Methods, media and equipment for colorectal cancer screening based on big data analysis
Through big data analysis and neural network models, the problem that the existing technology cannot effectively track and evaluate potential patients' personalized living habits, and provides colorectal cancer evaluation reports to help users reduce the risk of disease.
Patent Information
- Application Number
- CN202411125626.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-16
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-08-16
AI Technical Summary
The existing colorectal cancer screening programs are unable to effectively track and evaluate potential patients' personalized living and eating habits, resulting in the inability to provide long-term prevention and treatment recommendations, and lack of support for early screening and prevention of colorectal cancer.
Using a method based on big data analysis, by obtaining user information, generating preset screening conditions, performing database screening and clustering analysis, using neural network models to calculate the inducible probability of colorectal cancer, generating colorectal cancer evaluation reports, and providing personalized Diet and living habits advice.
Accurate assessment and personalized report on the user's probability of colorectal cancer are achieved, helping users to intervene in time to prevent measures and reduce the risk of colorectal cancer.
Smart Images

Figure CN118658632B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of big data technology, and in particular to a method, medium and equipment for screening a population for colorectal cancer based on big data analysis. Background Art
[0002] In the middle and late stages of colorectal cancer, not only does it cost patients a lot of money, but it also brings them great pain. However, a large number of studies and practices have shown that colorectal cancer screening, early diagnosis and treatment, and scientific follow-up intervention methods can effectively reduce the incidence and mortality of colorectal cancer. Even the 5-year survival rate after early colorectal cancer treatment can exceed 95%.
[0003] In the early stages, the main means of colorectal cancer screening is colonoscopy. The existing colorectal cancer screening program screens and tracks potential colorectal cancer populations from a third-party perspective, but this technical solution still cannot provide effective help for every potential patient. The reason is that every potential patient needs to cooperate in making changes in lifestyle and eating habits in the early stages of prevention and treatment. The existing screening program focuses on data processing and does not provide long-term tracking and advice for each potential patient. Summary of the invention
[0004] In view of the above problems, the present invention provides a method, medium and equipment for colorectal cancer population screening based on big data analysis, which solves the problem of lack of effective tracking and evaluation of colorectal cancer in potential patients in the early stages of the disease.
[0005] To achieve the above objectives, in a first aspect, the present invention provides a method for colorectal cancer screening based on big data analysis, comprising:
[0006] Obtain user information, including user name, ID number, age, permanent location, and diet information;
[0007] Generate a first preset screening condition according to the user information, and screen the database according to the first preset screening condition to obtain a first analysis data set, wherein the database includes initial predisposing data of historically diagnosed patients, the initial predisposing data includes initial predisposing information of multiple predisposing categories that induce colorectal cancer, and the predisposing categories include any one of age factors, geographical location factors, dietary factors, exercise factors, and genetic factors;
[0008] Performing cluster analysis on the first analysis data set according to preset classification conditions, the preset classification conditions including multiple classification levels, obtaining multiple inducement groups, each inducement group including classification feature data of multiple inducement categories at the same classification level, the classification feature data including initial inducement information and inducement categories;
[0009] Extracting an inducement group with the same similarity to the current user information from multiple inducement groups according to the user information, recording it as a first inducement group to be analyzed, and recording the classified feature data in the first inducement group to be analyzed as first feature data to be analyzed, wherein the first feature data to be analyzed includes first inducement information to be analyzed and a first inducement category to be analyzed;
[0010] Inputting the first predisposing factor group to be analyzed into the trained first neural network model, the output result of the first neural network model is the probability parameter of inducing colorectal cancer under each first predisposing factor category in the first predisposing factor group to be analyzed;
[0011] Collecting first actual inducement data of the user within a first preset time period corresponding to a current sampling period to obtain a second analysis data set, wherein the second analysis data set includes first actual inducement information of multiple inducement categories that induce colorectal cancer;
[0012] Screening the second analysis data set according to the first inducement category to be analyzed to obtain a second inducement group to be analyzed, the second inducement group to be analyzed includes second feature data to be analyzed, and the second feature data to be analyzed includes second inducement information to be analyzed;
[0013] The second predisposing factor group to be analyzed is input into the trained second neural network model, and the output result of the second neural network model is the adjusted weight value of the probability parameter of inducing colorectal cancer under each first predisposing factor category to be analyzed in the second predisposing factor group to be analyzed;
[0014] Obtaining the actual probability of colorectal cancer inducing of the user according to the probability parameter, the adjustment weight value and the second inducement group to be analyzed;
[0015] Adjusting the length of the first preset time period corresponding to the next sampling period according to the actual induction probability;
[0016] Generate a colorectal cancer assessment report based on the actual induction probability.
