AI-based weight allocation method for medical decision-making

Abstract

This invention relates to the field of smart healthcare, and more particularly to an AI-based method for weighting medical decisions. The method includes: determining the data extraction method based on data abundance; when selecting from multiple data sources, determining the data matching degree based on historical incidence rates and rhythm disorder coefficients, and designating data with a matching degree greater than a preset matching degree as supplementary data; determining the weight coefficients of relevant parameters of the data matching degree based on a critical threshold; determining the supplementary selection method based on the relevance of effective data, either by selecting medical record data based on feature keywords or by selecting medical record data based on a dynamic difference coefficient; determining the weight adjustment method corresponding to the relevance of effective data based on the treatment effectiveness coefficient, either by determining the weight adjustment amplitude based on a benchmark weight allocation method or by determining the weight adjustment amplitude based on the treatment effectiveness difference; and ending the weight adjustment when the weight adjustment amplitude reaches a preset adjustment amplitude. This improves the accuracy and efficiency of medical record matching.

Description

Technical Field

[0001] This invention relates to the field of smart healthcare, and more particularly to an AI-based method for allocating weights in medical decision-making. Background Technology

[0002] Existing deep learning-based medical decision weighting techniques fail to determine appropriate processing methods based on the characteristics of medical records during the actual matching process. This results in a lack of targeted matching methods in subsequent record matching processes, leading to low accuracy and efficiency in record matching.

[0003] Chinese Patent Publication No. CN119230108A discloses a medical decision support method and system based on medical rules, including: receiving patient medical data from various medical devices and information systems; standardizing the received data; mapping the standardized data to corresponding medical concepts; and acquiring and recording data association information during the data mapping process. This invention effectively solves the problem of misassociation of medical data from different sources through standardization processing and machine learning model analysis, improving the accuracy of diagnosis and treatment. By evaluating the quality of data association and filtering out misassociated data, misdiagnosis and overtreatment are prevented. Clustering algorithms are used to optimize the identification and exclusion of abnormal data, ensuring that diagnoses are based on reliable data, reducing unnecessary medical interventions caused by data errors, and improving the overall quality of care and medical service levels. However, the above technical solution has the following problems: it fails to determine the appropriate processing method based on the characteristic status of the medical records during the actual matching process, leading to a lack of targeted selection of matching methods in subsequent medical record matching processes, resulting in low accuracy and processing efficiency in medical record matching. Summary of the Invention

[0004] To address this issue, the present invention provides an AI-based medical decision weight allocation method to overcome the problem in existing technologies that fail to determine the appropriate processing method based on the characteristic status of medical records during the actual matching process, resulting in the inability to select a matching method in a targeted manner during subsequent record matching, thus leading to low accuracy and efficiency in record matching.

[0005] To achieve the above objectives, this invention provides an AI-based method for weight allocation in medical decision-making, comprising:

[0006] The data extraction method is determined based on the data abundance: either by extending the selection from multiple data sources or by supplementing the data based on the effective relevance.

[0007] When selecting multiple data sources, the data matching degree is determined based on historical incidence rate and rhythm disorder coefficient, and data with a matching degree greater than the preset matching degree are used as supplementary data.

[0008] The weight coefficients of relevant parameters for data matching degree are determined based on the critical threshold. The weight of any relevant parameter that is greater than the critical threshold is increased and adjusted. The adjustment range is determined based on the effective difference. The relevant parameter is the difference in historical incidence rate or the difference in rhythm disorder coefficient. The difference in historical incidence rate between the target medical record and the comparison medical record is recorded as the difference in historical incidence rate. The difference in rhythm disorder coefficient between the target medical record and the comparison medical record is recorded as the difference in rhythm disorder coefficient.

[0009] The supplementary selection method is determined based on the comparison results between the relevance of valid data and the preset relevance of valid data. The selection method is either based on the selection of medical record data according to the feature keywords or based on the dynamic difference coefficient.

