Hospital bed allocation system based on probability measurement to construct disease difficulty index
The hospital bed allocation system constructed through probability measurement uses Bayesian method to predict the relative weight and the probability of fourth-level surgical treatment of patients, which solves the problems of delayed admission and unfair manual allocation of difficult and critical patients in traditional allocation strategies, and improves the utilization of medical resources rate and operational efficiency.
Patent Information
- Application Number
- CN202411154977.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-22
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-08-22
AI Technical Summary
Traditional hospital bed allocation strategies are prone to delay admission of difficult and critical patients, and manual subjective allocation has problems such as high labor costs, subjectivity and unfairness, resulting in low efficiency in medical resource utilization and unreasonable bed allocation.
A hospital bed allocation system is constructed based on probability measurements. By obtaining patient complaint information and diagnostic results, using Bayesian method to predict the patient's relative weight and the probability of fourth-level surgical treatment, we scientifically determine the priority of bed allocation.
It has achieved scientific allocation of bed resources based on the patient's condition and treatment difficulty, improved the utilization rate and operational efficiency of medical resources, reduced human resource waste, and ensured the fairness and rationality of bed allocation.
Smart Images

Figure CN118675721B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of hospital bed management, and specifically to a hospital bed allocation system that constructs a disease difficulty index based on probability measurement. Background Art
[0002] Under the bed management model of large-scale general hospitals, bed resources are uniformly allocated by the management department. Traditional hospital bed allocation strategies usually include obtaining the bed allocation priority of all patients waiting to be admitted based on the "first come, first served" allocation strategy, or subjectively judging the patients who need to be allocated beds first based on manual experience. Although the "first come, first served" allocation strategy has clear rules and is simple to operate, large-scale general hospitals usually have a large number of patients waiting for beds, with mixed diseases and varying degrees of difficulty. The "first come, first served" allocation strategy is likely to lead to a large number of difficult and critical patients being delayed in admission, which in turn leads to a decrease in the utilization efficiency of high-quality medical resources. Although subjective allocation based on manual experience can classify patients who need to be admitted in advance, give priority to allocating beds to difficult and critical patients, and improve the utilization rate of medical resources, this method has a high labor cost, and the allocation rules are vague. There is subjectivity and unfairness in the allocation process, which leads to waste of human resources and unreasonable bed allocation. Summary of the invention
[0003] The purpose of the present invention is to solve the technical problem of how to reasonably allocate beds in medical institutions, and to provide a hospital bed allocation system that constructs a disease difficulty index based on probability measurement, so as to achieve the technical effect of allocating beds according to patients' admission needs and treatment priorities under the background of coordinated bed management, improve the utilization rate of medical resources and the operating efficiency of medical institutions, promote large-scale comprehensive hospitals to implement their functional positioning, and guide the goal of hierarchical diagnosis and treatment.
[0004] The present invention claims protection for a hospital bed allocation system that constructs a disease difficulty index based on probability measurement, comprising:
[0005] A first acquisition module, the first acquisition module includes a first acquisition submodule, a second acquisition submodule and a first processing submodule; wherein the first acquisition submodule is used to acquire the chief complaint information of the target patient to obtain the target chief complaint information, the second acquisition submodule is used to acquire the diagnosis result of the target patient to obtain the target diagnosis information, and the first processing submodule is used to obtain the target feature information of the target patient according to the target chief complaint information and the target diagnosis information;
[0006] A first prediction module, the first prediction module is used to predict the relative weight of the target patient using a Bayesian method according to the target feature information to obtain a target relative weight;
[0007] A second prediction module, the second prediction module is used to predict the probability of the target patient receiving level 4 surgery during hospitalization by using a Bayesian method according to the target feature information, to obtain a target surgery probability;
[0008] The first processing module is used to obtain the bed allocation priority rank of the target patient according to the target relative weight and the target surgery probability. The target patient with a larger target relative weight has a higher priority bed allocation rank, and the target patient with a larger target surgery probability has a higher priority bed allocation rank.
[0009] In one embodiment of the present application, the target characteristic information includes a target severity level, and the first processing submodule also includes a first prediction submodule, which is used to predict the severity level of the target patient's condition using a Bayesian method based on the target chief complaint information and the target diagnosis information to obtain the target severity level.
[0010] In one embodiment of the present application, the target feature information also includes a target expected effect, and the first processing submodule also includes a second prediction submodule, which is used to predict the treatment effect of the target patient at the target time point using a Bayesian method based on the target chief complaint information and the target diagnosis information to obtain the target expected effect.
[0011] In one embodiment of the present application, the second prediction module further includes:
[0012] Predicting the primary outcome of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain the target outcome;
[0013] Predicting whether the target patient will experience an adverse event at the target time point according to the target chief complaint information and the target diagnosis information, and obtaining a target event;
[0014] Predicting the duration of the target patient's stay in the ICU at the target time point according to the target chief complaint information and the target diagnosis information, and obtaining a target ICU duration;
[0015] Predicting the hospitalization duration of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain a target hospitalization duration;
[0016] Predicting the treatment cost of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain a target cost;
[0017] Predicting the satisfaction evaluation of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain a target satisfaction evaluation;
[0018] The target expected effect of the target patient is predicted using the Bayesian method based on the target outcome, the target event, the target ICU duration, the target hospitalization duration, the target cost and the target satisfaction evaluation.
[0019] In one embodiment of the present application, the target characteristic information also includes target resource usage, and the third prediction module is used to predict the resource usage of the target patient during hospitalization using a Bayesian method based on the target chief complaint information and the target diagnosis information to obtain the target resource usage.
[0020] In an embodiment of the present application, the third prediction module further includes:
[0021] Predicting the usage of different target medical resources by the target patient according to the target chief complaint information and the target diagnosis information, and obtaining corresponding target resource usage sub-amounts; wherein the target medical resources include medical resources other than bed resources;
[0022] Calculate the average medical resource usage of the target patient according to each target resource usage sub-amount of the target patient to obtain a predicted resource usage;
[0023] querying a database according to the disease code of the target diagnosis information to obtain a benchmark usage amount, wherein the benchmark usage amount is obtained according to an average of resource usage amounts of all historical patients having the same disease code as the target diagnosis information;
[0024] Obtaining the resource usage level of the target patient according to the difference between the target resource usage and the benchmark usage;
[0025] The target resource usage level of the target patient is predicted using a Bayesian method according to the resource usage level of the target patient to obtain the target resource usage.
[0026] In one embodiment of the present application, the target characteristic information also includes a target waiting degree, the first acquisition module also includes a third acquisition submodule, the third acquisition submodule is used to obtain the target patient's waiting time for admission to obtain the actual waiting time, the first processing submodule also includes a fourth prediction submodule, the fourth prediction submodule is used to estimate the target patient's waiting time for admission to the hospital based on the target chief complaint information and the target diagnosis information using a Bayesian method to obtain the target waiting time; the target waiting degree of the target patient is obtained based on the comparison result of the target waiting time and the actual waiting time.
