Calculation method for high-risk medication pathways
By calculating the risk value and risk score of the drug arrangement path and combining the data from the medical record database, the high-risk drug use paths are found, which solves the problem of complexity of drug interactions and improves the safety and treatment effect of drug use.
Patent Information
- Application Number
- CN202111060560.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2020-12-24
- Filing Date
- 2021-09-10
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2041-09-10
AI Technical Summary
Due to the wide variety of drugs and the complex paths of patients' medication, the arrangement and combination of drug interactions have shown nonlinear growth, and the amount of data is huge, making it difficult to find high-risk medication paths efficiently, affecting the treatment effect and safety.
By listing multiple drug arrangement paths in the medical record database, calculate the risk values and hazard scores of each path, and after sorting, take out the high-risk score paths and combine them, combining the chance of winning, reverse file frequency and occurrence frequency, find out the results of the designated medication use by finding the chance of being used by the drug with the specified one. high-risk paths and adjust the path combination according to the preset number of people.
Effectively save time, avoid searching for huge medication paths, remind doctors to pay attention to the patients' medication history, reduce adverse chain effects, and improve medication safety and treatment effects.
Smart Images

Figure CN114678140B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method for searching usage paths among multiple drugs, and in particular to a method for calculating high-risk medication paths. Background Art
[0002] In recent years, due to the aging population and the prevalence of chronic diseases, taking two or more drugs at the same time or combining them into a drug, Chinese herbal medicine or health food has made medication more complicated. When the effect of a drug is changed by other drugs, resulting in a decrease or increase in efficacy, it is called a drug interaction. Milder drug interactions may only affect the absorption of the drug and affect the efficacy, but in severe cases, it may be fatal. The chain effect of drug interactions may pass through many drugs, and the patient may have seen more than one doctor, making it difficult to trace the cause. Therefore, if the high-risk medication path can be found from the patient's medical record data, the doctor can be reminded to pay attention to the patient's medication history and avoid adverse chain effects. However, due to the wide variety of drugs, the longer the patient's medication path, the more its permutations and combinations will grow nonlinearly, resulting in a huge amount of data. Therefore, an efficient way is needed to find high-risk paths. Summary of the invention
[0003] The present invention is a method for calculating a high-risk medication pathway, which can save time and avoid searching for a huge medication pathway.
[0004] The method for calculating a high-risk medication pathway of the present invention comprises: listing a plurality of arrangement pathways consisting of two drugs each among a plurality of drugs included in a medical record database; calculating the risk value of each arrangement pathway by querying the medical record database based on a designated medication result; calculating the hazard score of each arrangement pathway according to the risk value, and sorting the arrangement pathways based on the hazard score; and taking out N arrangement pathways starting from the one with the highest hazard score for combination to obtain a plurality of serially connected pathways, wherein the number of drugs included in each serially connected pathway meets the designated medication quantity.
[0005] In one embodiment of the present invention, after the step of taking out N permutation paths starting from the one with the highest risk score and combining them to obtain the cascade path, it also includes: setting a preset number of people; determining whether the number of patients who meet the obtained cascade path and lead to the specified medication result meets the preset number; and if the number of patients does not meet the preset number, updating N to N+M, and re-executing the step of taking out N permutation paths starting from the one with the highest risk score and combining them to obtain the cascade path until the number of patients meets the preset number.
[0006] In one embodiment of the present invention, the risk value is an odds ratio, and the odds ratio of the i-th arrangement path is calculated by the following formula:
[0007]
[0008] Among them, O represents the odds ratio of the ith permutation path, D E H is the number of patients in their drug use records that meet the i-th permutation path and lead to the specified drug use result, E D is the number of patients whose drug use records meet the i-th permutation path and do not lead to the specified drug use result. N H is the number of patients whose drug use records do not have the i-th permutation path that leads to the specified drug use result, N The number of patients whose drug use records do not have the i-th permutation path and do not lead to the specified medication outcome.
[0009] In one embodiment of the present invention, the step of calculating the risk score of each arrangement path based on the risk value includes: setting the risk ranking of each arrangement path based on the odds ratio of each arrangement path, and using the risk ranking as the risk score; calculating the probability value of each arrangement path, and setting the probability ranking of each arrangement path based on the probability value; and adding the risk ranking to the probability ranking to obtain the risk score.