[0010] The weight adjustment method for the correlation of effective data is determined based on the comparison between the effective coefficient and the preset effective coefficient. This method is either based on the baseline weight allocation method or based on the difference in the effectiveness of treatment.

[0011] Data sources include medical records, laboratory data, and imaging reports;

[0012] When the weight adjustment amplitude reaches the preset adjustment amplitude, the weight adjustment ends.

[0013] Furthermore, the data extraction method is determined based on data richness, and the extraction methods include:

[0014] If the data abundance is greater than the preset data abundance, the data extraction method will be extended selection from multiple data sources;

[0015] If the data abundance is less than or equal to the preset data abundance, the data extraction method is to supplement the data based on the effective relevance.

[0016] Furthermore, in response to the first preset supplementation condition, the data matching degree is determined based on the historical incidence rate and rhythm disorder coefficient, and data with a matching degree greater than the preset data matching degree are used as supplementary data;

[0017] The first preset supplementary condition is to determine that the data extraction method is a multi-source data source extended selection.

[0018] Furthermore, the weight coefficients of the relevant parameters of data matching degree are determined based on the critical threshold; the weight of any relevant parameter that is greater than the critical threshold is increased, and the increase is determined based on the effective difference.

[0019] Among them, the relevant parameters are the difference in historical incidence rates or the difference in rhythm disorder coefficients. The difference in historical incidence rates between the target medical record and the control medical record is recorded as the difference in historical incidence rates, and the difference in rhythm disorder coefficients between the target medical record and the control medical record is recorded as the difference in rhythm disorder coefficients.

[0020] Furthermore, in response to the second preset supplementary condition, the supplementary selection method is determined based on the comparison result between the correlation of effective data and the preset correlation of effective data;

[0021] If the relevance of the effective data is less than the preset relevance of the effective data, then the supplementary selection method is determined to be to select medical record data based on feature keywords;

[0022] If the relevance of the effective data is greater than or equal to the preset relevance of the effective data, then the supplementary selection method is determined to be to select medical record data based on the dynamic difference coefficient;

[0023] The second preset supplementary condition is to determine the data extraction method as supplementing data based on effective relevance.

[0024] Furthermore, the relevance of effective data is determined based on keyword repetition and keyword proportion;

[0025] The relevance of the effective data is positively correlated with both keyword repetition and keyword proportion.

[0026] Furthermore, the weighting adjustment method corresponding to the relevance of effective data is determined based on the diagnostic effectiveness coefficient;

[0027] If the effective treatment coefficient is greater than the preset effective treatment coefficient, then the allocation will be based on the benchmark weighting method;

[0028] If the effective coefficient of diagnosis and treatment is less than or equal to the preset effective coefficient of diagnosis and treatment, the weight adjustment amplitude shall be determined according to the difference in the effectiveness of diagnosis and treatment.

[0029] The adjustment amplitude is positively correlated with the difference in diagnostic and treatment effectiveness.

[0030] Furthermore, the treatment effectiveness coefficient is determined based on the cure rate and recurrence rate;

[0031] The effective treatment coefficient and the cure rate are positively correlated.

[0032] The effective treatment coefficient and the recurrence rate are negatively correlated.

[0033] Furthermore, data sources include medical records, laboratory data, and imaging reports.

[0034] Furthermore, when the weight adjustment amplitude reaches the preset adjustment amplitude, the weight adjustment ends.

[0035] Compared with the prior art, the beneficial effect of the present invention is that the technical solution of the present invention determines two data extraction methods according to the comparison results of data richness and preset data richness, which avoids the problem of over-extraction or omission of key data caused by the single data extraction method in the prior art, thereby improving the accuracy and efficiency of data collection.

[0036] Furthermore, the technical solution of this invention determines the data matching degree by using historical incidence rate and rhythm disorder coefficient, avoiding the inability to accurately determine the data matching degree due to a single influencing factor, thereby improving the accuracy of the data matching degree. This facilitates subsequent adjustment of weights, thereby improving the accuracy and processing efficiency of medical record matching.