[0017] Further, a first preset screening condition is generated according to the user information, and the database is screened according to the first preset screening condition to obtain a first analysis data set including:
[0018] Generate a first screening condition according to the age information, screen out the initial inducement data satisfying the first screening condition in the database, record them as the first initial inducement data, and obtain a first initial inducement data set;
[0019] Generating a second screening condition according to the resident geographic location information, screening out second initial inducement data satisfying the second screening condition from the first initial inducement data set, and recording the second initial inducement data set;
[0020] A third screening condition is generated according to the dietary information, and the third initial inducement data satisfying the third screening condition is screened out from the second initial inducement data set to obtain a first analysis data set.
[0021] Furthermore, the method also includes:
[0022] Determine, based on the user information, whether there is a first preset historical time period associated with the current user information, where the first preset historical time period is the first preset time period of the current user in the previous sampling cycle;
[0023] If yes, the first preset time period in history is used as the first preset time period corresponding to the current sampling period, and the incentive data within the first preset time period corresponding to the current user information in the current sampling period up to now is obtained as the first actual incentive data;
[0024] If not, the initial cause information of the first analysis data set is sorted according to the severity of the colorectal cancer symptoms of the historically diagnosed patients, and a first preset time period corresponding to the sampling period of the historically diagnosed patients within the first preset disease threshold range is obtained, and recorded as a reference first preset time period;
[0025] Calculate the average of multiple reference first preset time periods, and use the calculation result as the first preset time period corresponding to the current user in the current sampling period, and obtain the incentive data within the first preset time period corresponding to the current user information in the current sampling period up to now as the first actual incentive data;
[0026] The user information also includes genetic disease history and past medical history. The first preset disease threshold is obtained by the following steps:
[0027] The threshold adjustment factor is obtained based on the genetic history and past medical history;
[0028] Obtaining a first basic disease threshold of the current user according to age information;
[0029] The first preset disease threshold is calculated based on the first basic disease threshold and the threshold adjustment factor.
[0030] Furthermore, collecting the first actual incentive data of the user within the first preset time period corresponding to the current sampling period to obtain the second analysis data set includes:
[0031] Dividing the first preset time period to obtain a plurality of second preset time periods;
[0032] Allocating the first actual inducement data according to a second preset time period to obtain a plurality of second actual inducement data;
[0033] Generate an inducement score according to the second actual inducement data, each second actual inducement data corresponds to an inducement score;
[0034] The plurality of second actual inducement data are sorted according to the inducement scores, and the second actual inducement data within a preset inducement threshold range are eliminated to obtain a second analysis data set.
[0035] Furthermore, the method also includes:
[0036] Obtaining medical information, which includes the symptoms and test reports of the current user under the colorectal cancer category, and the test reports include the values of multiple test indicators;
[0037] Generate initial diagnosis information based on the medical information and display it on the doctor's side, obtain the modified content on the doctor's side, and generate final diagnosis information based on the modified content and the initial diagnosis information;
[0038] According to the final diagnosis information, the treatment information and recovery information of the user under the same final diagnosis information are screened in the database, where the recovery information includes the recovery diet information and / or the recovery exercise information;
[0039] Generate and display recovery strategy recommendations within the first preset treatment cycle based on treatment information, post-recovery information and user information.