[0010] In one embodiment of the present invention, the method for calculating the high-risk medication pathway further includes: querying the number of patients who meet the i-th arrangement pathway in the medical record database based on the following setting, wherein the i-th arrangement pathway includes the first drug and the second drug in sequence. The setting is: querying the number of patients who used the first drug and the second drug in sequence within a specified time range in the medication use records of multiple patients.
[0011] In one embodiment of the present invention, the method for calculating the high-risk medication pathway further includes: by querying the medical record database, taking out multiple medication pathways recorded in the medical record database in the concatenated pathway, and obtaining a data set of multiple diseases corresponding to multiple patients with specified medication results, wherein one patient corresponds to one disease and has a set of pathway sets, and the pathway set includes at least one medication pathway; calculating the reverse file frequency value of each medication pathway based on the following formula,
[0012]
[0013] Where IDF(i) represents the inverse document frequency value of the i-th medication pathway, D is the total number of diseases, and t(i) is the number of i-th medication pathways included in the data set;
[0014] The frequency of occurrence of each medication pathway in the pathway set corresponding to each patient is calculated based on the following formula:
[0015]
[0016] Among them, TFi,j represents the frequency of occurrence of the i-th medication pathway in the pathway set corresponding to the j-th patient, n i,j represents the number of pathways corresponding to the i-th medication pathway appearing in the j-th patient, A j Represents the number of medication pathways included in the pathway set corresponding to the jth patient;
[0017] Calculate the estimated value of each patient corresponding to the i-th medication pathway based on the reverse file frequency value and the occurrence frequency of each medication pathway; and
[0018] Based on the estimated values, unique pathways were selected for each disease.
[0019] In one embodiment of the present invention, the step of selecting a unique pathway for each disease includes: calculating an average value based on the estimated values of all patients corresponding to each medication pathway, and using the medication pathway with the largest average value as the unique pathway.
[0020] In one embodiment of the present invention, the step of selecting a unique pathway for each disease includes: calculating an average value based on the estimated values of all patients corresponding to each medication pathway, and after determining a threshold value by the elbow method, selecting a medication pathway with an average value greater than the threshold value as a unique pathway.
[0021] Based on the above, the present disclosure proposes a method for efficiently finding high-risk pathways, which can save time and avoid searching for huge medication pathways. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 is a block diagram of an electronic device according to an embodiment of the present invention;
[0023] Figure 2 is a flow chart of a method for calculating a high-risk medication pathway according to an embodiment of the present invention;
[0024] Figure 3 is a flow chart of a method for finding unique pathways for different diseases according to an embodiment of the present invention;
[0025] Figure 4 FIG. 4 is a graph showing the average estimated values of the medication route according to an embodiment of the present invention.
[0026] Description of Reference Numerals
[0027] 100: Electronic devices
[0028] 110: Processor
[0029] 120: Storage device
[0030] 121: Medical Record Database
[0031] 130: Output device
[0032] T:Threshold
[0033] S205~S220: Steps of the calculation method for high-risk medication pathway
[0034] S305-S320: Steps in the method of finding unique pathways for different diseases DETAILED DESCRIPTION
[0035] Figure 1 is a block diagram of an electronic device according to an embodiment of the present invention. Figure 1 The electronic device 100 includes a processor 110, a storage device 120, and an output device 130. The processor 110 is coupled to the storage device 120 and the output device 130. The processor 110 is, for example, a central processing unit (CPU), a physical processing unit (PPU), a programmable microprocessor, an embedded control chip, a digital signal processor (DSP), an application specific integrated circuit (ASIC), or other similar devices.
[0036] The storage device 120 is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk or other similar device or a combination of these devices. The medical record database 121 records the drug use records of multiple patients. In other embodiments, the medical record database 121 can also be set up in a cloud server, and the cloud server downloads the medical record database 121 to the electronic device 100 in advance, or the electronic device 100 connects to the cloud server in real time to query the medical record database 121. In addition, the storage device 120 also stores a plurality of code segments, which, after being installed, will be executed by the processor 110 to implement the following method for calculating high-risk medication paths.
[0037] The calculation method of the high-risk medication path disclosed in the present invention is to find the medication path that may cause the specified medication result among multiple drugs. In the following embodiment, the specified number of medications is 3, and the medication path is represented by A→B→C, which means that the patient takes medication in the order of taking medication A first, taking medication B within a specified time range (e.g., 30 days) after taking medication A, and taking medication C within a specified time range (e.g., 30 days) after taking medication B. Other drugs can be taken after taking medication A and before taking medication B, and other drugs can be taken after taking medication B and before taking medication C.
