ray source and the isocenter to achieve an arbitrary fluence distribution of X-rays.

[0019] The testing device 4 is a device that analyzes blood collected from a patient and measures ALC. When the blood to be tested is introduced into the testing device 4, it measures the ALC in the blood and stores the measurement result in the storage device 5. The testing device 4 may also receive the measured ALC from another device. In this embodiment, the clinical test value used to verify the effectiveness of treatment is described as ALC, but other clinical test values ​​may be used as long as they are physical quantities that are affected by radiation therapy and related to the effectiveness of treatment.

[0020] The storage device 5 is, for example, an HDD (Hard Disk Drive) or an SSD (Solid State Drive). The storage device 5 has a database in which ALC measurement information is accumulated in chronological order for each of several patients, which is information that is a combination of the measured value, which is the clinical test value of ALC measured during the course of treatment, and the date on which it was measured. In the database, for example, the ALC measurement information is accumulated in chronological order corresponding to a patient identifier, which is a different identifier for each patient. The database may also store X-ray CT (Computed Tomography) images taken inside the body, including the affected area of ​​the patient, corresponding to the patient identifier.

[0021] The treatment planning device 2 is an information processing device such as a computer that performs various information processing. The treatment planning device 2 is, for example, a PC (Personal Computer). The treatment planning device 2 has a processor 11, memory 12, input device 13, output device 14, communication device 15, and disk drive 16. Each of the processor 11, memory 12, input device 13, output device 14, communication device 15, and disk drive 16 is connected via a bus 17.

[0022] Memory 12 stores programs executed by the processor 11, information necessary for various arithmetic operations performed by the processor 11, and the results of those operations. Memory 12 includes non-volatile memory such as flash memory or ROM (Read Only Memory). In addition to non-volatile memory, memory 12 may also include volatile memory such as RAM (Random Access Memory).

[0023] The processor 11 is, for example, a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit). When the treatment planning device 2 starts operating, the processor 11 reads a program from the memory 12 and executes the program. The start of operation of the treatment planning device 2 is, for example, when power is turned on to the treatment planning device 2. The control device 20 is configured when the processor 11 executes the program stored in the memory 12. The control device 20 controls the entire treatment planning device 2 and performs various processes related to treatment planning. The processor 11 reads the information necessary for calculation processing from the storage device 5 and stores it in the memory 12.

[0024] The control device 20 uses multiple ALC measurement data stored in the storage device 5 for each of the multiple patients as training data to train an ALC prediction model, which is a model for predicting ALC. The control device 20 also learns a parameter variation range prediction model, which is the variation range of the parameters set in the ALC prediction model. The parameters are, for example, the radiosensitivity of lymphocytes and / or the recovery rate of lymphocytes per unit time. Based on the above parameters, the control device 20 estimates the temporal progression of the patient's ALC during treatment. Based on the parameter variation range, the control device 20 estimates a prediction range, which is the range that the temporal progression of the predicted value, which is the estimated ALC, can probabilistically take.

[0025] The input device 13 is a device that receives instructions from the operator to the treatment planning device 2. In this embodiment, the operator is a medical professional, including a physician. The input device 13 is, for example, a mouse and a keyboard. The output device 14 is a device that notifies the operator of various information, such as the treatment plan. The output device 14 is, for example, a display that outputs images and a speaker that outputs sound. In this embodiment, the case where the output device 14 is a display will be described.

[0026] The disk drive 16 is a device that reads information from and writes information to optical discs such as DVDs (Digital Versatile Discs). The recording medium from which the treatment planning device 2 reads and writes information is not limited to optical discs. A program executed by the processor 11 may be stored on the recording medium, and the program may be installed in the memory 12 via the disk drive 16 from the recording medium. The communication device 15 is a communication interface that connects to external devices in a communicative manner. In this embodiment, the communication device 15 sends and receives information to and from the radiation irradiation device 3, the examination device 4, and the storage device 5 via the network 100 according to a predetermined communication protocol.

[0027] Furthermore, some or all of the functions of the control device 20 may be executed by a dedicated circuit such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field-Programmable Gate Array). Also, although Figure 1 shows a configuration in which the processor 11 is connected to other devices such as memory 12 via a bus 17, the connection means is not limited to the bus 17. For example, the processor 11 may be connected to other devices via a LAN or WAN. In addition, the information processing system 10 may be provided with multiple processors 11. In this case, the multiple processors 11 may be distributed among multiple information processing devices such as PCs or servers.

[0028] Furthermore, Figure 1 shows a configuration in which the treatment planning device 2 is communicated with other devices such as the radiation irradiation device 3 and the examination device 4 via the network 100, but these connection configurations are not limited to the configuration shown in Figure 1. The communication device 15 may be connected to each of the other devices, including the radiation irradiation device 3, via dedicated signal lines. The information transmission medium may be wired, radio waves, or a combination thereof. In addition, the storage device 5 may be provided separately from the information processing system 10, which includes the treatment planning device 2.