[0040] Furthermore, the final diagnosis information includes symptom image information, and the method further includes:
[0041] Acquiring lesion distribution information in the disease image information, the lesion distribution information including at least one of the canceration point of colon cancer, the degree of colon canceration, the canceration point of rectal cancer, and the degree of rectal canceration;
[0042] generating first predicted inducement data within a first preset treatment cycle according to the first actual inducement data;
[0043] Generate colorectal cancer simulation evolution information of the current user according to the first predicted cause data and the symptom image information, and dynamically display the simulation evolution information, wherein the colorectal cancer simulation evolution information includes a predicted evolution animation image of the cancerous symptom degree of the current user under the first predicted cause data;
[0044] The method also includes:
[0045] Obtain the medical information associated with the current user based on the user information. If the current user has visited the doctor multiple times, obtain the historical test report of the current user in the previous medical record and the real-time test report of the current user under the current medical record based on the user information;
[0046] Compare historical inspection reports and real-time inspection reports to obtain the numerical change trends of multiple inspection indicators;
[0047] Determine one by one whether the numerical change trend of the inspection index is within the preset change trend range. If not, mark the inspection index that is not within the preset change trend as an abnormal index, and record the numerical change trend of the abnormal index as an abnormal change trend;
[0048] The simulated trend of abnormal indicators is generated according to the abnormal change trend, and the disease prediction results are generated and displayed according to the simulated trend of abnormal indicators.
[0049] Furthermore, the method also includes:
[0050] According to the first actual inducement data, a trigger condition for triggering the update of the adjustment weight value is obtained, wherein the trigger condition includes a geographical trigger condition, a time trigger condition and a pathological trigger condition, and the geographical trigger condition is obtained by the following steps:
[0051] Acquire the first geographical location information in the first actual inducement data, and arrange the first geographical location information in the first actual inducement data in chronological order;
[0052] Determine whether the plurality of first geographic location information are consistent, and if not, obtain a change record of the first geographic location information;
[0053] It is determined whether the duration of the change record exceeds a preset time period. If so, a geographic trigger condition is generated, and the geographic trigger condition is used to update the adjustment weight value, and to update the first geographic location information to the permanent geographic location information in the user information.
[0054] Furthermore, extracting an inducement group having the same similarity with the current user information from multiple inducement groups according to the user information includes:
[0055] Performing vector transformation on each inducement group to obtain a group of multidimensional vectors corresponding to each inducement group, and recording the group of multidimensional vectors corresponding to each inducement group as a reference vector group to obtain multiple reference vector groups;
[0056] Perform vector conversion on the user information to obtain a set of multi-dimensional vectors corresponding to the user information, and record it as a user vector group;
[0057] Perform inner product calculations on the reference vector group and the user vector group one by one to obtain multiple inner product calculation results, each inner product calculation result corresponding to a reference vector group;
[0058] Calculate the modulus of the reference vector group one by one to obtain multiple reference vector moduli, calculate the modulus of the user vector group to obtain the user vector modulus;
[0059] Perform product operations on the reference vector modulus and the user vector modulus one by one to obtain multiple product calculation results, each product calculation result corresponds to a reference vector group;
[0060] Pairing the inner product calculation results with the product calculation results according to the reference vector group to obtain multiple similarity calculation groups;
[0061] Divide the inner product calculation result and the product calculation result in the same similarity calculation group to obtain a similarity value, and each reference vector group corresponds to a similarity value;
[0062] Arrange the multiple similarity values in order of their values, select the reference vector group corresponding to the similarity value ranked first, and record it as the first reference vector group to be analyzed;
[0063] The inducement group corresponding to the first reference vector group to be analyzed is obtained, and recorded as the first inducement group to be analyzed.
[0064] In a second aspect, the present invention further provides a computer-readable storage medium storing computer program instructions, which implement the method described in the first aspect when executed by a processor.
[0065] In a third aspect, the present invention further provides an electronic device, comprising a memory and a processor, wherein the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method described in the first aspect.