[0038] Figure 2 This is a flow chart of a method for calculating a high-risk medication path according to an embodiment of the present invention. Figure 2 In step S205, multiple arrangement paths consisting of two drugs in the multiple drugs included in the medical record database 121 are listed. The processor 110 executes the relevant code segment to find each used drug recorded in the medical record database 121, and form multiple arrangement paths in groups of two.
[0039] For example, assuming that the drug use records of multiple patients recorded in the medical record database 121 include a total of K drugs, K×(K-1) arrangement paths can be obtained. Among them, the two drugs included in each arrangement path have different usage orders. For drug A10BA, drug C10AA, and drug D01AC, they can be combined to obtain 6 arrangement paths, namely, A10BA→C10AA, A10BA→D01AC, C10AA→A10BA, C10AA→D01AC, D01AC→A10BA, and D01AC→C10AA.
[0040] Next, in step S210, based on the specified medication results, the risk value of each arrangement path is calculated by querying the medical record database 121. Here, the risk value can be calculated using odds ratio or logistic regression. In this embodiment, the odds ratio is used as the risk value, that is, the odds ratio O of the i-th arrangement path is calculated by the following formula:
[0041]
[0042] Among them, D E is the number of patients in the drug use record that meet the i-th permutation path and lead to the specified drug use result, H E D is the number of patients whose drug use records meet the i-th permutation path and do not lead to the specified drug use result. NH is the number of patients whose drug use records do not have the i-th permutation path that leads to the specified drug use result, N The number of patients whose drug use records do not have the i-th permutation path and do not lead to the specified medication outcome.
[0043] For the arrangement path A10BA→C10AA (first take drug A10BA and then take drug C10AA), query the drug use records of each patient in the medical record database 121 to find out the patients with the arrangement path A10BA→C10AA and the patients without the arrangement path A10BA→C10AA. Among the patients with the arrangement path A10BA→C10AA, find out the number of patients who are hospitalized D E and the number of patients who were not hospitalized E And, among the patients who do not have the arrangement path A10BA→C10AA, find the number of patients who are hospitalized D N and the number of patients who were not hospitalized N Then, the risk value (odds ratio) of the arrangement path A10BA→C10AA is obtained by the above formula. A10BA→C10AA ). And so on, calculate the risk value corresponding to each arrangement path.
[0044] Here, the number of patients who match the i-th arrangement path (including the first drug and the second drug in sequence) in the medical record database 121 is queried based on the following setting: the number of patients who used the first drug and the second drug in sequence within a specified time range (for example, 30 days) is queried in the drug use records of multiple patients.
[0045] Next, in step S215, the risk score of each arrangement path is calculated according to the risk value, and the arrangement paths are sorted based on the risk score. Further, the risk ranking of each arrangement path is set based on the odds ratio of each arrangement path, and the probability value (probability value, p value) of each arrangement path is calculated, and the probability ranking of each arrangement path is set based on the p value. Afterwards, the risk ranking is added to the probability ranking to obtain the risk score. For example, Table 1 shows the risk ranking, probability ranking and risk score (risk ranking + probability ranking) corresponding to the arrangement path of one embodiment.
[0046] Table 1
[0047] Arrange Path Risk Ranking Probability Ranking Hazard score A10BA→C10AA 1 17 18 … D01AC→H03BA 20 14 34 … … … … N07AA→S01LA 15 8 23
[0048] After obtaining the risk score, in step S220, N arrangement paths are taken out starting from the one with the highest risk score and combined to obtain multiple serial paths. Here, the number of drugs included in each serial path meets the specified number of medications. For example, assuming that the specified number of medications is 3, it means that the serial path is composed of two arrangement paths. For example, the arrangement path A→B and the arrangement path B→C can form the serial path A→B→C.
[0049] In this embodiment, a higher risk score means that the medication order represented by the arrangement path has a higher probability of leading to a specified medication result (e.g., hospitalization). The hospitalization risk of the concatenated path obtained by the arrangement path with a high risk score is usually also greater. Accordingly, starting from the arrangement path with a high risk score to find a high-risk medication path can save time in searching all paths.