[0029] Next, the operation of the adaptive radiotherapy system 1 of this embodiment will be described. First, the process by which the treatment planning device 2 creates a predictive model to predict the range of parameter fluctuations will be explained with reference to Figure 2. Figure 2 is a flowchart showing an example of the operation procedure of machine learning by the treatment planning device according to this embodiment.

[0030] The storage device 5 contains time-series data y of ALC obtained from patients n who have received treatment in the past. （ｍｅａｓ） ｎ (t ｍ ) is remembered. n = 1, 2, ..., N. In this embodiment, we assume that the total number of people treated is 500. That is, N = 500. ｍis the measurement time of the m-th ALC. m = 1, 2, ···, M. For example, when the ALC is measured 10 times during the treatment period, M = 10. The measurement interval of ALC and the number of measurement points M may vary for each patient.

[0031] In step S1, as a preparation before the start of treatment, when an operator inputs an instruction for parameter adjustment of the ALC prediction model f(y ０ , α, k ｒ , d, t ｍ ) via the input device 13, the control device 20 adjusts the parameters. The ALC prediction model f is a mathematical model for predicting the ALC at time t ｍ . In this embodiment, the parameters to be adjusted are α and k ｒ . α is a parameter representing the radiation sensitivity of lymphocytes. k ｒ is a parameter representing the recovery rate of lymphocytes per unit time.

[0032] Also, y ０ is the numerical value of the baseline ALC. d represents the three-dimensional dose distribution applied to the patient. Although d is represented as a vector, due to limitations in the character notation available in the specification, the vector symbol is omitted. As will be described later, d is calculated by the control device 20 and stored by the storage device 5. Note that in actual treatment, d may be corrected during the treatment period due to tumor shrinkage or organ movement, etc. Therefore, the ALC prediction model f may use different d for each fraction as input.

[0033] Regarding the process of step S1, an example of a specific procedure will be described. First, the control device 20 obtains the parameters α ｎ and k ｎ for each patient n so as to minimize the objective function O shown in Equation (1). At this time, the control device 20 sets y （ｎ） ｒ to the value y of ALC obtained in the first measurement for y ０ （ｍｅａｓ） ｎ (t １ ​The parameter search method is not limited to the method described in this embodiment. Various algorithms, such as the quasi-Newton method, can be used as the parameter search method.

[0034]

[0035] Next, the control device 20 calculates the n α values ​​obtained. ｎ From α ｎ The average value Ave[α ｎ Calculate the n k values ​​obtained （ｎ） ｒ k （ｎ） ｒ The average value Ave[k （ｎ） ｒ The control device 20 calculates α. The control device 20 stores the two calculated average values ​​in memory 12 as the initial parameters of the ALC prediction model f. Furthermore, the control device 20 sets α as the parameter variation range. ｎ Standard deviation σ α and k （ｎ） ｒ Standard deviation σ ｋｒ The control device 20 calculates the two standard deviations σ. α and σ ｋｒ This is stored in memory 12. In this embodiment, the parameter variation range is explained as the standard deviation of the parameter, but a numerical value such as the 95% confidence interval or the full width at half maximum may be used as the parameter variation range.

[0036] In step S2, the control device 20 controls the parameters α and k of the ALC prediction model f. ｒ Regarding the parameter variation range prediction model g σα and g σｋｒ The control device 20 learns the measured value y of the ALC. （ｍｅａｓ） When readjusting during the treatment period based on this, predictive model g of parameter variation range σα and g σｋｒ Refer to the following. An example of a specific procedure for the process in step S2 will be explained.

[0037] First, the control device 20 uses the objective function O shown in equation (2). （Ｍ’） ｎThe parameter α' is minimized for each patient n. ｎ and k' （ｎ） ｒ The following is obtained. Here, M' is the number of times ALC is measured up to the time when the parameters of the ALC prediction model f are adjusted. The control device 20, similar to equation (1), y ０ The value of ALC obtained in the first measurement y （ｍｅａｓ） ｎ (t １ ) is used. In this embodiment, when there are two parameters, M' is 3 ≤ M' < M. That is, parameter α ｎ and k （ｎ） ｒ While α' is a parameter determined from all measurement points based on equation (1), α' ｎ and k' （ｎ） ｒ This represents the parameters that are optimized using data up to the point of treatment.