[0027] In one embodiment of the present application, the system also includes a second acquisition module and a second processing module, the second acquisition module is used to acquire the allocatable amount of the target medical resources to obtain a reference resource amount, and the second processing module is used to determine whether each patient to be admitted meets the preset allocation conditions in order of bed allocation priority from high to low, and the preset allocation conditions include that the current patient's expected usage of the target medical resources is not greater than the corresponding reference resource amount; if the current patient meets the preset allocation conditions, the current patient's bed application information is included in the allocation queue, and the reference resource amount is updated according to the current patient's expected usage of each target medical resource; if the current patient does not meet the preset allocation conditions, the current patient's bed application information is included in the waiting queue, and the current patient is updated according to the priority order of bed allocation until there are no allocatable bed resources or the bed allocation results of all patients to be admitted are obtained.
[0028] In one embodiment of the present application, the second processing module further includes:
[0029] If the number of patients in the current priority position who meet the preset allocation conditions is greater than 1, and the current reference resource amount cannot simultaneously meet the needs of all patients who meet the preset allocation conditions, the target medical resource with an expected usage amount not equal to 0 corresponding to the patients who meet the preset allocation conditions shall be used as the first candidate medical resource, and according to the classification rules of medical resources, the first candidate medical resource with the highest level shall be used as the second candidate medical resource for the corresponding patient, and the bed application information of the patient corresponding to the second candidate medical resource with the highest level shall be included in the allocation queue. After updating the reference resource amount, it shall be re-judged whether other patients corresponding to the current priority position meet the preset allocation conditions.
[0030] This application has the following beneficial effects:
[0031] 1. The RW value reflects the resource consumption degree of different DRG groups. The resource consumption degree of the DRG group is related to the severity, complexity, difficulty of treatment, recovery time and treatment risk of the patient's condition. The more severe or complex the patient's condition, the greater the difficulty of treatment, the longer the recovery time and the higher the risk, the larger the corresponding RW value. According to the risk degree, difficulty, resource consumption degree or ethical risk of the operation, operations with high risk, complex process, high difficulty, high resource consumption or involving major ethical risks are divided into four levels of operations. Level 4 operations are often used to treat patients with more serious conditions. If the probability of the target patient being treated with level 4 surgery is greater, it means that the target patient's condition is more serious and a bed should be allocated to him first. Therefore, by predicting the RW value of the DRG group to which the patient to be admitted belongs and the probability of using level 4 surgery for treatment based on the target chief complaint information and target diagnosis information, the bed allocation priority of all patients to be admitted is obtained according to the prediction results, thereby realizing scientific allocation of bed resources according to the patient's condition and treatment difficulty.
[0032] 2. Since the DRG grouping standard usually needs to determine the patient's DRG grouping results based on multiple factors such as the patient's diagnosis results, comorbidities and complications during the treatment process, and then obtain the patient's corresponding relative weight based on the patient's DRG grouping results. However, in actual application scenarios, as the patient's condition develops, comorbidities and complications do not appear at the time of admission. Directly obtaining the patient's DRG grouping results based on the patient's preliminary diagnosis results and symptoms before admission is prone to inaccurate grouping, which in turn affects the patient's RW value. The Bayesian method estimates the RW value of the target patient based on the prior knowledge of historical hospitalized patients and the target patient's chief complaint information and diagnosis results. In the estimation process, the probability model is used to model the uncertainty in the historical data. The probability model is used to incorporate the unknown changes of the target patient during the treatment process into the estimation basis, and the RW value of the target patient is accurately estimated. Similarly, due to the influence of recovery speed and other factors during the treatment process, the probability of the patient adopting level 4 surgical treatment during the treatment process will change. The probability model is established by the Bayesian method, and the unknown changes are incorporated into the estimation basis to achieve an accurate estimation of the probability of the target patient adopting level 4 surgical treatment.
[0033] 3. Based on the patient's chief complaint information, diagnosis information and severity of the disease in the historical admission records, the Bayesian method is used to evaluate the severity of the target patient's disease, and the target patient's RW value and the probability of using level 4 surgical treatment are predicted based on the target severity, so as to prioritize the allocation of beds to patients with more severe conditions.
[0034] 4. Obtain an estimate of the waiting time for hospitalization of historical patients whose chief complaint information is the same as the target chief complaint and whose diagnosis information is the same as the target diagnosis information, and compare the estimated result with the actual waiting time to obtain the target waiting level of the target patient. According to the target waiting level, predict the RW value of the target patient and the probability of adopting level 4 surgical treatment. On the one hand, under the condition that other conditions are the same, beds are allocated preferentially to patients with higher target waiting levels. On the other hand, the higher the waiting level of the target patient, the greater the risk of worsening or delay of the patient's condition, and the corresponding RW value and the probability of adopting level 4 surgical treatment will also increase, so as to achieve accurate prediction of the patient's RW value and the probability of adopting level 4 surgical treatment.
[0035] 5. According to the historical patient admission information, the target outcome, the target event, the target ICU duration, the target hospitalization duration, the target cost and the target satisfaction evaluation are predicted respectively, and the expected treatment effect of the target patient is estimated according to the prediction results, so as to provide a reference for the formulation of the patient's hospitalization treatment plan while realizing the prediction of the target patient's hospitalization treatment effect.
[0036] 6. Predict the usage of all types of medical resources required by the patient based on the target chief complaint information and the target diagnosis information. When the usage of medical resources is the same, a higher level of medical resource usage indicates that the patient's treatment is more difficult and / or the condition is more serious and / or the probability of a difficult and complicated disease is greater. The corresponding RW value of the patient and the probability of using level 4 surgery are greater, so as to predict the amount and type of resources that the patient will use during hospitalization.
[0037] 7. Compare the patient's expected resource usage with the reference resource usage according to the bed allocation priority, and obtain the patient's bed allocation result based on the comparison result, so as to allocate beds to patients according to the allocation priority based on the medical resources that can be allocated by the medical institution. For patients with abnormal bed allocation results, prompt information such as waiting for admission or recommending transfer will be adopted to improve the utilization rate of medical resources in medical institutions.
[0038] 8. When allocating beds for each patient to be admitted, priority will be given to matching beds with patients who use higher-level medical resources, thereby achieving the goal of improving the effective utilization rate of advanced medical resources in medical institutions while promoting tiered diagnosis and treatment. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.
[0040] Figure 1A schematic diagram of the structure of a hospital bed allocation system for constructing a disease difficulty index based on probability measurement according to an embodiment of the present application;
[0041] Figure 2 A schematic diagram of a method for obtaining a bed allocation priority order involved in an embodiment of the present application;
[0042] Figure 3 A schematic diagram of a method for obtaining bed allocation results according to an embodiment of the present application;
[0043] Figure 4 A schematic diagram of the structure of an electronic device involved in an embodiment of the present application;
[0044] Symbols in the figure: q-priority order, Q-number of patients waiting to be admitted. DETAILED DESCRIPTION
[0045] The present invention provides a hospital bed allocation system for constructing a disease difficulty index based on a probability measure. In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the technical scheme in the embodiment of the present application will be clearly and completely described in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, with reference to the terms "one embodiment", "some embodiments", "implementation", "embodiment", "illustrative embodiment", "example", "specific example" or "some examples", the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application for protection, but only indicates that the specific features, structures or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. And the specific features, structures or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.