[0037] Furthermore, in the technical solution of the present invention, the supplementary selection method is determined based on the comparison result of the correlation between the effective data and the preset effective data, which avoids the fact that a single supplementary selection method is difficult to meet the actual application needs, thereby improving the rationality of the supplementary data. The weight adjustment method corresponding to the correlation between the effective data is determined based on the diagnostic and treatment effectiveness coefficient, thereby improving the rationality of the weight adjustment, and thus improving the accuracy of medical record matching and processing efficiency. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the AI-based medical decision-making weight allocation method of the present invention;

[0039] Figure 2 This is a flowchart illustrating how the present invention determines the data extraction method based on data abundance.

[0040] Figure 3 This is a flowchart illustrating the supplementary selection method based on the relevance of valid data in this invention. Detailed Implementation

[0041] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0042] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0043] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0044] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0045] Please see Figures 1 to 3 As shown, this invention provides an AI-based method for weight allocation in medical decision-making, comprising:

[0046] The data extraction method is determined based on the data abundance: either by extending the selection from multiple data sources or by supplementing the data based on the effective relevance.

[0047] When selecting multiple data sources, the data matching degree is determined based on historical incidence rate and rhythm disorder coefficient, and data with a matching degree greater than the preset matching degree are used as supplementary data.

[0048] The weight coefficients of relevant parameters for data matching degree are determined based on the critical threshold; the weight of any relevant parameter that is greater than the critical threshold is increased, and the increase is determined based on the effective difference.

[0049] Among them, the relevant parameters are the difference in historical incidence rates or the difference in rhythm disorder coefficients. The difference in historical incidence rates between the target medical record and the control medical record is recorded as the difference in historical incidence rates, and the difference in rhythm disorder coefficients between the target medical record and the control medical record is recorded as the difference in rhythm disorder coefficients.

[0050] The supplementary selection method is determined based on the comparison results between the relevance of valid data and the preset relevance of valid data. The selection method is either based on the selection of medical record data according to the feature keywords or based on the dynamic difference coefficient.

[0051] The weight adjustment method for the correlation of effective data is determined based on the comparison between the effective coefficient and the preset effective coefficient. This method is either based on the baseline weight allocation method or based on the difference in the effectiveness of treatment.

[0052] Data sources include medical records, laboratory data, and imaging reports;

[0053] When the weight adjustment amplitude reaches the preset adjustment amplitude, the weight adjustment ends.

[0054] The application scenario of this invention is to use AI to screen and match medical records for medical decision-making, so as to improve the efficiency of assisted decision-making.

[0055] Specifically, the data extraction method is determined based on data abundance, and the extraction methods include:

[0056] If the data abundance is greater than the preset data abundance, the data extraction method will be extended selection from multiple data sources;

[0057] If the data abundance is less than or equal to the preset data abundance, the data extraction method is to supplement the data based on the effective relevance.

[0058] The method for confirming data adequacy is to detect the number of people suffering from the target disease for that disease, and the data adequacy is calculated as: number of people suffering from the disease / total population.

[0059] The preset data sufficiency value can be set adaptively by the user according to the actual application scenario. It can be understood that the higher the preset data sufficiency value, the greater the user's need to extend the selection from multiple data sources. A preset data sufficiency value is provided, and the historical records of the user's extended selection from multiple data sources are detected. The average value of the data sufficiency corresponding to the historical records that can meet the user's needs is recorded as the preset data sufficiency value.

[0060] Specifically, in response to the first preset supplementation condition, the data matching degree is determined based on the historical incidence rate and the rhythm disorder coefficient, and data with a matching degree greater than the preset data matching degree are used as data to be supplemented.

[0061] The first preset supplementary condition is to determine that the data extraction method is a multi-source data source extended selection.

[0062] Data matching degree = α1 × historical incidence rate + α2 × rhythm disorder coefficient; where α1 is the first weight coefficient and α2 is the second weight coefficient. The values ​​of α1 and α2 can be set directly by the user based on domain experience, or by using statistical methods such as regression analysis or principal component analysis (PCA) to determine the contribution of historical incidence rate and rhythm disorder coefficient to data matching degree, thereby determining the corresponding weight coefficient values. The greater the contribution, the larger the weight coefficient value. The weights can also be adjusted through training on historical records (such as machine learning).