[0038]

[0039] Next, the control device 20 calculates the parameter α obtained from all measurement points. ｎ and the parameter α' obtained using data up to the end of treatment. ｎ The difference σ' （ｎ） α The parameter k obtained from all measurement points is calculated. This calculation formula is shown in equation (3). The control device 20 also calculates the parameter k. （ｎ） ｒ and the parameter k' obtained using data up to the middle of treatment. （ｎ） ｒ The difference σ' （ｎ） ｋｒ We calculate this. This calculation formula is shown in equation (4).

[0040]

[0041]

[0042] Furthermore, the control device 20 uses the objective function O shown in equation (5). σα A predictive model g of the parameter variation range minimizes the following: σα The parameters are adjusted. In addition, the control device 20 controls the objective function O shown in equation (6).σｋｒ A predictive model g of the parameter variation range minimizes the following: σｋｒ Adjust the parameters of these predictive models g σα and g σｋｒ This can be adjusted in any way. For example, the predictive model g σα and g σｋｒ When a neural network is used to determine this, the weights of the connected layers are subject to adjustment.

[0043]

[0044]

[0045] The control device 20 controls the predictive model g once the learning process is complete. σα and g σｋｒ The parameters α and k of the ALC prediction model f are stored in memory 12. ｒ The explanation was given for the case where there is only one adjustment timing, but similar effects can be obtained even if there are multiple adjustment timings. When there are multiple adjustment timings, the predictive model of the parameter variation range g σα and g σｋｒ This is created for each M' number of ALC measurements taken before parameter adjustment. Parameter variation range prediction model g σα and g σｋｒ This may be constructed using a recurrent neural network with long-term or short-term memory.

[0046] Next, the operating procedure of the treatment planning device 2 in adaptive radiotherapy will be explained. After the preliminary preparations described above are completed, the operator starts treatment for a new patient using the adaptive radiotherapy system 1. Figure 3 is a flowchart showing an example of the operating procedure of the treatment planning device in adaptive radiotherapy using the adaptive radiotherapy system according to the embodiment.

[0047] First, the operator collects blood from the patient and puts the collected blood into the testing device 4. The testing device 4 measures ALC from the patient's blood and then uses the measured ALC as the initial ALC measurement information y （ｍｅａｓ） (t １Store it in the storage device 5. In step S11, the control device 20 acquires the first ALC measurement information from the storage device 5 and writes it into the memory 12.

[0048] Next, the operator inputs the content of the treatment plan in order to formulate the treatment plan. In step S12, the control device 20 sets the treatment plan according to the operator's input. When information specifying the region of interest (ROI) and the exclusion region (OAR) to avoid radiation is input, the control device 20 sets the respective information of the ROI and the OAR. The ROI is the target region irradiated with radiation. The ROI is, for example, a tumor.

[0049] An example of a specific operation will be described. The control device 20 reads the X-ray CT image of the patient from the storage device 5 and causes the output device 14 to display each slice image from the X-ray CT image composed of a plurality of slice images. When the ROI and the OAR are specified for each slice image by the operator via the input device 13, the control device 20 sets the respective regions of the specified ROI and OAR in three-dimensional coordinates. For one patient, there may be two or more ROIs and OARs respectively. The control device 20 stores the three-dimensional coordinates of the set ROI and OAR respectively in the memory 12 as three-dimensional position information.

[0050] Subsequently, when the operator inputs dose constraint information for the set ROI into the treatment planning device 2, the control device 20 sets the dose constraint according to the input content. The dose constraint is, for example, the target dose D （ｊ） Ｔａｒｇｅｔ , the maximum dose D of the OAR （ｊ） ｍａｘ－ＯＡＲ , and the maximum dose volume ratio V of the OAR （ｊ） ｍａｘ－ＯＡＲ . In this embodiment, in addition to these dose constraints, the operator adds a constraint on the maximum number of days Tmax of ALC reduction. The operator assigns a weight w （ｊ） Ｔａｒｇｅｔ , w （ｊ） ＯＡＲ－Ｄｍａｘ , w （ｊ） ＯＡＲ－Ｖｍａｘ and w Ｔ－ｍａｘSet each of them. The control device 20 causes the memory 12 to store the constraint content and weight for each of the plurality of constraints. The subscript j of each value represents an identification number when each of the target and OAR is plural. The control device 20 causes the memory 12 to store the constraints and weights for each of the set plurality of ROIs.

[0051] Then, based on the information stored in the memory 12, the control device 20 generates the objective function F(x) shown in Equation (7) as follows. In Equation (7), x is a parameter to be optimized and is a vector having the irradiation dose for each spot as an element. Although x is represented as a vector, the vector symbol is omitted because there are limitations in the character notation that can be used in the specification.