[0066] Different from the prior art, in the above technical solution, user information is obtained, a first preset screening condition is generated according to the user information, screening is performed in the database to obtain a first analysis data set, and then the first analysis data set is clustered and analyzed according to the preset classification condition to obtain multiple predisposing groups, and a predisposing group with consistent similarity with the user information is screened out from the multiple predisposing groups, recorded as the first predisposing group to be analyzed, and the first predisposing group to be analyzed is input into the first neural network model to obtain the probability parameter of inducing colorectal cancer under each first predisposing category to be analyzed in the first predisposing group to be analyzed. At the same time, the first actual predisposing data of the user in the first preset time period corresponding to the current sampling period is collected to obtain a second analysis data set, and the second analysis data set is screened to obtain a second predisposing group to be analyzed, and the second predisposing group to be analyzed is input into the second neural network model to obtain the adjustment weight value of the probability parameter of the second predisposing group to be analyzed inducing colorectal cancer under each first predisposing category to be analyzed, and the probability parameter, the adjustment weight value and the second predisposing group to be analyzed are used to obtain the actual induction probability of the user's colorectal cancer, and a colorectal cancer evaluation report is generated according to the actual induction probability. This technical solution tracks the data generated by the user's living and eating habits by collecting user information and the first actual inducement data, and rationally uses the data of confirmed patients in the database to analyze the first actual inducement data generated by the current user to obtain the actual inducement probability of the current user. A reference colorectal cancer evaluation report can be output to the current user, which is convenient for the user to determine whether to seek medical treatment, the time of treatment, adjustment of eating habits, etc. according to the actual situation, and timely intervene in colorectal cancer prevention operations in the early stage to reduce the probability of colorectal cancer.
[0067] The above-mentioned records related to the content of the invention are only an overview of the technical solution of the present invention. In order to enable ordinary technicians in the field to more clearly understand the technical solution of the present invention, and then implement it according to the text of the specification and the contents recorded in the drawings, and in order to make the above-mentioned purpose and other purposes, features and advantages of the present invention easier to understand, the following is a description in combination with the specific implementation mode of the present invention and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0068] The drawings are only used to illustrate the principles, implementations, applications, features and effects of the specific embodiments of the present invention and other related contents, and shall not be considered as limiting the present invention.
[0069] In the drawings of the specification:
[0070] Figure 1 This is a diagram of the first step of the screening method according to a specific embodiment of the present invention;
[0071] Figure 2 This is a diagram of the second step of the screening method according to a specific embodiment of the present invention;
[0072] Figure 3 This is a diagram of the third step of the screening method according to a specific embodiment of the present invention;
[0073] Figure 4 This is a diagram of the fourth step of the screening method according to a specific embodiment of the present invention;
[0074] Figure 5 A schematic diagram of an electronic device for a screening method according to a specific embodiment of the present invention.
[0075] The reference numerals in the above drawings are described as follows:
[0076] 1. Electronic equipment;
[0077] 11. Memory;
[0078] 12. Processor. DETAILED DESCRIPTION
[0079] In order to explain in detail the possible application scenarios, technical principles, specific schemes that can be implemented, and the purposes and effects of the present invention, the following is a detailed description in combination with the specific embodiments listed and the accompanying drawings. The embodiments described herein are only used to more clearly illustrate the technical solutions of the present invention, and are therefore only used as examples, and cannot be used to limit the scope of protection of the present invention.
[0080] Reference to "embodiment" herein means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The word "embodiment" appearing in various places in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or association with other embodiments. In principle, in the present invention, as long as there is no technical contradiction or conflict, the various technical features mentioned in each embodiment can be combined in any way to form a corresponding implementable technical solution.
[0081] Unless otherwise defined, the technical terms used herein have the same meanings as those generally understood by those skilled in the art to which the present invention belongs; the use of relevant terms herein is only for describing specific embodiments and is not intended to limit the present invention.
[0082] In the description of the present invention, the term "and / or" is an expression used to describe the logical relationship between objects, indicating that three relationships may exist, for example, A and / or B, which means: A exists, B exists, and A and B exist at the same time. In addition, the character " / " in this article generally indicates that the objects before and after are in a logical relationship of "or".