[0063] This invention detects the incidence rate of a target disease in several patients. Specifically, for each patient with the target disease, the incidence rate is measured from the initial diagnosis to the most recent medical visit. The incidence rate is calculated as the number of occurrences divided by the interval, where the interval is the time between the most recent medical visit and the initial diagnosis of the target disease. The average incidence rate of the several patients is recorded as the historical incidence rate. The number of patients can be adaptively set by the user according to actual needs. It is understood that the higher the accuracy requirement for the historical incidence rate, the larger the number of patients should be. This invention provides a value for the number of patients: 500 patients.

[0064] The rhythm disorder coefficient is calculated as follows: |Peak coefficient corresponding to the initial consultation time of the target disease - Preset peak coefficient|. The preset peak coefficient is determined by setting the target reference value corresponding to the initial consultation time of the target disease as a1, and the average peak coefficient of the target reference value data corresponding to the target reference value a1 corresponding to the initial consultation time in the historical records as the preset peak coefficient. The peak coefficients for different diseases are different, and users can make adaptive settings according to actual application needs.

[0065] The preset data matching degree can be set by the user according to the actual application needs. It can be understood that the higher the user’s requirement for the matching degree of the data to be supplemented, the larger the preset data matching degree will be. This invention provides a way to set the preset data matching degree by detecting the average value of the data matching degree in the historical records that meet the user’s needs and recording it as the preset data matching degree.

[0066] Specifically, the weight coefficients of relevant parameters of data matching degree are determined based on the critical threshold; the weight of any relevant parameter that is greater than the critical threshold is increased, and the increase is determined based on the effective difference.

[0067] Among them, the relevant parameters are the difference in historical incidence rates or the difference in rhythm disorder coefficients. The difference in historical incidence rates between the target medical record and the control medical record is recorded as the difference in historical incidence rates, and the difference in rhythm disorder coefficients between the target medical record and the control medical record is recorded as the difference in rhythm disorder coefficients.

[0068] The value of the critical threshold can be adaptively set by the user according to the actual application requirements. It can be understood that the higher the user's requirements for the accuracy of data matching, the smaller the value of the critical threshold. This invention provides a method for determining the value of the critical threshold, which extracts the corresponding threshold from the historical records that meet the user's requirements, filters out outliers, and records the average value of the thresholds after removing outliers as the critical threshold.

[0069] The effective difference is the larger of the historical incidence rate difference and the rhythm disorder coefficient difference;

[0070] The weights of effective differences greater than the critical threshold are increased based on the base weights. In this invention, the base weights are α1=0.5 and α2=0.5, and the increase is equal to the adjustment coefficient × (effective difference - critical threshold). The user can adaptively set the value of the adjustment coefficient according to the actual application requirements. It can be understood that the greater the influence of the difference between the effective difference and the critical threshold on the increase, the larger the value of the adjustment coefficient. This invention provides a value for the adjustment coefficient, which is 0.5.

[0071] Specifically, in response to the second preset supplementary condition, the supplementary selection method is determined based on the comparison result between the correlation of effective data and the preset correlation of effective data;

[0072] If the relevance of the effective data is less than the preset relevance of the effective data, then the supplementary selection method is determined to be to select medical record data based on feature keywords;

[0073] If the relevance of the effective data is greater than or equal to the preset relevance of the effective data, then the supplementary selection method is determined to be to select medical record data based on the dynamic difference coefficient;

[0074] The second preset supplementary condition is to determine the data extraction method as supplementing data based on effective relevance.

[0075] Machine learning is used to identify keywords in medical records. Data that meet the format "keyword: value" are recorded as a combination and this combination is recorded as a feature keyword. For example, body temperature is a keyword in medical record information, and body temperature: 36.5℃ is a feature keyword.