[0046] It should be noted that similar reference numerals and letters represent similar items in the following figures, so once an item is defined in one figure, it does not need to be further defined and explained in the subsequent figures. At the same time, in the description of this application, the terms "first", "second" and other relational terms are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.
[0047] The present invention claims protection for a hospital bed allocation system that constructs a disease difficulty index based on probability measurement, referring to the attached Figure 1As shown, it includes a first acquisition module, a first prediction module, a second prediction module and a first processing module.
[0048] The first acquisition module includes a first acquisition submodule, a second acquisition submodule and the first processing submodule. The first acquisition submodule is used to acquire the main complaint information of the target patient to obtain the target main complaint information. The main complaint information includes the description information of the patient's symptoms by the patient and / or other accompanying personnel, such as the patient himself and / or the patient's family members describing "recent chest tightness and difficulty breathing" and "headache lasting for a week", etc., of course, it is not limited to this.
[0049] The second acquisition submodule is used to acquire the diagnosis result of the target patient to obtain target diagnosis information. The diagnosis result includes the doctor's diagnosis result of the patient's condition, such as the name of the disease, the cause of the disease, and the severity of the condition, etc., but it is not limited to this.
[0050] The first processing submodule is used to obtain the target feature information of the target patient according to the target chief complaint information and the target diagnosis information. The target chief complaint information is converted into a word vector to obtain a target chief complaint vector. A natural language processing algorithm is used to perform a word segmentation operation on the target chief complaint information to obtain a first word segmentation result; interference words in the first word segmentation result are removed according to a stop word list, such as adverbs, function words, punctuation marks, and ultra-low frequency words, to obtain a second word segmentation result; a neural network model is used according to one-hot embedding technology to convert the second word segmentation result into a word vector to obtain the target chief complaint vector. The target feature information is obtained according to the target chief complaint vector and the target diagnosis information.
[0051] The first prediction module is used to predict the relative weight of the target patient according to the target chief complaint vector and the target diagnosis information using a Bayesian method to obtain a target relative weight.
[0052] It should be noted that the relative weight is the RW value, which is used to represent the comparison value between the average resource consumption of historical patients in the DRG group to which the patient belongs and the average resource consumption of all historical patients. According to the "National Medical Insurance DRG Grouping and Payment Technical Specifications" and the "National Medical Insurance DRG (CHS-DRG) Grouping Scheme", all historical patients are grouped to obtain corresponding DRG groups. According to the hospitalization expenses of all historical patients, the average hospitalization expenses of historical patients in each DRG group and the average hospitalization expenses of all historical patients are calculated respectively. The RW value of the DRG group is calculated based on the average hospitalization expenses of each DRG group and the average hospitalization expenses of all historical patients.
[0053] In this embodiment, the RW values of all DRG groups are obtained, and the probability that the RW value of the target patient is equal to the RW value of each DRG group is calculated according to the Bayesian method, and the RW value corresponding to the maximum probability is taken as the target relative weight of the target patient.
[0054] It should be noted that the probability RW of the target patient's RW value equals the RW value of a certain DRG group i The calculation methods include:
[0055] ;
[0056] Among them, word represents the target complaint vector, and class represents the target diagnosis information. It represents the class prior probability of historical patients belonging to the i-th DRG group, which can be calculated according to the proportion of the number of historical patients in the i-th DRG group to the number of all historical patients. The conditional probability that the chief complaint vector of the historical patients in the i-th DRG group is the same as the target chief complaint vector can be calculated according to the proportion of the number of historical patients whose chief complaint vector is word in the i-th DRG group to the number of all historical patients in the DRG group. The conditional probability that the diagnosis results of historical patients in the i-th DRG group are the same as the target diagnosis information can be calculated based on the proportion of the number of historical patients with the diagnosis results of class in the i-th DRG group to the number of all historical patients in the DRG group.
[0057] It should be noted that the similarity between the chief complaint information of each historical patient and the target chief complaint information is calculated. If the similarity is greater than a preset threshold, the chief complaint information of the corresponding historical patient is regarded as the same as the target chief complaint information. The chief complaint information of the target patient is matched with the chief complaint information of the historical patient by similarity to adapt different expressions of the same expression content. The calculation is based on the proportion of the number of historical patients in each DRG group whose chief complaint vector has a similarity with the target chief complaint vector greater than a preset threshold to the number of all historical patients in the DRG group. In this embodiment, the preset threshold is 90%.
[0058] The second prediction module is used to predict the probability of the target patient receiving level 4 surgery during hospitalization using a Bayesian method according to the target chief complaint vector and the target diagnosis information, so as to obtain a target surgery probability.
[0059] It should be noted that level 4 surgeries include surgeries that are high-risk, complex, difficult, resource-intensive, or involve major ethical risks. According to the "Regulations on the Grading of Surgery in Medical Institutions," historical patients' surgeries are graded, and historical patients are classified according to whether level 4 surgery is used. Different categories correspond to different probability levels of level 4 surgery. For example, patients who do not undergo level 4 surgery are classified into a low-probability level 4 surgery probability level, and patients who undergo level 4 surgery are classified into a high-probability level 4 surgery probability level. The method for calculating the target surgery probability H includes:
[0060] ;
[0061] ;
[0062] Among them, h1 represents one of the four surgical probability levels among all four surgical probability levels h, i.e., low probability or high probability; It represents the proportion of historical patients with a level 4 surgical probability of h1 among all historical patients; The ratio of historical patients whose chief complaint information is the same as the target chief complaint information among historical patients whose fourth-level surgery probability level is h1 can be calculated according to the ratio of the number of historical patients whose chief complaint vectors among historical patients whose fourth-level surgery probability level is h1 and whose similarity with the target chief complaint vector is greater than the preset threshold to the number of all historical patients whose fourth-level surgery probability level is h1; It represents the proportion of historical patients whose diagnosis results are the same as the target diagnosis information among historical patients with the fourth-level surgery probability level h1.
[0063] The first processing module is used to obtain the bed allocation priority rank of the target patient according to the target relative weight and the target surgery probability. The target patient with a larger target relative weight has a higher priority bed allocation rank, and the target patient with a larger target surgery probability has a higher priority bed allocation rank.
[0064] In this embodiment, the first processing module includes a first evaluation submodule, which is used to evaluate the target relative weight to obtain a first evaluation subresult; evaluate the target surgical probability to obtain a second evaluation subresult; and calculate the evaluation result of the target patient based on the first evaluation subresult and the second evaluation subresult.
[0065] It should be noted that the scoring rules or grading standards can be set according to the actual application scenario, and the relative weight and the target surgery probability can be evaluated respectively according to the evaluation rules. For example, if the larger the value of the evaluation result, the higher the priority of the target patient's bed allocation, then the target patient with a larger value of the target relative weight will use the first evaluation sub-result with a larger value, and the target patient with a larger probability of the target surgery will use the second evaluation sub-result with a larger value; if the smaller the value of the evaluation result, the higher the priority of the target patient's bed allocation, then the target patient with a larger value of the target relative weight will use the first evaluation sub-result with a smaller value, and the target patient with a larger probability of the target surgery will use the second evaluation sub-result with a smaller value.