[0050] Specifically, the starting value of N and the stop condition of the search are set in advance. Here, the stop condition of the search is that the number of patients who meet the obtained serial path and lead to the specified medication result meets the preset number. For example, the preset number is 50% of the total number of hospitalized patients. Afterwards, the processor 110 queries the drug use record of each patient in the medical record database 121 to find out the total number of patients who have been hospitalized. In addition, it is determined whether the number of patients who meet the obtained serial path and lead to the specified medication result meets the preset number. If the number of patients who meet the obtained serial path and lead to the specified medication result does not meet the preset number, N=N+M is reset, and the step of taking out N permutation paths from the one with the highest risk score and combining them to obtain the serial path that meets the specified medication number is re-executed until the number of patients who meet the obtained serial path and lead to the specified medication result meets the preset number. If N+M is greater than the obtained permutation path, all permutation paths are taken for combination to obtain multiple serial paths that meet the specified medication number.
[0051] Table 2 shows the search results of searching in the arrangement paths. In the two arrangement paths to be connected in series, the second drug in the first arrangement path must be the same as the first drug in the second arrangement path to be connected in series. Taking the arrangement paths R05FA→N02BE, N02BE→A02BA, and N02BE→M01AB in Table 2 as an example, the combinable series paths are R05FA→N02BE→A02BA and R05FA→N02BE→M01AB.
[0052] Table 2
[0053]
[0054] In Table 2, the preset number of patients is 50% of the total number of hospitalized patients, the initial value of N is 100, and M = 100. When N is 100, the number of patients whose search results meet the obtained concatenated path and lead to the specified medication result is only 7%, so N is reset to 200, until N is 1500, the search result (53%) is greater than the preset number of patients 50%, so the stop condition is met and the search is stopped.
[0055] After obtaining the final concatenated path, unique medication paths (hereinafter referred to as unique paths) can be further found for different diseases. In the following embodiment, term frequency and inverse document frequency techniques are used to find unique paths for different diseases.
[0056] Figure 3 is a flow chart of a method for finding unique pathways for different diseases according to an embodiment of the present invention. Figure 3 In step S305, the IDF value of each medication path is calculated. Specifically, the processor 110 retrieves multiple medication paths recorded in the medical record database 121 in the concatenated path by querying the medical record database 121, and obtains a data set of multiple diseases corresponding to multiple patients with specified medication results.
[0057] For example, assuming that the serial paths include A1 to A10, the processor 110 determines whether the serial paths A1 to A10 exist in the medical record database 121 by querying the medical record database 121, so as to find out those that match the serial paths A1 to A10 as the medication paths. Here, it is assumed that there are medication paths P1 to P5 that match the serial paths A1 to A5 in the medical record database 121, and it is assumed that the designated medication result is hospitalization. Next, the patients with these medication paths P1 to P5 and hospitalization records are taken out, as well as the diseases (hospitalization reasons) caused by these medication paths, and the data set is obtained.
[0058] For example, Table 3 illustrates a data set of an embodiment. As shown in Table 3, each patient corresponds to a disease (reason for hospitalization) and has a corresponding path set, and the path set includes at least one medication path. In the data set shown in Table 3, different patients are distinguished by patient numbers, and the diseases listed here include pneumonia, peptic ulcer and stroke, but are not limited to this. For patient number U01, the patient with patient number U01 corresponds to pneumonia, and its path set includes medication paths P1, P2, and P4.
[0059] Table 3
[0060] Patient number Disease (reason for hospitalization) Path Collection U01 pneumonia {P1;P2;P4} U02 pneumonia {P3} U03 Peptic ulcer {P1;P4} U04 Stroke {P2;P4;P5} U05 Stroke {P3;P4}
[0061] After obtaining the data set, the processor 110 calculates the IDF value of each medication route based on the following formula (1).
[0062] Formula (1):
[0063] Among them, IDF(i) represents the IDF value of the i-th medication pathway, D is the total number of diseases, and t(i) is the number of i-th medication pathways included in the data set. Taking Table 3 as an example, the medication pathways include P1 to P5, and D = 3. For medication pathway P1, the number of medication pathways P1 in the data set is 2, so its IDF(P1) = log(3 / 2) = 0.41. By analogy, the IDF values of medication pathways P2 to P5 can be obtained respectively, as shown in Table 4.
[0064] Here, the larger the IDF value is, the fewer hospitalization reasons (diseases) the corresponding medication pathway appears in, indicating that it is more unique.