[0076] Data containing characteristic keywords in the medical record scheme are categorized as first-class selection data. First-class selection data is selected in descending order of the number of characteristic keywords until the number of selected first-class selection data reaches a preset quantity. The number of characteristic keywords refers to the number of characteristic keywords contained in a single medical record. Medical record data is selected based on a dynamic difference coefficient, including: medical record data with a dynamic difference coefficient greater than a preset dynamic difference coefficient from the target medical record are categorized as second-class selection data. Second-class selection data is selected in descending order of the dynamic difference coefficient until the number of selected second-class selection data reaches a preset quantity. The preset quantity can be adaptively set by the user according to the actual application scenario. This invention provides a method for determining the preset quantity by extracting the corresponding quantity from historical records that meet the user's needs, filtering out outliers, and recording the average value of the quantity after removing outliers as the preset quantity.

[0077] The dynamic difference coefficient between a single matched medical record and the target medical record is the average of the fluctuation differences of each feature keyword corresponding to the matched medical record and the target medical record. The fluctuation difference of each feature keyword corresponding to the matched medical record and the target medical record = |the diagnosis and treatment fluctuation value of the feature keyword corresponding to the matched data - the diagnosis and treatment fluctuation value of the feature keyword corresponding to the target medical record|.

[0078] Treatment fluctuation value = standard deviation of parameter value under the consultation time corresponding to a single matched data point / percentage of decrease frequency, percentage of decrease frequency = number of consultation times with decrease / (number of consultation times in a single matched data point - 1).

[0079] Different diseases correspond to different parameter values, which include, but are not limited to, heart rate, blood sugar, and tumor size.

[0080] The preset effective data relevance value can be adaptively set by the user according to the actual application needs. It can be understood that the user determines two different supplementary selection methods based on the comparison results of the effective data relevance and the preset effective data relevance. This invention provides a method for setting the preset effective data relevance value, extracts the corresponding effective data relevance from the historical records that meet the user's needs, filters out outliers, and records the average value of the effective data relevance after removing outliers as the preset effective data relevance.

[0081] Specifically, the relevance of effective data is determined based on keyword repetition and keyword proportion;

[0082] The relevance of the effective data is positively correlated with both keyword repetition and keyword proportion.

[0083] Effective data relevance = β1 × keyword repetition + β2 × keyword proportion; where β1 is the first weight coefficient and β2 is the second weight coefficient. The keyword repetition is determined by the number of identical keywords between the target medical record data and any matching medical record data. Keyword repetition = number of identical keywords / number of keywords in the target medical record data; Keyword proportion = number of keywords in the matching medical record data / number of characters in the matching medical record data.

[0084] Specifically, the weighting adjustment method corresponding to the relevance of effective data is determined based on the diagnostic effectiveness coefficient;

[0085] If the effective treatment coefficient is greater than the preset effective treatment coefficient, then the allocation will be based on the benchmark weighting method;

[0086] If the effective coefficient of diagnosis and treatment is less than or equal to the preset effective coefficient of diagnosis and treatment, the weight adjustment amplitude shall be determined according to the difference in the effectiveness of diagnosis and treatment.

[0087] The adjustment amplitude is positively correlated with the difference in diagnostic and treatment effectiveness.

[0088] Treatment effectiveness difference = Preset treatment effectiveness coefficient - Treatment effectiveness coefficient;

[0089] The user can adaptively set the value of the preset treatment effectiveness coefficient according to actual needs. It can be understood that the treatment effectiveness coefficient reflects the treatment effect. When the treatment effectiveness coefficient is greater than the preset treatment effectiveness coefficient, it indicates that the treatment effect is better. Therefore, the weight allocation method is not adjusted. This invention provides a value of the benchmark weight. In this invention, β1 is 0.5 and β2 is 0.5.