[0066] It should be noted that the linear weighted result of the first evaluation sub-result and the second evaluation sub-result may be calculated by setting weights to obtain the evaluation result of the target patient.
[0067] In this embodiment, the calculation method of the evaluation result is described in detail by taking the scoring rule that the larger the value of the evaluation result is, the higher the bed allocation priority of the target patient is as an example:
[0068] ;
[0069] ;
[0070] ;
[0071] ;
[0072] Wherein, λ1 represents the weight of the first evaluation sub-result, s1 represents the first evaluation sub-result, λ2 represents the weight of the second evaluation sub-result, s2 represents the second evaluation sub-result, rw represents the target relative weight, max(rw1, rw2, ..., rw i ,…,rw n ) means taking the maximum value of RW values in all DRG groups, rw i represents the RW value of the i-th DRG group, n represents the number of DRG groups, Indicates the preset score corresponding to the low probability level 4 surgery probability level, Indicates the preset score corresponding to the four-level surgical probability level with high probability.
[0073] It should be noted that the weight of the first evaluation sub-result and the weight of the second evaluation sub-result can be preset according to the relative weight and the importance of the probability of adopting level 4 surgical treatment. In this embodiment, the weight of the first evaluation sub-result and the weight of the second evaluation sub-result are both 0.5.
[0074] The first processing module is used to obtain the bed allocation priority rank of the target patient according to the target relative weight and the target surgery probability. The target patient with a larger target relative weight has a higher priority bed allocation rank, and the target patient with a larger target surgery probability has a higher priority bed allocation rank.
[0075] In this embodiment, the bed allocation priority ranking of the target patient among all patients to be admitted is obtained according to the evaluation result. Beds are allocated to patients to be admitted according to the bed allocation priority based on the allocatable bed quantity of the medical institution. If the priority ranking of the target patient is not greater than the allocatable bed quantity of the medical institution, the bed application of the target patient is included in the allocation queue; if the priority ranking of the target patient is greater than the allocatable bed quantity of the medical institution, the bed application of the target patient is included in the waiting queue. Beds are allocated to patients according to the allocation queue.
[0076] It should be noted that the RW value reflects the resource consumption degree of different DRG groups. The resource consumption degree of the DRG group is related to the difficulty and criticality of the patient. The larger the RW value of the patient, the greater the difficulty and / or criticality of the patient's condition, that is, the more serious or complex the condition, the greater the difficulty of treatment, the longer the recovery time, and the higher the risk. Level 4 surgery is usually used to treat patients with more serious conditions. If the probability of the target patient being treated with level 4 surgery is greater, it means that the target patient's condition is more serious and the treatment is more difficult, and beds should be allocated to them first. Therefore, by predicting the RW value of the DRG group to which the patient to be admitted belongs and the probability of using level 4 surgery for treatment based on the target chief complaint information and target diagnosis information, the bed allocation priority of all patients to be admitted is obtained according to the prediction results, thereby realizing scientific allocation of bed resources according to the difficulty of the patient. At the same time, since the DRG grouping standard usually needs to determine the patient's DRG grouping results based on multiple factors such as the patient's diagnosis results, comorbidities and complications during treatment, and then the relative weight corresponding to the patient is obtained according to the patient's DRG grouping results. However, in actual application scenarios, as the patient's condition progresses, comorbidities and complications do not appear at the time of admission. Directly obtaining the DRG grouping results of the patient based on the patient's preliminary diagnosis results and symptoms before admission is prone to inaccurate grouping, which in turn affects the patient's RW value. The Bayesian method estimates the RW value of the target patient based on the prior knowledge of historical hospitalized patients and the target patient's chief complaint information and diagnosis results. In the estimation process, the probability model is used to model the uncertainty in the historical data. The probability model is used to incorporate the unknown changes of the target patient during the treatment process into the estimation basis, thereby improving the accuracy of the estimation of the target patient's RW value. Similarly, due to changes in the patient's condition during the treatment process, the probability of the patient undergoing level 4 surgery will change during the treatment process. The probability model is established through the Bayesian method, and the unknown changes are incorporated into the estimation basis, thereby improving the accuracy of the estimation of the probability of the target patient undergoing level 4 surgery.
[0077] In a feasible embodiment, the target characteristic information includes a target severity level, and the first processing submodule also includes a first prediction submodule, which is used to predict the severity level of the target patient's condition using a Bayesian method based on the target chief complaint information and the target diagnosis information to obtain the target severity level.
[0078] It should be noted that the severity of the patient's condition can be scored, or the corresponding severity level can be matched according to the severity of the patient's condition, or other methods can be used. If the severity of the patient's condition is obtained by scoring, the Bayesian method can be used to regress and predict the target severity based on the target chief complaint information and the target diagnosis information; if the severity of the patient's condition is obtained by level classification, the Bayesian classifier can be used to obtain the target patient's severity level based on the target chief complaint information and the target diagnosis information for classification prediction.
[0079] In this embodiment, historical patients are divided into four severity levels according to the severity of their illness. A naive Bayes classifier is used to predict the severity of the target patient's illness, and the calculation method of the target severity level includes:
[0080] ;
[0081] ;
[0082] Among them, c1 represents a critical level among all critical levels y1. represents the conditional probability that the target patient belongs to the critical condition level c1, The function is used to obtain the severity level of the disease corresponding to the maximum value of the conditional probability. The class prior probability of historical patients with a critical condition level of c1 can be calculated according to the proportion of historical patients with a critical condition level of c1 in all historical patients. The conditional probability that the chief complaint vector of historical patients with a critical illness level of c1 is the same as the target chief complaint vector can be calculated based on the proportion of historical patients whose chief complaint vectors of historical patients with a critical illness level of c1 have a similarity with the target complaint vector greater than the preset threshold. The conditional probability that the diagnosis results of historical patients with a critical condition level of c1 are the same as the target diagnosis information can be calculated based on the proportion of the number of historical patients with a diagnosis result of class among the historical patients with a critical condition level of c1.
[0083] In this embodiment, the first prediction module is used to predict the target relative weight using the Bayesian method according to the target severity. The probability that the RW value of the target patient is equal to the RW value corresponding to the DRG group is calculated based on the class prior probability of the RW value corresponding to each DRG group in the historical patients, the conditional probability of the historical patients in the corresponding DRG group with the same chief complaint information as the target chief complaint information, and the conditional probability of the historical patients in the corresponding DRG group with the same diagnosis information as the target diagnosis information, and the RW value corresponding to the maximum probability is taken as the target relative weight.
[0084] In this embodiment, the second prediction module is used to predict the target surgical probability using the Bayesian method according to the target criticality. The probability that the target patient belongs to each probability category is calculated based on the class prior probability of the RW value corresponding to the probability category of the historical patients who adopted the fourth-level surgical treatment, the conditional probability of the historical patients whose chief complaint information in the corresponding probability category is the same as the target chief complaint information, and the conditional probability of the historical patients whose diagnosis information in the corresponding probability category is the same as the target diagnosis information, and the RW value corresponding to the maximum probability is taken as the target surgical probability.