[0052]

[0053] In Equation (7), t is the number of days when the ALC becomes less than or equal to the threshold value. The threshold value is the lower limit value of the ALC. The threshold value is also the target value for maximizing the dose of radiation irradiated by the medical staff to the ROI. The control device 20 obtains the number of days t by the ALC prediction model f. At this time, the control device 20 uses Ave[α ｎ and Ave[k （ｎ） ｒ as parameters of the ALC prediction model f. In Equation (7), the subscript i indicates the number of the voxel included in each ROI. d ｉ，ｊ is the dose of the i-th voxel. The function θ(x) is a step function. When x is a positive value, θ(x) = 1, and when x is a non-positive value, θ(x) = 0. Also, the function C ［ａ，ｂ］ (x) is an interval constraint function. The function C ［ａ，ｂ］ (x) shown in Equation (7) is defined by Equation (8).

[0054]

[0055] FIG. 4 is a graph showing an example of the DVH of an arbitrary ROI among a plurality of ROIs. The vertical axis of the graph shown in FIG. 4 is volume (%), and the horizontal axis is dose (Gy). FIG. 4 shows the DVH of the j-th ROI. ΔD is the dose D(V （ｊ） ｍａｘ at the point of the maximum dose volume ratio V in the DVH curve shown in FIG. 4.（ｊ） ｍａｘ ) and maximum dose D （ｊ） ｍａｘ It is defined as the difference between the two. The control device 20 calculates d, which represents the dose to each voxel, using equation (9) based on its relationship with a vector x whose elements are the beam irradiation amount for each spot. In equation (9), matrix A is a dose matrix that represents the dose that the pencil beam of particle beams irradiated to each spot delivers to each voxel.

[0056]

[0057] The control device 20 generates an objective function F(x) and then finds the spot irradiation dose x that minimizes the objective function F(x) through iterative search. An example of this procedure is described below. The control device 20 calculates a dose matrix A based on the patient's three-dimensional internal information obtained from the X-ray CT image and the beam irradiation angle of the radiation irradiation device 3 set by the operator. The control device 20 stores the calculated dose matrix A in the memory 12. Subsequently, the control device 20 finds the spot irradiation dose x that minimizes the objective function F(x) through iterative calculation. The termination condition is determined according to indicators such as the total calculation time of the iterative calculation, the number of iterations, or the amount of change in the objective function per iteration. The control device 20 stores the calculated spot irradiation dose x in the memory 12.

[0058] The control device 20 calculates a three-dimensional dose distribution d based on the determined spot irradiation dose x. The control device 20 stores the determined dose distribution d in the memory 12 and the storage device 5. Furthermore, the control device 20 takes the dose distribution d as input and calculates the parameter Ave[α] stored in the memory 12. ｎ ] and Ave[k （ｎ） ｒ Using the ALC prediction model f, the temporal changes in the predicted ALC values ​​are calculated. The control device 20 stores the parameter variation range σ in the memory 12. α and σ ｋｒ Based on this, the control device 20 calculates a prediction range which is the range that the predicted value of ALC can take over time probabilistically. The control device 20 displays the calculated predicted value of ALC over time and the prediction range on the output device 14 (step S13).

[0059] Figure 5 is a graph showing an example of the temporal progression of ALC as displayed on the output device shown in Figure 1. In Figure 5, the vertical axis represents ALC, and the horizontal axis represents the number of observation days. In Figure 5, the measured value REV is indicated by a star, and the temporal progression of the predicted value PDV is shown by a dotted line. In addition, the predicted range PDR, which is the range that the temporal progression of the predicted value can take probabilistically, is shown by a dot pattern. As shown in Figure 5, the control device 20 causes the output device 14 to display the calculated predicted range PDR overlaid on the graph showing the temporal progression of ALC. The control device 20 also causes the output device 14 to display a threshold value that has been set in advance based on clinical experience overlaid on the graph showing the temporal progression of ALC.

[0060] Furthermore, the control device 20 calculates the probability of achieving the threshold based on the predicted range and the threshold. As mentioned above, the threshold corresponds to the target upper limit of the radiation dose irradiated to the patient. Therefore, the greater the lower limit of the predicted range is than the threshold and the closer it is to the threshold, the more effective the treatment will be. The probability of achievement serves as a criterion for the operator to judge whether the treatment plan is successful or not. The control device 20 may also display the calculated probability of achievement on the output device 14 along with a graph showing the temporal changes in ALC.

[0061] Referring to Figure 5, the output device 14 displays on its display unit 14a a graph showing the temporal progression of ALC, overlaid with the measured value of ALC, the predicted range, and the threshold value. By referring to the information displayed on the display unit 14a, the operator can judge the validity of the treatment plan from the relationship between the predicted range, target value, and measured value. The output device 14 also displays the probability of achievement along with the graph showing the temporal progression of ALC on its display unit 14a. In the example shown in Figure 5, since the lower limit of the predicted range is greater than the threshold value, the probability of achieving the target is 100%. By referring to the probability of achievement displayed on the display unit 14a, the operator can efficiently judge whether the treatment plan is successful or not.