[0065] Table 4
[0066] Medication route P1 P2 P3 P4 P5 IDF(i) log(3 / 2)=0.41 log(3 / 2)=0.41 log(3 / 2)=0.41 log(3 / 3)=0 log(3 / 1)=1.1
[0067] Then, in step S310, the processor 110 calculates the frequency of occurrence of each medication path in the path set corresponding to each patient. Here, the frequency of occurrence is calculated by using a word frequency calculation method. That is, the frequency of occurrence of each medication path in the path set corresponding to each patient is calculated based on the following formula (2).
[0068] Formula (2):
[0069] Among them, TF i,j represents the frequency of occurrence of the i-th medication pathway in the pathway set corresponding to the j-th patient, n i,j represents the number of pathways corresponding to the i-th medication pathway appearing in the j-th patient, A j Represents the number of medication pathways included in the pathway set corresponding to the jth patient.
[0070] In Table 3, the path set {P1; P2; P4} of patient number U01 includes 3 medication paths, namely A U01 is 3. The number n of the path set of medication path P1 in patient number U01 P1,U01 is 1, based on which, TF is calculated P1,U01 =1 / 3. The number of path sets n for medication path P2 in patient number U01 P2,U01 is 1, based on which, TF is calculated P2,U01 is 1 / 3. The number of path sets n of medication path P3 in patient number U01 P3,U01 is 0, based on which, TF is calculatedP3,U01 is 0 / 3. The number n of the path set of medication path P4 in patient number U01 P4,U01 is 1, based on which, TF is calculated P4,U01 is 1 / 3. The number of path sets n of medication path P5 in patient number U01 P5,U01 is 0, based on which, TF is calculated P5,U01 is 0 / 3. By analogy, we can obtain the occurrence frequency TF of each path set as shown in Table 5. i,j .
[0071] Then, in step S315, the processor 110 calculates the estimated value of each patient corresponding to each medication pathway. That is, based on the IDF value and the occurrence frequency of each medication pathway, the estimated value of each patient corresponding to the i-th medication pathway is calculated. Here, the estimated value = TF i,j ×IDF(i). As shown in Table 4, IDF(i) and TF(i) in Table 5. i,j For , the estimated values shown in Table 6 can be obtained.
[0072] Table 5
[0073]
[0074] Table 6 (TF i,j ×IDF(i)
[0075]
[0076] In step S320, the processor 110 selects a unique path. That is, based on the estimated values, a unique path is selected for each disease. Here, the processor 110 calculates an average value based on the estimated values of all patients corresponding to each medication path, and uses the medication path with the largest average value as the unique path.
[0077] Referring to Table 3 and Table 6, pneumonia can be organized as Table 7, and stroke can be organized as Table 8. From Table 7, the unique path for pneumonia can be selected as P3, which means that in the case of hospitalization due to pneumonia, the medication path P3 has a high risk and the probability of causing other reasons for hospitalization is low. From Table 8, the unique path for stroke can be selected as P5, which means that in the case of hospitalization due to stroke, the medication path P5 has a high risk and the probability of causing other reasons for hospitalization is low. In addition, based on the example of Table 3, peptic ulcer only corresponds to patient number U03, so, referring to Table 6, the medication path P1 corresponding to the maximum value in the estimated value corresponding to patient number U03 is taken as the unique path.
[0078] Table 7
[0079]
[0080] Table 8
[0081]
[0082] In another approach, the average value is calculated based on the estimated values of all patients corresponding to each medication pathway, and after selecting a threshold using the elbow method, all medication pathways with an average value greater than the threshold are selected as unique pathways. Here is another example to illustrate. In the embodiment shown in Table 9, it is assumed that the cause of hospitalization is pneumonia, the patients include Ua, Ub, and Uc, and the medication pathways include P11 to P16.
[0083] Table 9
[0084]
[0085] The average values are calculated based on the estimated values of the medication routes P11 to P16 corresponding to all patients shown in Table 9, and a curve graph can be obtained based on these average values, such as Figure 4 shown. Figure 4 is a graph showing the average value of the estimated value of the medication path according to an embodiment of the present invention. Figure 4 In the example, the elbow method is used to take the elbow position (corresponding to medication path P13) in the curve graph as the threshold T, and medication paths with an average value greater than the threshold T are regarded as unique paths. That is, medication paths P11, P12, and P15 are determined to be unique paths.
[0086] In summary, the present invention uses the above-mentioned calculation method of high-risk medication pathways to find out high-risk medication pathways that lead to designated medication results, and can also find out unique pathways that lead to designated medication results for different reasons.