[0085] It should be noted that the Bayesian method is used to evaluate the severity of the target patient's condition based on the patient's chief complaint information, diagnosis information and severity of the condition in the historical admission records, and the target patient's RW value and the probability of using level 4 surgical treatment are predicted based on the target severity of the condition, so as to prioritize the allocation of beds to patients with more severe conditions.
[0086] In a feasible embodiment, the target feature information also includes a target expected effect, and the first processing submodule also includes a second prediction submodule, which is used to predict the treatment effect of the target patient at the target time point using a Bayesian method based on the target chief complaint information and the target diagnosis information to obtain the target expected effect.
[0087] It should be noted that the treatment effect of the patient can be scored, or the corresponding treatment effect level can be matched according to the treatment effect of the patient, or other methods can be used. If the treatment effect of the patient is scored, the Bayesian method can be used to regress and predict the target expected effect according to the target chief complaint information and the target diagnosis information; if the treatment effect of the patient is evaluated by grade classification, the Bayesian classifier can be used to classify and predict the expected treatment effect of the target patient according to the target chief complaint information and the target diagnosis information.
[0088] In this embodiment, historical patients are divided into two categories according to the treatment effect: good treatment effect and poor treatment effect. A naive Bayes classifier is used to predict the expected treatment effect of the target patient, and the calculation method of the target expected effect includes:
[0089] ;
[0090] ;
[0091] Among them, c2 represents a treatment effect category among all treatment effect categories y2, that is, good treatment effect or poor treatment effect; represents the conditional probability that the target patient belongs to treatment effect category c2, The function is used to obtain the treatment effect category corresponding to the maximum conditional probability. The class prior probability of historical patients with treatment effect category c2 can be calculated according to the proportion of historical patients with treatment effect category c2 in all historical patients. The conditional probability that the chief complaint vector of historical patients with treatment effect category c2 is the same as the target chief complaint vector can be calculated based on the proportion of historical patients whose chief complaint vector of historical patients with treatment effect category c2 has a similarity with the target complaint vector greater than the preset threshold. The conditional probability that the diagnosis results of historical patients with treatment effect category c2 are the same as the target diagnosis information can be calculated based on the proportion of the number of historical patients with diagnosis results of class among historical patients with treatment effect category c2.
[0092] In this embodiment, the first prediction module is used to predict the target relative weight according to the target criticality and the target expected effect using the Bayesian method. The specific prediction method and predicting the target relative weight according to the target criticality, the target expected effect, the target resource usage and the target waiting time will not be described in detail here. The second prediction module is used to predict the target surgical probability according to the target criticality and the target expected effect using the Bayesian method. The specific prediction method and predicting the target surgical probability according to the target criticality, the target expected effect, the target resource usage and the target waiting time will not be described in detail here.
[0093] It should be noted that the probability that the expected treatment effect of the target patient is a good treatment effect and the probability that the treatment effect is a poor treatment effect are calculated respectively, and the treatment effect category corresponding to the maximum probability is taken as the expected treatment effect of the target patient to obtain the target expected effect.
[0094] In a feasible embodiment, the target characteristic information also includes target resource usage, and the first processing submodule also includes a third prediction submodule, which is used to predict the resource usage of the target patient during hospitalization using a Bayesian method based on the target chief complaint information and the target diagnosis information to obtain the target resource usage.
[0095] It should be noted that the Bayesian method can be used to perform regression prediction on the resource usage based on the target chief complaint information and the target diagnosis information; the Bayesian classifier can also be used to obtain the resource usage level of the target patient based on the target chief complaint information and the target diagnosis information for classification prediction.
[0096] In this embodiment, historical patients are divided into four resource utilization levels according to resource usage. A naive Bayes classifier is used to predict the resource utilization level of the target patient, and the calculation method of the target expected effect includes:
[0097] ;
[0098] ;
[0099] Wherein, c3 represents a resource usage level among all resource usage levels y3; represents the conditional probability that the target patient belongs to resource utilization level c3, The function is used to obtain the resource usage level corresponding to the maximum conditional probability. The class prior probability of historical patients with resource utilization level c3 can be calculated according to the proportion of historical patients with resource utilization level c3 in all historical patients. The conditional probability that the chief complaint vector of historical patients with resource utilization level c3 is the same as the target chief complaint vector can be calculated according to the proportion of the number of historical patients with resource utilization level c3 whose chief complaint vector has a similarity with the target accommodation vector greater than the preset threshold. The conditional probability that the diagnosis results of historical patients with resource usage level c3 are the same as the target diagnosis information can be calculated according to the proportion of the number of historical patients with diagnosis results of class among the historical patients with resource usage level c3.
[0100] In this embodiment, the first prediction module is used to predict the target relative weight using the Bayesian method based on the target criticality, the target expected effect, and the target resource usage. The specific prediction method is similar to predicting the target relative weight based on the target criticality, the target expected effect, the target resource usage, and the target waiting time, and will not be described in detail here. The second prediction module is used to predict the target surgical probability using the Bayesian method based on the target criticality, the target expected effect, and the target resource usage. The specific prediction method is similar to predicting the target surgical probability based on the target criticality, the target expected effect, the target resource usage, and the target waiting time, and will not be described in detail here.
[0101] It should be noted that the probability that the resource usage of the target patient belongs to each resource usage level is calculated respectively, and the resource usage level corresponding to the maximum probability is taken as the resource usage level of the target patient to obtain the target resource usage.
[0102] In a possible implementation, refer to the attached Figure 2 As shown, the target characteristic information also includes a target waiting degree, the first acquisition module also includes a third acquisition submodule, the third acquisition submodule is used to obtain the target patient's waiting time for admission to obtain the actual waiting time, the first processing submodule also includes a fourth prediction submodule, the fourth prediction submodule is used to estimate the target patient's waiting time for admission to the hospital according to the target chief complaint information and the target diagnosis information using the Bayesian method to obtain the target waiting time; the target waiting degree of the target patient is obtained according to the comparison result of the target waiting time and the actual waiting time.
[0103] It should be noted that the waiting time for hospital admission can be estimated by regression based on the Bayesian method according to the target chief complaint information and the target diagnosis information.
[0104] It should be noted that the target waiting degree can be obtained by setting a threshold value. For example, if the actual waiting duration is not greater than 80% of the target waiting duration, the target waiting degree is a low duration level, if the actual waiting duration is between 80%-120% of the target waiting duration, the target waiting degree is a medium duration level, and if the actual waiting duration is not less than 120% of the target waiting duration, the target waiting degree is a high duration level. The target waiting degree can also be obtained according to the ratio of the actual waiting duration to the target waiting duration.
[0105] It should be noted that, by obtaining the estimated waiting time for hospitalization of historical patients whose chief complaint information is the same as the target chief complaint and whose diagnosis information is the same as the target diagnosis information, and comparing the estimated result with the actual waiting time, the target waiting degree of the target patient is obtained, and the RW value of the target patient and the probability of adopting level 4 surgical treatment are predicted according to the target waiting degree. On the one hand, under the condition that other conditions are the same, beds are allocated preferentially to patients with higher target waiting degrees. On the other hand, the higher the waiting degree of the target patient, the greater the risk of worsening or delay of the patient's condition, and the corresponding RW value and the probability of adopting level 4 surgical treatment will also increase, so as to improve the accuracy of predicting the patient's RW value and the probability of adopting level 4 surgical treatment.