[0062] From the perspective of lymphocyte protection, the criteria for determining the success or failure of a treatment plan are whether the ALC predicted by the ALC prediction model f falls below a threshold, or whether the number of days the ALC remains below the threshold exceeds a predetermined upper limit. However, it is difficult for operators to determine the success or failure of a treatment plan based solely on a graph showing the temporal changes in the predicted ALC value, and whether the predicted value is reliable. Operators may fear that the predicted value is overly optimistic and may make overly strict judgments. In this case, the treatment plan is less likely to be approved, leading to the need to revise the treatment plan. As a result, the treatment throughput decreases.

[0063] In contrast, according to this embodiment, as shown in Figure 5, the prediction range is displayed on the display unit 14a of the output device 14, superimposed on a graph of the temporal changes in the predicted values. Furthermore, the display unit 14a of the output device 14 displays the achievement probability, which serves as the pass / fail criterion for the treatment plan. Therefore, the operator can determine the pass / fail status of the treatment plan, taking into account the accuracy and reliability of the ALC prediction model f.

[0064] For example, consider a case where the operator sets the success or failure criterion for a treatment plan as a probability of ALC falling below a threshold of 20% or less. This corresponds to allowing a reasonable margin in the treatment plan based on the accuracy of the prediction model. In this embodiment, there is an advantage that even prediction models with different accuracies can be used in the treatment plan without changing the judgment criteria. According to this embodiment, appropriate judgment criteria can be set, and the number of times the treatment plan needs to be revised can be reduced, thereby improving the treatment throughput.

[0065] The operator determines whether the treatment plan formulated in step S12 is acceptable or unacceptable based on the information displayed on the output device 14. In step S14, the control device 20 determines whether the information input for the treatment plan is acceptable or unacceptable. If, as a result of the determination in step S14, the operator inputs information indicating that the treatment plan is unacceptable, the control device 20 returns to the process of step S12. In step S12, the control device 20 determines the weight w of the dose constraint for each ROI. （ｊ） Ｔａｒｇｅｔ , lol （ｊ）ＯＡＲ－Ｄｍａｘ , lol （ｊ） ＯＡＲ－Ｖｍａｘ and w Ｔ－ｍａｘ The settings are readjusted, and the dose distribution d is optimized again. Meanwhile, if the operator inputs information that the treatment plan is acceptable as a result of the judgment in step S14, the control device 20 proceeds to the process in step S15. In step S15, the control device 20 informs the operator that it will proceed to the first fraction of radiation irradiation.

[0066] However, even if the ALC prediction results indicate that the treatment plan is likely to be unsuitable, the operator may proceed with radiation therapy based on an overall assessment. In actual treatment, the operator does not judge the suitability of the treatment plan solely based on the ALC prediction results, but rather makes a comprehensive judgment based on whether a sufficient dose is being delivered to the ROI, whether the dose to the OAR is sufficiently reduced, etc.

[0067] When the operator receives notification from the treatment planning device 2 that it is time to proceed with radiation therapy, the operator administers the first fraction of radiation to the patient. The operator then inputs information into the treatment planning device 2 indicating that the first fraction of radiation therapy has been administered. After the radiation therapy is completed, the control device 20 determines whether or not the planned fraction remains (step S16). If the planned fraction remains, the control device 20 notifies the operator that the fraction remains. When the operator receives notification from the treatment planning device 2 that the fraction remains, the operator takes another blood sample from the patient and puts the collected blood into the testing device 4.

[0068] When the blood to be tested is introduced into the testing device 4, the number of ALCs in the blood y （ｍｅａｓ） Measure the number of ALCs y （ｍｅａｓ） The ALC measurement information, combined with the date of measurement, is stored in the storage device 5. The inspection device 4 transmits data update information to the treatment planning device 2 indicating that new ALC measurement information has been stored in the storage device 5. When the control device 20 receives the data update information from the inspection device 4, it retrieves the ALC measurement information from the storage device 5 and stores the ALC measurement information in the memory 12 (step S17).

[0069] On the other hand, if the determination in step S16 indicates that no fractions remain as planned, the control device 20 notifies the operator that no fractions remain. When the operator receives notification from the treatment planning device 2 that no fractions remain, they terminate treatment for the patient. The operator repeats radiation irradiation and ALC measurement for the patient until the planned number of fractions is completed. Note that the ALC measurement and radiation irradiation do not need to be performed on the same day or at the same time. Therefore, the operator may measure multiple ALCs between fractions, and does not necessarily need to measure ALCs between fractions.