[0090] When the treatment effectiveness coefficient is less than or equal to the preset treatment effectiveness coefficient, the first weight coefficient is increased. The first weight coefficient = 0.5 + increase value; where the increase value = proportional coefficient × treatment effectiveness difference. The value of the proportional coefficient can be adaptively set by the user according to the actual application needs. It can be understood that the greater the influence of the treatment effectiveness difference on the increase value, the larger the value of the proportional coefficient. The ratio of the increase value to the treatment effectiveness difference corresponding to the historical records that meet the user's needs is extracted, outliers are filtered out, and the average value of the ratio after removing outliers is recorded as the proportional coefficient.

[0091] It is understandable that the first weighting coefficient + the second weighting coefficient = 1, and the increase in the first weighting coefficient is the same as the decrease in the second weighting coefficient.

[0092] Specifically, the treatment effectiveness coefficient is determined based on the cure rate and recurrence rate;

[0093] The effective treatment coefficient and the cure rate are positively correlated.

[0094] The effective treatment coefficient and the recurrence rate are negatively correlated.

[0095] The treatment effectiveness coefficient = cure rate + 1 / recurrence rate. In the medical record data of the target patient, if the key parameter data indicators are met and the duration of achievement is greater than the preset duration, then the target patient is recorded as a cured patient. The cure rate = number of cured patients / total number of patients. The probability of a cured patient relapsing within the preset recovery time is recorded as the recurrence rate. The value of the preset recovery time can be adaptively set by the user according to the actual application needs. It can be understood that different types of diseases correspond to different recovery times, and the recovery time for chronic diseases is set to 1 year.

[0096] Specifically, data sources include medical records, laboratory data, and imaging reports.

[0097] Specifically, the weight adjustment ends when the weight adjustment amplitude reaches the preset adjustment amplitude.

[0098] The preset adjustment amplitude value can be adaptively set by the user according to actual application needs. This invention provides a method for setting the preset adjustment amplitude value by extracting the corresponding adjustment amplitude value from the historical records that meet the user's needs, filtering out outliers, and recording the average value of the adjustment amplitude after removing outliers as the preset adjustment amplitude value. In this invention, the preset adjustment amplitude value is 60% of the base weight.

[0099] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. An AI-based method for weighting medical decisions, characterized in that, include: The data extraction method is determined based on the data abundance: either by extending the selection from multiple data sources or by supplementing the data based on the effective relevance. When selecting multiple data sources, the data matching degree is determined based on historical incidence rate and rhythm disorder coefficient, and data with a matching degree greater than the preset matching degree are used as supplementary data. The weight coefficients of relevant parameters for data matching degree are determined based on the critical threshold; the weight of any relevant parameter that is greater than the critical threshold is increased and adjusted, and the adjustment range is determined based on the effective difference; the relevant parameters are the difference in historical incidence rate or the difference in rhythm disorder coefficient. The difference in historical incidence rate between the target medical record and the comparison medical record is recorded as the difference in historical incidence rate, and the difference in rhythm disorder coefficient between the target medical record and the comparison medical record is recorded as the difference in rhythm disorder coefficient. The supplementary selection method is determined based on the comparison results between the relevance of valid data and the preset relevance of valid data. The selection method is either based on the selection of medical record data according to the feature keywords or based on the dynamic difference coefficient. The weight adjustment method for the correlation of effective data is determined based on the comparison between the effective coefficient and the preset effective coefficient. This method is either based on the baseline weight allocation method or based on the difference in the effectiveness of treatment. Data sources include medical records, laboratory data, and imaging reports; When the weight adjustment amplitude reaches the preset adjustment amplitude, the weight adjustment ends. The method for confirming data adequacy is to detect the number of people suffering from the target disease for that disease, and the data adequacy is calculated as: number of people suffering from the disease / total population. Effective data relevance = β1 × keyword repetition + β2 × keyword proportion; where β1 is the first weight coefficient, β2 is the second weight coefficient, and the keyword repetition is determined by the number of identical keywords between the target medical record data and any matching medical record data. Keyword repetition = number of identical keywords / number of keywords in the target medical record data; Keyword proportion = number of keywords in the matching medical record data / number of characters in the matching medical record data. Rhythm Disorder Coefficient = |Peak Coefficient corresponding to the first visit time of the target disease - Preset Peak Coefficient|; The treatment effectiveness coefficient = cure rate + 1 / recurrence rate. In the medical record data of the target patient, if the key parameter data indicators are met and the duration of achievement is greater than the preset duration, then the target patient is recorded as a cured patient. The cure rate = number of cured patients / total number of patients. The probability of a cured patient relapsing within the preset recovery time is recorded as the recurrence rate. The number of recurrences of several patients with the target disease is detected. For a single patient with the target disease, the number of recurrences from the initial diagnosis of the target disease to the most recent visit is detected, and the number of recurrences / the interval is recorded as the patient's incidence rate. The interval is the time from the most recent visit to the time of initial diagnosis containing the target disease. The average incidence rate of several patients is recorded as the historical incidence rate.