[0106] In this embodiment, the first prediction module predicts the target relative weight of the target patient according to the target severity, the target expected effect, the target resource usage and the target waiting time. The calculation method of the target relative weight includes:
[0107] ;
[0108] ;
[0109] Where rw represents a value of RW in the set R consisting of the RW values of all DRG groups. It indicates the proportion of historical patients whose RW value is rw among all historical patients. It indicates the proportion of historical patients whose critical illness level is the same as the target critical illness level among historical patients whose RW value is rw. It indicates the proportion of historical patients whose RW value is rw in the treatment effect category that is the same as the target expected effect. It indicates the proportion of historical patients whose resource usage level is the same as the target resource usage among historical patients whose RW value is rw. It indicates the proportion of historical patients whose admission waiting time is the same as the target waiting time among historical patients whose RW value is rw.
[0110] In this embodiment, the second prediction module predicts the target surgical probability of the target patient based on the target severity, the target expected effect, the target resource usage and the target waiting time. Historical patients are classified according to whether or not level 4 surgical treatment is adopted. Different categories correspond to different probability levels of adopting level 4 surgical treatment. For example, patients who do not adopt level 4 surgical treatment are classified into a low probability level of level 4 surgical probability, and patients who adopt level 4 surgical treatment are classified into a high probability level of level 4 surgical probability. The method for calculating the target surgical probability includes:
[0111] ;
[0112] ;
[0113] Among them, h1 represents one of the four surgical probability levels among all four surgical probability levels h, namely, low probability and high probability; It represents the proportion of historical patients with a probability level of h1 for level 4 surgery among all historical patients. It represents the proportion of historical patients with the same critical condition as the target critical condition among historical patients with the probability level of fourth-level surgery h1, It represents the proportion of historical patients with the same treatment effect category as the target expected effect among historical patients with the fourth-level surgery probability level h1, It represents the proportion of historical patients with the same resource usage level as the target resource usage among historical patients with the probability level of level 4 surgery h1, It represents the proportion of historical patients with a level 4 surgery probability level h1 whose hospital admission waiting time is the same as the target waiting time.
[0114] In this embodiment, the target relative weight and the target surgical probability are scored respectively, and the scoring results are linearly weighted and then calculated to obtain the evaluation result of the target patient.
[0115] In a feasible implementation manner, the second prediction module further includes:
[0116] Predict the main outcome of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain the target outcome; predict whether the target patient will have adverse events at the target time point according to the target chief complaint information and the target diagnosis information to obtain the target event; predict the duration of the target patient's ICU stay at the target time point according to the target chief complaint information and the target diagnosis information to obtain the target ICU duration; predict the hospitalization duration of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain the target hospitalization duration; predict the treatment cost of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain the target cost; predict the satisfaction evaluation of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain the target satisfaction evaluation; use the Bayesian method to predict the target expected effect of the target patient according to the target outcome, the target event, the target ICU duration, the target hospitalization duration, the target cost and the target satisfaction evaluation.
[0117] It should be noted that the target time point can be customized according to different disciplines and diseases to observe the treatment effect, such as patient discharge, etc. The main outcomes include cure, improvement and transfer, but they are not limited to these.
[0118] In this embodiment, the target time point is described in detail by taking the discharge of the patient as an example. The target outcome includes the main outcome of the target patient at discharge, but it may not be limited to this. The target event includes whether the target patient has adverse events during hospitalization, but it may not be limited to this. The target ICU duration includes the duration of the target patient's stay in the ICU during hospitalization, but it may not be limited to this. The target hospitalization duration includes the hospitalization duration of the target patient, but it may not be limited to this. The target cost includes the medical expenses of the target patient during hospitalization, but it may not be limited to this. The target satisfaction evaluation includes the satisfaction evaluation results of the target patient on the treatment process at discharge, but it may not be limited to this.
[0119] It should be noted that after extracting the feature vectors of the target chief complaint information and the target diagnosis information through the neural network model, the target outcome, the target event, the target ICU duration, the target hospitalization duration, the target cost and the target satisfaction evaluation can be predicted according to the admission information of historical patients. For example, a neural network model can be trained according to the chief complaint information, diagnosis information and main outcomes at discharge of historical patients, and the target chief complaint information and the target diagnosis information can be input into the trained neural network model to obtain the target outcome. The conditional probability that the target patient belongs to each of the main outcomes can also be calculated according to the prior probability of historical patients through the Bayesian method, and the main outcome with the largest conditional probability is taken as the target outcome. The prediction method for the target event, the target ICU duration, the target hospitalization duration, the target cost and the target satisfaction evaluation is similar to the above method, and will not be described in detail here.
[0120] It should be noted that the target outcome is represented by z1, the target event is represented by z2, the target ICU duration is represented by z3, the target hospitalization duration is represented by z4, the target cost is represented by z5, and the target satisfaction evaluation is represented by z6. The calculation method of the probability that the target patient belongs to the treatment effect category c2 includes:
[0121] ;
[0122] Among them, the value of c2 is good treatment effect or poor treatment effect. Indicates that among all historical patients, z j The prior probability of the class appearing, represents the class prior probability of c2 appearing in all historical patients, Indicates that among the historical patients with treatment effect category c2, z j The conditional probability of occurrence.
[0123] It should be noted that and The treatment effect category corresponding to the maximum value obtains the target expected effect, which provides a reference for the formulation of the patient's hospitalization treatment plan while predicting the target patient's hospitalization treatment effect.
[0124] In a feasible embodiment, the third prediction submodule also includes: predicting the target patient's usage of different medical resources according to the target chief complaint information and the target diagnosis information, and obtaining the corresponding target resource usage sub-amounts; calculating the target patient's average medical resource usage according to each of the target resource usage sub-amounts of the target patient, and obtaining the predicted resource usage; querying the database according to the disease code of the target diagnosis information to obtain a benchmark usage, and the benchmark usage is obtained according to the average resource usage of all historical patients with the same disease code as the target diagnosis information; obtaining the target patient's resource usage level according to the difference between the target resource usage and the benchmark usage; and predicting the target patient's target resource usage level using the Bayesian method according to the target patient's resource usage level to obtain the target resource usage.
[0125] It should be noted that medical resources can be graded according to actual application scenarios, and the usage of medical resources at each level can be predicted based on the target chief complaint information and the target diagnosis information. The usage of each item of medical resources can also be predicted based on the target chief complaint information and the target diagnosis information, or other feasible methods can be used.
[0126] In this embodiment, medical resources are divided into four categories according to their value and usage scenarios, including first-class medical resources, second-class medical resources, third-class medical resources and fourth-class medical resources in order from low to high, of course, they are not limited to this. The first-class medical resources include non-high-value consumables, medicines and equipment used in conventional treatment methods, of course, they are not limited to this. The second-class medical resources include high-value consumables, medicines and equipment used in conventional treatment methods, of course, they are not limited to this. The third-class medical resources include non-high-value consumables, medicines and equipment used in unconventional treatment methods, of course, they are not limited to this. The fourth-class medical resources include high-value consumables, medicines and equipment used in unconventional treatment methods, of course, they are not limited to this.