[0070] In step S18, the control device 20 determines whether the measurement of ALC at the preset measurement point M' has been completed. If the measurement of ALC at the measurement point M' has not been completed, the control device 20 proceeds to step S15. If the determination in step S18 indicates that the measurement of ALC at the measurement point M' has been completed, the control device 20 proceeds to step S19. This is to perform adaptive radiotherapy. In step S19, the control device 20 stores the measured value y of ALC at the measurement point M' in the memory 12. （ｍｅａｓ） (t １ ) ~ y （ｍｅａｓ） (t Ｍ’ The parameters α and k of the ALC prediction model f are set to reproduce the above. ｒ The parameters α and k are readjusted. Specifically, the control device 20 updates the parameters α and k so that the measured value of ALC matches the predicted value of ALC. The control device 20 then readjusts the parameters α'' and k''. ｒ The control device 20 obtains the parameters α'' and k''. ｒ This is stored in memory 12. The parameters are updated so that the predicted values ​​match the measured values ​​of ALC, thus improving the estimation accuracy of ALC.

[0071] In step S20, the control device 20 uses a predictive model g of the parameter variation range. σα and g σｋｒ Measured value of ALC y （ｍｅａｓ） (t １ ) ~ y （ｍｅａｓ）(t Ｍ’ Enter the parameter variation range σ''. α and σ'' ｋｒ The parameter variation range σ'' is recalculated (step S20). The control device 20 then adjusts the parameter variation range σ'' α and σ'' ｋｒ This is stored in memory 12. As a result, the parameter variation range is estimated based on the measured values ​​of the ALC. Therefore, the accuracy of the prediction range is improved by the newly estimated parameter variation range.

[0072] After adjusting various parameters related to the ALC prediction model f, the control device 20 returns to the process in step S13. In step S13, the control device 20 uses the dose d obtained in step S12 and the newly adjusted parameters α'' and k'' ｒ Using this, the ALC prediction model f newly calculates the temporal progression of ALC. The control device 20 displays a graph showing the temporal progression of ALC on the output device 14.

[0073] Furthermore, the control device 20 controls the variation range σ'' of the newly adjusted parameters. α and σ'' ｋｒ Based on this, the control device 20 calculates the predicted range. The control device 20 displays the predicted range on the output device 14, overlaid on the graph showing the temporal progression of ALC. The control device 20 displays the ALC threshold, which is set in advance based on clinical experience, on the output device 14, overlaid on the graph showing the temporal progression of ALC. Furthermore, the control device 20 calculates the probability of achieving the threshold based on the newly calculated predicted range and threshold. The control device 20 displays the calculated probability of achieving the threshold together with the graph showing the temporal progression of ALC on the output device 14.

[0074] The control device 20 may control the output device 14 so that the previous prediction result and the latest prediction result are displayed on the same screen when displaying the latest calculation result on the output device 14. Figure 6 is a diagram showing an example of an image when the output device shown in Figure 1 displays the previous prediction result and the latest prediction result.

[0075] Figure 6 shows the initially determined parameter Ave[α ｎ ] and Ave[k（ｎ） ｒ ] and the initially determined parameter variation range σ α and σ ｋｒ Based on the above, the time progression and prediction range of the calculated ALC are shown side by side with the two prediction results. As shown in Figure 6, the display unit 14a of the output device 14 displays the two measurement results. The upper graph in Figure 6 shows the prediction results before the parameter update, and the lower graph in Figure 6 shows the prediction results after the parameter update. Focusing on the achievement probabilities of the two prediction results, the achievement probability after the parameter update is smaller than the achievement probability before the parameter update. In this case, the control device 20 may notify the operator of warning information prompting a review of the treatment. For example, if the control device 20 displays "Target achievement probability = 100%" in black, it may display "Target achievement probability = 80%" in red to draw the operator's attention. The notification of warning information to the operator may also be an audio output via a speaker. Because the operator is notified of information prompting a review of the treatment, they will not forget to review the treatment plan. In addition, the operator can judge the appropriateness of the parameter update by comparing the achievement probability before the parameter update with the achievement probability after the parameter update.

[0076] The operator refers to the information displayed on the output device 14 and, based on the displayed information, determines whether the treatment plan formulated in step S12 is successful or not. By referring to the two prediction results shown in Figure 6, the operator can compare the success probability before and after parameter adjustment. In the second and subsequent judgments in step S14, the operator can establish a criterion to pass the treatment plan if the success probability obtained after parameter adjustment is greater than the success probability before parameter adjustment.