[0083] In the present invention, terms such as "first" and "second" are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship of quantity, priority or sequence between these entities or operations.
[0084] Without further restrictions, in the present invention, the words "include", "comprises", "has" or other similar open-ended expressions used in sentences are intended to cover non-exclusive inclusion. These expressions do not exclude the presence of additional elements in the process, method or product including the elements, so that the process, method or product including a series of elements may include not only those limited elements, but also other elements not explicitly listed, or also include elements inherent to such process, method or product.
[0085] Similar to the understanding in the Examination Guidelines, in the present invention, expressions such as "greater than", "less than", "exceed" and the like are understood to exclude the number itself; expressions such as "above", "below", "within" and the like are understood to include the number itself. In addition, in the description of the embodiments of the present invention, "multiple" means more than two (including two), and similar expressions related to "multiple" are also understood in this way, such as "multiple groups", "multiple times", etc., unless otherwise clearly and specifically limited.
[0086] In the description of the embodiments of the present invention, the space-related expressions used, such as "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "vertical", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicate the orientation or position relationship based on the orientation or position relationship shown in the specific embodiments or drawings, and are only for the convenience of describing the specific embodiments of the present invention or facilitating the reader's understanding, and do not indicate or imply that the referred device or component must have a specific position, a specific orientation, or be constructed or operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present invention.
[0087] Unless otherwise expressly specified or limited, in the description of the embodiments of the present invention, the terms such as "install", "connect", "connect", "fix", "set" and the like used should be understood in a broad sense. For example, the "connection" can be a fixed connection, a detachable connection, or an integrated setting; it can be a mechanical connection, an electrical connection, or a communication connection; it can be a direct connection, or an indirect connection through an intermediate medium; it can be the internal connection of two elements or the interaction relationship between two elements. For those skilled in the art of the technical field to which the present invention belongs, the specific meanings of the above terms in the embodiments of the present invention can be understood according to the specific circumstances.
[0088] See also Figure 1 In a first aspect, the present invention provides a method for colorectal cancer screening based on big data analysis, comprising:
[0089] S11, obtaining user information, the user information including user name, ID number, age information, permanent location information and dietary information;
[0090] S12, generating a first preset screening condition according to the user information, and screening the database according to the first preset screening condition to obtain a first analysis data set, wherein the database includes initial predisposing data of historically diagnosed patients, the initial predisposing data includes initial predisposing information of multiple predisposing categories that induce colorectal cancer, and the predisposing categories include any one of age factors, geographical location factors, dietary factors, exercise factors, and genetic factors;
[0091] S13, performing cluster analysis on the first analysis data set according to preset classification conditions, the preset classification conditions including multiple classification levels, obtaining multiple inducement groups, each inducement group including classification feature data of multiple inducement categories at the same classification level, the classification feature data including initial inducement information and inducement categories;
[0092] S14, extracting an inducement group with the same similarity to the current user information from multiple inducement groups according to the user information, recording it as a first inducement group to be analyzed, and recording the classified feature data in the first inducement group to be analyzed as first feature data to be analyzed, wherein the first feature data to be analyzed includes first inducement information to be analyzed and a first inducement category to be analyzed;
[0093] S15, inputting the first predisposing factor group to be analyzed into the trained first neural network model, and the output result of the first neural network model is the probability parameter of inducing colorectal cancer under each first predisposing factor category in the first predisposing factor group to be analyzed;
[0094] S16, collecting first actual inducement data of the user within a first preset time period corresponding to the current sampling period to obtain a second analysis data set, the second analysis data set including first actual inducement information of multiple inducement categories inducing colorectal cancer;
[0095] S17, screening the second analysis data set according to the first inducement category to be analyzed to obtain a second inducement group to be analyzed, the second inducement group to be analyzed including second feature data to be analyzed, and the second feature data to be analyzed including second inducement information to be analyzed;
[0096] S18, inputting the second predisposing factor group to be analyzed into the trained second neural network model, and the output result of the second neural network model is the adjusted weight value of the probability parameter of inducing colorectal cancer under each first predisposing factor category to be analyzed in the second predisposing factor group to be analyzed;
[0097] S19. Obtain the actual induction probability of colorectal cancer of the user according to the probability parameters, the adjusted weight values and the second induction group to be analyzed; and adjust the length of the first preset time period corresponding to the next sampling cycle according to the actual induction probability; and generate a colorectal cancer evaluation report according to the actual induction probability.