[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for calculating high-risk medication pathways, It is characterized in that include: Listing a plurality of arrangement paths consisting of two drugs in a plurality of drugs included in a medical record database, wherein the number of drugs included in the medical record database is K, the number of arrangement paths is K×(K-1), and each arrangement path represents a first drug and a second drug based on a medication order; Based on the designated medication results, the risk value of each of the arranged pathways is calculated by querying the medical record database; Calculating a risk score for each of the arrangement paths according to the risk value, and sorting the arrangement paths based on the risk score; as well as Starting from the one with the highest risk score, N of the arrangement paths are taken out for combination to obtain multiple serial connection paths, wherein the number of drugs included in each serial connection path meets the specified medication quantity, and each serial connection path is obtained by serial connection of the arrangement paths whose number is the specified medication quantity minus 1, and the second drug in the first arrangement path of the two serial connection arrangement paths is the same as the first drug in the second arrangement path.
2. The method for calculating a high-risk medication pathway according to claim 1, wherein after the step of taking out N arranged pathways starting from the one with the highest risk score and combining them to obtain the concatenated pathway, further comprising: include: Set the preset number of people; Determining whether the number of patients who meet the obtained concatenation path and lead to the designated medication result meets the preset number; as well as If the number of patients does not meet the preset number, N is updated to N+M, and the step of taking out N arrangement paths from the one with the highest risk score and combining them to obtain the concatenated path is re-executed until the number of patients meets the preset number.
3. The method for calculating a high-risk medication pathway according to claim 1, wherein the risk value is an odds ratio, and the odds ratio of the i-th arrangement pathway is calculated by the following formula: in, O represents the odds ratio of the i-th arrangement path, D E is the number of patients in their drug use records that match the i-th arrangement path and lead to the specified drug use result, H E is the number of patients whose drug use records match the i-th arrangement path and do not lead to the specified drug use result, D N is the number of patients whose drug use records do not have the i-th permutation path and lead to the specified drug use result, H N The number of patients whose drug use records do not have the i-th arrangement path and do not lead to the specified drug use result.
4. The method for calculating high-risk medication pathways according to claim 3, wherein the step of calculating the risk score of each of the arranged pathways according to the risk value is: include: Setting a risk ranking for each of the arrangement paths based on the odds ratio of each of the arrangement paths; Calculating a probability value of each of the arrangement paths, and setting a probability ranking of each of the arrangement paths based on the probability value; as well as The risk ranking is added to the probability ranking to obtain the hazard score.
5. The method for calculating high-risk medication pathways according to claim 1, further comprising: include: The number of patients matching the i-th arrangement path in the medical record database is queried based on the following setting, wherein the setting is: querying the number of patients who used the first drug and the second drug sequentially within a specified time range in the drug use records of multiple patients.
6. The method for calculating high-risk medication pathways according to claim 1, further comprising: include: By querying the medical record database, taking out a plurality of medication pathways recorded in the medical record database in the concatenated pathway, and obtaining a data set of a plurality of diseases corresponding to a plurality of patients having the specified medication results, wherein each of the patients corresponds to one of the diseases and has a pathway set, and the pathway set includes at least one of the medication pathways; The reverse file frequency value of each medication route is calculated based on the following formula: Wherein, IDF(i) represents the inverse file frequency value of the i-th medication pathway, D is the total number of the diseases, and t(i) is the number of the i-th medication pathways included in the data set; The occurrence frequency of each medication pathway in the pathway set corresponding to each patient is calculated based on the following formula: Among them, TF i,j represents the occurrence frequency of the i-th medication pathway in the pathway set corresponding to the j-th patient, n i,j represents the number of the path set corresponding to the i-th medication path appearing in the j-th patient, A j represents the number of medication pathways included in the pathway set corresponding to the jth patient; Calculate the estimated value of each of the patients corresponding to the i-th medication pathway based on the reverse file frequency value and the occurrence frequency of each of the medication pathways; and Based on the estimates, unique pathways are selected for each of the diseases.
7. The method for calculating high-risk medication pathways according to claim 6, wherein the step of selecting the unique pathway for each of the diseases is: include: Calculate the average value based on the estimated values of all patients corresponding to each medication route; as well as The medication route with the largest average value is taken as the unique route.
8. The method for calculating high-risk medication pathways according to claim 6, wherein the step of selecting the unique pathway for each of the diseases is: include: Calculate the average value based on the estimated values of all patients corresponding to each medication route; The threshold is determined by the elbow method; as well as All medication pathways whose average values are greater than the threshold are selected as the unique pathways.