[0077] In step S13, the control device 20 adjusts the parameters α'' and k'' of the ALC prediction model f. ｒ Use the newly adjusted parameter variation range σ''. α and σ'' ｋｒHowever, if the value increases significantly compared to the previous value, it means that the reliability of the adjusted model is not high. In this case, the control device 20 stops updating the parameters. The operator may continue treatment without updating the parameters. Since the ALC prediction result is the same as before parameter adjustment, the operator does not need to create a treatment plan again. This prevents predictions for ALC that deviate from the measured value from the actual value. The processing from step S13 onward is the same as described above, so a detailed explanation is omitted.

[0078] As explained with reference to Figure 3, the treatment planning device 2 updates its model parameters to reproduce the measured ALC values ​​obtained during the treatment period and then calculates the predicted ALC value again. Healthcare professionals periodically measure the patient's ALC to understand changes in the patient's condition during the treatment period. If the re-prediction reveals that the predicted ALC value is lower than initially expected, healthcare professionals can revise the treatment plan.

[0079] However, the measured ALC value does not necessarily decrease monotonically as the observation period progresses. Possible causes include not only measurement accuracy and radiation therapy, but also changes in the patient's physical condition. Therefore, parameter adjustment alone does not necessarily improve the accuracy of the ALC prediction model. In contrast, the information processing system 10 in this embodiment not only estimates the temporal progression of ALC, but also estimates the predicted range of ALC based on the parameter variation range. As a result, medical professionals can understand the range of error of ALC from the predicted range of ALC, and can utilize the predicted range of ALC in determining the treatment plan. Consequently, the burden on medical professionals is reduced, and the treatment throughput is improved.

[0080] In this embodiment, the parameters to be adjusted are α and k. ｒ Although we have explained this using the case of two parameters, the types of two parameters are α and k ｒ This is not limited to two parameters. Also, although the explanation used two parameters, the number of parameters is not limited to two. There may be one parameter, or three or more. The parameters to be adjusted are modified according to the predictive model being applied.

[0081] Furthermore, in each process described with reference to Figures 2 and 3, the control device 20 may store the information to be stored in the memory 12 in the storage device 5, or the information to be stored in the storage device 5 may be stored in the memory 12. Also, although this embodiment was described assuming that the treatment is radiation therapy and the clinical test value is ALC, this embodiment is not limited to radiation therapy. The information processing method of this embodiment may be applied to treatments other than radiation therapy. The information processing method of this embodiment is particularly effective for treatments in which individual differences affect the treatment plan.

[0082] The information processing system 10 of this embodiment includes a memory 12 for storing a program and a processor 11 for executing processing according to the program. The processor 11 operates as follows when executing the program: The processor 11 estimates the temporal changes in the patient's clinical test values ​​during treatment based on pre-set parameters. The processor 11 then estimates a predicted range, which is the range that the temporal changes in the estimated clinical test values ​​(predicted values) can take probabilistically, based on the range of parameter fluctuations.

[0083] According to this embodiment, the predicted range of clinical test values ​​in treatment is estimated based on the fluctuation range of a parameter that estimates the temporal changes in clinical test values. Healthcare professionals can understand the range of error in clinical test values ​​from the predicted range of clinical test values ​​and use the predicted range of clinical test values ​​to determine treatment plans. When the clinical test value is ALC, healthcare professionals do not need to estimate the error of ALC by their own judgment when determining the radiation dose by utilizing the predicted range of ALC. In addition, since the overestimation of the error of ALC is suppressed, the effectiveness of a single radiation therapy is improved and the number of radiation therapy sessions performed on a single patient is reduced. As a result, the burden on healthcare professionals is reduced and the treatment throughput is improved.

[0084] Furthermore, according to this embodiment, in radiation therapy that protects ALC, an ALC prediction model whose parameters are adjusted based on measured ALC values ​​is appropriately used in determining the treatment plan. As a result, the burden on medical personnel is further reduced and the treatment throughput is further improved.

[0085] The embodiments described above are illustrative for explaining the present invention and are not intended to limit the scope of the invention to those embodiments only. Those skilled in the art can implement the present invention in various other forms without departing from the scope of the invention.

[0086] Furthermore, the embodiments described above include the following items. However, the items included in these embodiments are not limited to those listed below.

[0087] (Item 1) A program for performing the following: estimating the temporal changes in a patient's clinical test values ​​during treatment based on parameters pre-set in the computer; and estimating a predicted range, which is the range that the estimated temporal changes in the predicted clinical test values ​​can take probabilistically, based on the range of variation of the parameters.

[0088] According to this method, the predicted range of clinical test values ​​in treatment is estimated based on the range of variation of parameters that estimate the temporal changes in clinical test values. Healthcare professionals can understand the range of error in clinical test values ​​from the predicted range of clinical test values ​​and use the predicted range of clinical test values ​​to determine treatment plans. This reduces the burden on healthcare professionals and improves treatment throughput.