2. The AI-based medical decision-making weight allocation method according to claim 1, characterized in that, The data extraction method is determined based on data abundance. Extraction methods include: If the data abundance is greater than the preset data abundance, the data extraction method will be extended selection from multiple data sources; If the data abundance is less than or equal to the preset data abundance, the data extraction method is to supplement the data based on the effective relevance.

3. The AI-based medical decision-making weight allocation method according to claim 2, characterized in that, Under the first preset supplementation condition, the data matching degree is determined based on the historical incidence rate and the rhythm disorder coefficient, and the data with a matching degree greater than the preset data matching degree is regarded as data to be supplemented. The first preset supplementary condition is to determine that the data extraction method is a multi-source data source extended selection.

4. The AI-based medical decision-making weight allocation method according to claim 3, characterized in that, The weight coefficients of relevant parameters for data matching degree are determined based on the critical threshold; the weight of any relevant parameter that is greater than the critical threshold is increased, and the increase is determined based on the effective difference. Among them, the relevant parameters are the difference in historical incidence rates or the difference in rhythm disorder coefficients. The difference in historical incidence rates between the target medical record and the control medical record is recorded as the difference in historical incidence rates, and the difference in rhythm disorder coefficients between the target medical record and the control medical record is recorded as the difference in rhythm disorder coefficients.

5. The AI-based medical decision-making weight allocation method according to claim 2, characterized in that, In response to the second preset supplementary condition, the supplementary selection method is determined based on the comparison result between the correlation of effective data and the preset correlation of effective data; If the relevance of the effective data is less than or equal to the preset relevance of the effective data, then the supplementary selection method is determined to be to select medical record data based on feature keywords; If the relevance of the effective data is greater than the preset relevance of the effective data, then the supplementary selection method is determined to be to select medical record data based on the dynamic difference coefficient; The second preset supplementary condition is to determine the data extraction method as supplementing data based on effective relevance.

6. The AI-based medical decision-making weight allocation method according to claim 5, characterized in that, Determine the relevance of effective data based on keyword repetition and keyword proportion; The relevance of the effective data is positively correlated with both keyword repetition and keyword proportion.

7. The AI-based medical decision-making weight allocation method according to claim 6, characterized in that, The weighting adjustment method corresponding to the relevance of effective data is determined based on the effective coefficient of diagnosis and treatment. If the effective treatment coefficient is greater than the preset effective treatment coefficient, then the allocation will be based on the benchmark weighting method; If the effective coefficient of diagnosis and treatment is less than or equal to the preset effective coefficient of diagnosis and treatment, the weight adjustment amplitude shall be determined according to the difference in the effectiveness of diagnosis and treatment. The adjustment amplitude is positively correlated with the difference in diagnostic and treatment effectiveness.

8. The AI-based medical decision-making weight allocation method according to claim 7, characterized in that, The treatment effectiveness coefficient is determined based on the cure rate and recurrence rate. The effective treatment coefficient and the cure rate are positively correlated. The effective treatment coefficient and the recurrence rate are negatively correlated.

9. The AI-based medical decision-making weight allocation method according to claim 8, characterized in that, When the weight adjustment amplitude reaches the preset adjustment amplitude, the weight adjustment ends.

Images