[0127] It should be noted that the target patient's usage of the four types of medical resources can be predicted by the neural network model according to the target complaint information and the target diagnosis information to obtain the target resource usage sub-amount. Each type of medical resource is linearly weighted according to the weight to obtain the resource usage of the target patient. The weight of the medical resource u k The calculation methods include:
[0128] ;
[0129] in, represents the actual usage intensity value of the k-th type of medical resources, and U represents the vector , l1 is represented as l1 norm.
[0130] It should be noted that the actual usage intensity value can be preset according to the importance that the medical institution attaches to this type of medical resources.
[0131] It should be noted that, in querying the database based on the disease code in the target diagnostic information, it also includes obtaining the disease code corresponding to the disease type in the target diagnostic information according to the target diagnostic information, that is, the coding result of the target patient's disease encoded using ICD-10, and querying the benchmark usage based on the coding result.
[0132] In this embodiment, the resource usage of the target patient is divided into three resource usage levels according to the benchmark usage. When the resource usage of the target patient is not more than 80% of the benchmark usage, the resource usage level of the target patient is lower than the average usage level; when the resource usage of the target patient is between 80%-120% of the benchmark usage, the resource usage level of the target patient is equal to the average usage level; when the resource usage of the target patient is not less than 120% of the benchmark usage, the resource usage level of the target patient is higher than the average usage level. A Bayesian classifier is used to predict the target resource usage level of the target patient according to the resource usage level of the target patient.
[0133] It should be noted that the usage of all types of medical resources required by the patient is predicted based on the target chief complaint information and the target diagnosis information. When the usage of medical resources is the same, the higher the level of medical resource usage, the more difficult the treatment of the patient is and / or the more serious the condition is and / or the greater the probability of a difficult and complicated disease, etc. The corresponding RW value of the patient and the probability of using level 4 surgery are greater, so as to predict the amount and category of resources that the patient will use during hospitalization.
[0134] In a possible implementation, refer to the attached Figure 3As shown, the system also includes a second acquisition module and a second processing module, the second acquisition module is used to acquire the allocatable amount of the target medical resources to obtain a reference resource amount, and the second processing module is used to determine whether each patient to be admitted meets the preset allocation conditions in order of bed allocation priority from high to low, and the preset allocation conditions include that the current patient's expected usage of the target medical resources is not greater than the corresponding reference resource amount; if the current patient meets the preset allocation conditions, the bed application information of the current patient is included in the allocation queue, and the reference resource amount is updated according to the current patient's expected usage of each target medical resource; if the current patient does not meet the preset allocation conditions, the bed application information of the current patient is included in the waiting queue, and the current patient is updated according to the priority order of bed allocation until there are no allocatable bed resources or the bed allocation results of all patients to be admitted are obtained.
[0135] In this embodiment, the second processing module is used to obtain the corresponding bed allocation result according to whether the patient in the current priority order meets the preset allocation conditions in the order of bed allocation priority from high to low. If the reference resource amount meets the expected resource usage of the patient in the current priority order, the bed application information of the current patient is included in the allocation queue, and the reference resource amount is updated according to the difference between the current reference resource amount and the expected resource usage of the current patient, and the bed allocation result of the patient in the next priority order is obtained according to the updated reference resource amount; if the reference resource amount does not meet the expected resource usage of the patient in the current priority order, the bed allocation result of the patient in the current priority order is abnormal allocation, and the patient in the next priority order is updated to the patient in the current priority order, and the updated patient in the current priority order is obtained, until there are no allocable bed resources or the bed allocation results of all patients to be admitted are obtained.
[0136] It should be noted that when the patient's expected usage of each medical resource is not greater than the allocatable amount of the corresponding target medical resource, the reference resource amount meets the expected resource usage of patients in the current priority order, and a bed is allocated to the patient; when the patient's expected usage of existing medical resources is greater than the allocatable amount of the corresponding medical resources, the reference resource amount does not meet the expected resource usage of patients in the current priority order, and a bed is not allocated to the patient.
[0137] In this embodiment, the patient's expected resource usage is compared with the reference resource amount according to the bed allocation priority, and the bed allocation result of the patient is obtained based on the comparison result, so as to allocate beds to patients according to the allocation priority based on the medical resources that can be allocated by the medical institution. For patients with abnormal bed allocation results, prompt information such as waiting for admission or recommending transfer will be adopted to improve the utilization rate of medical resources in the medical institution.
[0138] In this embodiment, the second processing module also includes: the second processing module also includes: if the number of patients in the current priority position who meet the preset allocation conditions is greater than 1, and the current reference resource amount cannot simultaneously meet the needs of all patients who meet the preset allocation conditions, the target medical resource whose expected usage corresponding to the patients who meet the preset allocation conditions is not 0 is used as the first candidate medical resource, and according to the classification rules of medical resources, the first candidate medical resource with the highest level is used as the second candidate medical resource for the corresponding patient, and the bed application information of the patient corresponding to the second candidate medical resource with the highest level is included in the allocation queue, and after updating the reference resource amount, it is re-determined whether other patients corresponding to the current priority position meet the preset allocation conditions.
[0139] It should be noted that, for example, when the number of patients in the current priority position is 2, the amount of medical resources currently available for allocation by the medical institution includes a1 of resources A1 (belonging to the first medical level), a2 of resources A2 (belonging to the second medical level) and a3 of resources A3 (belonging to the third medical level), the expected resource usage of the first patient includes b1 of resources A1 and b2 of resources A2, and the expected resource usage of the first patient includes b3 of resources A1 and b4 of resources A3. Among them, b1 and b3 are both less than a1, but b1+b3 is greater than a1, b2 is less than a2, and b4 is less than a3, that is, the reference resource amount satisfies the expected resource usage of the first patient and the expected resource usage of the second patient, but cannot satisfy the expected resource usage of the first patient and the second patient at the same time. Since the highest level of medical resources corresponding to the expected resource usage of the first patient is the second medical level, and the highest level of medical resources corresponding to the expected resource usage of the second patient is the third medical level, therefore, the second patient is given priority in allocating beds. When allocating beds for each patient to be admitted, priority is given to matching beds with patients who use higher-level medical resources, thereby achieving the goal of improving the effective utilization rate of advanced medical resources in medical institutions while promoting tiered diagnosis and treatment.
[0140] See attached Figure 4As shown, an embodiment of the present application provides an electronic device, including: a processor and a memory, the processor and the memory are interconnected and communicate with each other through a communication bus and / or other forms of connection mechanisms (not shown), the memory stores a computer program executable by the processor, and when the computing device is running, the processor executes the computer program to execute the system in any optional implementation of the above embodiments.
[0141] The embodiment of the present application provides a storage medium, and when the computer program is executed by the processor, the system in any optional implementation of the above embodiment is executed. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.
[0142] In the embodiments provided in the present application, it should be understood that the disclosed system can be implemented in other ways. The system embodiments described above are merely schematic. For example, the division of the modules is only a logical function division, and can be implemented in another way. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some communication interface, system or unit, which can be electrical, mechanical or other forms.