[0089] (Item 2) A program for performing the following in the program described in Item 1: calculating the probability of achieving the target value based on the target value set for the clinical test value and the predicted range. According to this, an achievement probability useful as an indicator for determining the success or failure of a treatment plan can be obtained.

[0090] (Item 3) A program for updating the parameters in the program described in Item 2 so that the measured clinical test value matches the predicted value. According to this, the parameters are updated so that the predicted value matches the measured clinical test value, thus improving the estimation accuracy of the clinical test value.

[0091] (Item 4) A program to perform the following in the program described in Item 3: estimating the range of variation of the updated parameters based on the measured values. According to this, the range of variation of the updated parameters is estimated based on the measured values ​​of clinical test values. Therefore, the accuracy of the prediction range is improved by the newly estimated range of variation.

[0092] (Item 5) A program to perform the following in the program described in Item 4: to determine the achievement probability corresponding to the parameters and the range of variation before the update, and the achievement probability corresponding to the parameters and the range of variation after the update. By comparing the achievement probabilities before and after the update of the parameters and the range of variation, the appropriateness of updating the parameters and the range of variation can be determined.

[0093] (Item 6) A program to enable the program described in Item 5 to notify the patient of information prompting a review of treatment if the probability of achieving the parameter after updating the parameter is lower than the probability of achieving the parameter before updating the parameter. This ensures that healthcare professionals are notified of information prompting a review of treatment and will not forget to review the treatment plan.

[0094] (Item 7) A program to stop updating the parameter if the update of the parameter increases the range of variation in any one of Items 4 to 6. If the updated parameter increases the range of variation of the parameter, the reliability of the predicted value based on the updated parameter is not high, so the parameter update is stopped. This prevents predictions of clinical test values ​​from deviating from actual values.

[0095] (Item 8) A program that causes the program described in any one of Items 1 to 7 to display the predicted range, the target value set for the clinical test value, and the measured value which is the clinical test value on a display. This allows healthcare professionals to judge the validity of the treatment plan from the relationship between the predicted range, the target value, and the measured value by referring to the information displayed on the display.

[0096] (Item 9) A program for causing the program described in Item 8 to perform the following actions: calculate the probability of achieving the target value based on the target value and the predicted range, and display the probability of achievement on the display. This allows healthcare professionals to efficiently determine the success or failure of a treatment plan by referring to the probability of achievement displayed on the display.

[0097] 1 Adaptive radiotherapy system, 2 Treatment planning device, 3 Radiation irradiation device, 4 Examination device, 5 Storage device, 10 Information processing system, 11 Processor, 12 Memory, 13 Input device, 14 Output device, 14a Display unit, 15 Communication device, 16 Disk drive, 17 Bus, 20 Control device, 100 Network.

Claims

1. An information processing system comprising: a memory for storing a program; and a processor for executing processing according to the program, wherein the processor, by executing the program, estimates the temporal changes in a patient's clinical test values ​​during treatment based on pre-set parameters, and estimates a predicted range, which is the range that the temporal changes in the estimated clinical test values ​​can take probabilistically, based on the range of variation of the parameters.

2. An information processing method performed by an information processing device, comprising: estimating the temporal changes in a patient's clinical test values ​​during treatment based on pre-set parameters; and estimating a predicted range, which is the range that the temporal changes in the estimated clinical test values ​​can probabilistically take, based on the range of variation of the parameters.

3. A program for performing the following: estimating the temporal changes in a patient's clinical test values ​​during treatment based on parameters pre-set in the computer; and estimating a predicted range, which is the range that the estimated temporal changes in the predicted clinical test values ​​can probabilistically take, based on the range of variation of the parameters.

4. A program for causing the program according to claim 3 to perform the following: calculate the probability of achieving the target value based on the target value set for the clinical test value and the predicted range.

5. A program for updating the parameters in the program according to claim 4 so that the measured clinical test values ​​match the predicted values.

6. A program for causing the program according to claim 5 to estimate the range of variation of the updated parameter based on the measured value.

7. A program for causing the program according to claim 6 to perform the following: determining the achievement probability corresponding to the parameters and the range of variation before the update, and determining the achievement probability corresponding to the parameters and the range of variation after the update.

8. A program for causing the program according to claim 7 to perform the following actions: if the probability of achieving the parameter after updating the parameter is smaller than the probability of achieving the parameter before updating the parameter, the program provides information prompting a review of the treatment.

9. A program for causing the program according to claim 6 to stop updating the parameter if the update of the parameter increases the range of variation.

10. A program for causing the program according to claim 3 to display the predicted range, the target value set for the clinical test value, and the measured value which is the measured clinical test value on a display.

11. A program according to claim 10, which causes the program to calculate the probability of achieving the target value based on the target value and the predicted range, and to display the probability of achieving the target value on the display.

Images