[0098] It should be noted that the method steps shown in this embodiment are suitable for people who have noticed their own colorectal cancer or are willing to track their own colorectal cancer symptoms on their own. This allows users to track their own eating habits and living habits, and pay more attention to changes and adjustments in their own habits to reduce the probability of colorectal cancer, or, if colorectal cancer is diagnosed, to alleviate the symptoms of colorectal cancer.
[0099] Specifically, in step S11, the user information may be information data input by the user himself, or may be user information obtained by associating with databases of other platforms after the user registers a user account. For example, the user may associate the user information with the database corresponding to the takeaway platform and the hospital database. When the information of the current user needs to be obtained, the permanent geographical location information and dietary information of the current user may be obtained quickly and accurately through data interaction between multiple platforms. It should be noted that the age information is the real age information of the user, the permanent geographical location information may be understood as the geographical location information of the user's residence, and the dietary information may be understood as the user's three meals a day dietary records, as well as drinking water records, beverage records, etc. Because in real life, in addition to genetic factors, the probability of colorectal cancer is also related to eating habits, sleeping habits, exercise habits, and the residential area of the geographical location. For example, some users live in Sichuan and Chongqing, and their diet is mainly spicy, so they have a higher probability of gastrointestinal diseases, which is also one of the causes of colorectal cancer; for example, when the user stays up late for a long time and has a low exercise frequency, resulting in obesity, the probability of colorectal cancer will also increase. When quantitative changes lead to qualitative changes, that is when users find themselves feeling unwell or are diagnosed with colorectal cancer. Therefore, it is necessary to track users' daily eating habits. At the same time, it is also necessary to combine geographical factors with the water source and eating habits of the user's place of residence, and through consideration of multiple inducements, obtain a comprehensive assessment of the current user's probability of colorectal cancer.
[0100] In step S12, a first preset screening condition is generated according to the user information, as described below. In this embodiment, a database is introduced, which contains the initial predisposing data of users who have been diagnosed with colorectal cancer, as well as the initial predisposing data of users who are being observed. The database can be obtained by associating with the hospital database. By setting up the database, it is convenient to use the existing big data in the field of colorectal cancer to make a reasonable assessment of the status of the current user. Furthermore, the database is screened according to the first preset screening condition to obtain a first analysis data set. It should be noted that the first preset screening condition screens the initial predisposing data of historically diagnosed patients in the database, so that the analysis of the user's predisposing data can be better realized in the early stages of colorectal cancer. In this embodiment, the initial inducement data includes initial inducement information of multiple inducement categories that induce colorectal cancer, wherein the inducement categories include age factors, geographical location factors, dietary factors, exercise factors, and genetic factors. The initial inducement information can be embodied using different indicators according to actual needs. For example, dietary factors can be subdivided into several first-level subcategories of sour, sweet, bitter, spicy, and salty, and then specific degrees are distinguished in these first-level subcategories. Taking acidity as an example, if the user's dietary information contains a high proportion of pickled vegetables and dishes with high acidity, this proportion is corresponding to the acidity ratio, and the acidity is embodied in the form of a ratio. Then, the initial inducement information under the dietary factor includes the ratio of the five categories of sour, sweet, bitter, spicy, and salty. For another example, under the exercise factor, the initial inducement information can be embodied by the user's average daily calorie consumption. If the user has a fixed exercise every week, the user's average daily calorie consumption in this week can be obtained by collecting the calorie consumption for a week and averaging it. By analogy, different inducement classifications can correspond to initial inducement information in different embodied forms.