[0143] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0144] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
[0145] Flowcharts are used herein to illustrate the steps of the method of the embodiments of the present disclosure. It should be understood that the preceding or following steps are not necessarily performed precisely in order. On the contrary, various steps may be evaluated in reverse order or simultaneously. At the same time, other operations may also be added to these processes.
[0146] Unless otherwise defined, all terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the present disclosure belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or highly formal sense, unless explicitly defined as such herein.
[0147] The above is a detailed introduction to the hospital bed allocation system for constructing a disease difficulty index based on probability measurement. This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only an embodiment of this application, which is only used to help understand the hospital bed allocation system for constructing a disease difficulty index based on probability measurement in this application, and is not used to limit the scope of protection of this application. At the same time, for those skilled in the art, this application can have various changes and variations. Any modifications and equivalent substitutions made within the spirit and principles of this application should be included in the scope of protection of this application.
Claims
1. A hospital bed allocation system based on probability measurement and disease difficulty index is constructed, characterized by: include: A first acquisition module, the first acquisition module includes a first acquisition submodule, a second acquisition submodule and a first processing submodule; wherein the first acquisition submodule is used to acquire the chief complaint information of the target patient to obtain the target chief complaint information, the second acquisition submodule is used to acquire the diagnosis result of the target patient to obtain the target diagnosis information, and the first processing submodule is used to obtain the target feature information of the target patient according to the target chief complaint information and the target diagnosis information; The target characteristic information includes a target severity level, and the first processing submodule further includes a first prediction submodule, the first prediction submodule being used to predict the severity level of the target patient's condition using a Bayesian method according to the target chief complaint information and the target diagnosis information to obtain the target severity level; A first prediction module, the first prediction module is used to predict the relative weight of the target patient using a Bayesian method according to the target feature information to obtain a target relative weight; A second prediction module, the second prediction module is used to predict the probability of the target patient receiving level 4 surgery during hospitalization by using a Bayesian method according to the target feature information, to obtain a target surgery probability; A first processing module, the first processing module is used to obtain the bed allocation priority rank of the target patient according to the target relative weight and the target surgery probability, wherein the target patient with a larger target relative weight has a higher bed allocation priority rank, and the target patient with a larger target surgery probability has a higher bed allocation priority rank; Wherein, the first processing module includes a first evaluation submodule, the first evaluation submodule is used to evaluate the target relative weight to obtain a first evaluation subresult; evaluate the target surgery probability to obtain a second evaluation subresult; and calculate the evaluation result of the target patient according to the first evaluation subresult and the second evaluation subresult; The target feature information also includes a target expected effect, and the first processing submodule also includes a second prediction submodule, and the second prediction submodule is used to predict the treatment effect of the target patient at the target time point using the Bayesian method according to the target chief complaint information and the target diagnosis information to obtain the target expected effect; The target feature information also includes a target resource usage, and the first processing submodule also includes a third prediction submodule, and the third prediction submodule is used to predict the resource usage of the target patient during hospitalization using a Bayesian method according to the target chief complaint information and the target diagnosis information to obtain the target resource usage; The target characteristic information also includes a target waiting degree. The first acquisition module also includes a third acquisition submodule, and the third acquisition submodule is used to obtain the target patient's waiting time for admission to obtain the actual waiting time. The first processing submodule also includes a fourth prediction submodule, and the fourth prediction submodule is used to estimate the target patient's waiting time for admission to the hospital based on the target chief complaint information and the target diagnosis information using a Bayesian method to obtain the target waiting time; the target waiting degree is obtained based on the comparison result of the target waiting time and the actual waiting time.
2. The hospital bed allocation system based on probability measurement to construct disease difficulty index according to claim 1 is characterized in that: The second prediction module also includes: Predicting the primary outcome of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain the target outcome; Predicting whether the target patient will experience an adverse event at the target time point according to the target chief complaint information and the target diagnosis information, and obtaining a target event; Predicting the duration of the target patient's stay in the ICU at the target time point according to the target chief complaint information and the target diagnosis information, and obtaining a target ICU duration; Predicting the hospitalization duration of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain a target hospitalization duration; Predicting the treatment cost of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain a target cost; Predicting the satisfaction evaluation of the target patient at the target time point according to the target chief complaint information and the target diagnosis information to obtain a target satisfaction evaluation; The target expected effect of the target patient is predicted using the Bayesian method based on the target outcome, the target event, the target ICU duration, the target hospitalization duration, the target cost and the target satisfaction evaluation.
3. The hospital bed allocation system based on probability measurement to construct disease difficulty index according to claim 1 is characterized in that: The third prediction module also includes: Predicting the target patient's usage of different target medical resources according to the target chief complaint information and the target diagnosis information, and obtaining corresponding target resource usage sub-amounts; wherein the target medical resources include medical resources other than bed resources; Calculate the average medical resource usage of the target patient according to each target resource usage sub-amount of the target patient to obtain a predicted resource usage; querying a database according to the disease code of the target diagnosis information to obtain a benchmark usage amount, wherein the benchmark usage amount is obtained according to an average of resource usage amounts of all historical patients having the same disease code as the target diagnosis information; Obtaining the resource usage level of the target patient according to the difference between the target resource usage and the benchmark usage; The target resource usage level of the target patient is predicted using a Bayesian method according to the resource usage level of the target patient to obtain the target resource usage.
4. The hospital bed allocation system for constructing disease difficulty index based on probability measurement according to claim 1 or 3, characterized in that: The system further includes a second acquisition module and a second processing module, wherein the second acquisition module is used to acquire the allocatable amount of the target medical resource to obtain a reference resource amount, and the second processing module is used to determine whether each patient to be admitted meets a preset allocation condition in descending order of the priority of bed allocation, wherein the preset allocation condition includes that the expected usage of the target medical resource by the current patient is not greater than the corresponding reference resource amount; If the current patient meets the preset allocation conditions, the bed application information of the current patient is included in the allocation queue, and the reference resource amount is updated according to the expected usage of each target medical resource by the current patient; If the current patient does not meet the preset allocation conditions, the bed application information of the current patient will be added to the waiting queue, and the current patient will be updated according to the priority order of bed allocation until there are no allocatable bed resources or the bed allocation results of all patients to be admitted are obtained.
5. The hospital bed allocation system based on probability measurement to construct disease difficulty index according to claim 4 is characterized in that: The second processing module also includes: If the number of patients in the current priority position who meet the preset allocation conditions is greater than 1, and the current reference resource amount cannot simultaneously meet the needs of all patients who meet the preset allocation conditions, the target medical resource with an expected usage amount not equal to 0 corresponding to the patients who meet the preset allocation conditions shall be used as the first candidate medical resource, and according to the classification rules of medical resources, the first candidate medical resource with the highest level shall be used as the second candidate medical resource for the corresponding patient, and the bed application information of the patient corresponding to the second candidate medical resource with the highest level shall be included in the allocation queue. After updating the reference resource amount, it shall be re-judged whether other patients corresponding to the current priority position meet the preset allocation conditions.