[0101] In step S13, the preset classification condition refers to distinguishing the initial inducement data of multiple inducement categories in the first analysis data set during the clustering process. The preset classification condition is one of the inducement categories. The preset classification condition can be preset or obtained after big data screening. Specifically, if the preset classification condition is preset, the preset classification condition includes multiple classification degrees, and the classification degree is the classification degree under this preset classification condition. For example, when the age factor is used as the preset classification condition, the classification degree of the preset classification condition is divided into an age span of 30-40, an age span of 40-50, an age span of 50-60, etc. Furthermore, taking the K-means cluster analysis model as an example, step S13 is specifically explained: the preset classification condition in step S13 is pre-set as the age factor in the inducement category, and the first data set to be analyzed is input into the K-means cluster analysis. The classification index of the K-means cluster analysis is the age factor, and the output result of the K-means cluster analysis model is the inducement group composed of different age groups: in the 30-40 age group, in addition to the inducement category of the age factor, the initial inducement information under other inducement categories is concentrated to form A classification cluster; in the 40-50 age group, in addition to the inducement category of the age factor, the initial inducement information under other inducement categories is concentrated to form a classification cluster; in the 50-60 age group, in addition to the inducement category of the age factor, the initial inducement information under other inducement categories is concentrated to form a classification cluster; the final inducement group is divided into 0-100 years old, and the total number of inducement groups is 10 groups, the first group is the inducement group of 0-10 years old, the second group is the inducement group of 10-20 years old, and so on. It can be known that the number of inducement groups is calculated according to the K value in the K-means cluster analysis model, which can be 3 groups or 5 groups. The length of the age group distinction can also be set according to actual needs, which can be clustered every 10 years or every 20 years. In some optional embodiments, the age groups of different inducement groups can also have different age intervals, such as the first inducement group has an age span of 10 years, and the second inducement group has an age span of 20 years.
[0102] In step S13, classification feature data of multiple predisposing categories under a unified classification degree can be obtained. For example, using age as a distinction, initial predisposing information of multiple predisposing categories under the age span of 30-40 is obtained. That is to say, in the predisposing group composed of the 30-40 age group, initial predisposing information of geographical factors, initial predisposing information of dietary factors, initial predisposing information of exercise factors, etc. are included. The pre-set preset classification conditions can be manually selected according to the doctor's analysis needs; or, after performing big data analysis on patients with historical diagnosis of colorectal cancer, a predisposing category that has the closest impact on colorectal cancer in the current status of colorectal cancer can be obtained, and this predisposing category can be used as a preset classification condition for distinction. This embodiment is not limited to this.
[0103] In step S14, based on the user information, a similarity analysis is performed on the classified multiple incentive groups, and the incentive group that is most similar to the current user information is extracted as the first incentive group to be analyzed. The specific similarity analysis process is described later. In this embodiment, the incentive group that is most similar to the user information of the current user is screened out from the multiple incentive groups, and then analyzed, so that the analysis result is more in line with the personal attributes of the current user, and the accuracy of the analysis result can also be improved.
[0104] Step S15 shows the analysis steps for the first predisposing factor group to be analyzed. Specifically, this embodiment constructs a first neural network model, the input data of the first neural network model is the first predisposing factor group to be analyzed, and the output result of the first neural network model is the probability parameter of inducing colorectal cancer under each first predisposing factor category to be analyzed in the first predisposing factor group to be analyzed. It should be noted that this probability parameter refers to the influence proportion of the first predisposing factor category to be analyzed in the entire colorectal cancer development process. For example, the first predisposing factor category to be analyzed includes geographical location factors, dietary factors, exercise factors and genetic factors. The probability parameter under dietary factors is 30%, the probability parameter under exercise factors is 10%, the probability parameter under genetic factors is 40%, and the probability parameter under geographical location factors is 20%. This percentage is obtained through the first neural network model. Combined with the above example, that is, assuming that the current user's age information is 34 years old, and the predisposing group of 30-40 is taken as the first predisposing group to be analyzed, the first neural network model can obtain the distribution of the factors that cause colorectal cancer in the 30-40 age group through the first predisposing group to be analyzed: the probability parameter under dietary factors is 30%, the probability parameter under exercise factors is 10%, the probability parameter under genetic factors is 40%, and the probability parameter under geographic location factors is 20%. It should be noted that the output of the first neural network model is the analysis of the proportion of various predisposing categories in the first predisposing group to be analyzed, and this proportion is reflected using probability parameters.
[0105] Furthermore, in this embodiment, it is also necessary to make a more accurate estimate of the user's probability of suffering from colorectal cancer in combination with the user's personal information.
[0106] Specifically, in step S16, the first actual inducement data of the user in the first preset time period corresponding to the current sampling period is collected. In this embodiment, the sampling period can be understood as the sampling period of the actual inducement data of the current user, and the first preset time period can be understood as the selection section of the actual inducement data. The sampling period can be in units of years, months, quarters, etc., and this embodiment is not limited to this. The first preset time period is the selection time section of the actual inducement data when executing the method steps shown in this embodiment, and the first preset time period is less than the sampling period. For example, the user has started tracking his own colorectal cancer related information two years ago, then the user's sampling period can be in units of years. In these two years, the user has long-term autonomous tracking of his own actual inducement data, including geographical location factors, dietary factors, exercise factors and genetic factors. So far, the user has two sampling periods of actual inducement data. The first preset time period can be 3 months. When the user wants to predict the probability of his own colorectal cancer, he can select the actual inducement data within 3 months so far, which is recorded as the first actual inducement data. The setting of the first preset time period can be pre-set, or it can be obtained by the method described below. In this embodiment, a second analysis data set is obtained based on the first actual predisposing data, and the second analysis data set includes the first actual predisposing information under multiple predisposing categories that induce colorectal cancer. It should be noted that the first actual predisposing data includes multiple categories, and there may be predisposing categories associated with colorectal cancer, and there may also be other noise data. To facilitate distinction, the first actual predisposing information of the predisposing categories associated with colorectal cancer is screened out from the first actual predisposing data, and the second analysis data set is obtained by sorting. The embodiment of the first actual predisposing information is the same as that of the initial predisposing information, and an adapted embodiment can be specifically selected according to different predisposing categories, such as calorie consumption, percentage value of dietary taste, etc.
[0107] In step S17, the second analysis data set is screened according to the first predisposing factor category to be analyzed. Specifically, the first predisposing factor category to be analyzed is the predisposing factor category contained in the first predisposing factor group to be analyzed. The second analysis data set contains the first actual predisposing factor information of all predisposing factor categories associated with colorectal cancer. In steps S13 and S14, the most similar predisposing factor group has been selected for user information. Therefore, the second analysis data set can be further screened using the first predisposing factor category to be analyzed to accurately divide the first actual predisposing factor information that is closely related to the user, so as to facilitate more accurate subsequent analysis. For ease of distinction, the result after the second screening is recorded as the second predisposing factor group to be analyzed, and the data in the second predisposing factor group to be analyzed is the personal data of the current user.
[0108] In step S18, the second predisposing factor group to be analyzed is input into the trained second neural network model, and the output result of the second neural network model is the adjusted weight value of the probability parameter of inducing colorectal cancer under each first predisposing factor category to be analyzed in the second predisposing factor group to be analyzed. In this step, the difference between the second neural network model and the first neural network model is that the input data of the second neural network model is the second predisposing factor group to be analyzed, and the second predisposing factor group to be analyzed is obtained after condition screening based on the actual predisposing data of the current user, and the output result of the second neural network model is the adjusted weight value of the probability parameter, so the first neural network model and the second neural network model are two different neural network models. Further, the second neural network model can output the adjusted weight value of the probability parameter under the first predisposing factor category to be analyzed, that is, in step S18, the objective factors such as the personal living habits and eating habits of the current user can be integrated through the actual predisposing data of the current user, and the adjusted weight value is output, so that the final actual induction probability can be derived by combining the initial predisposing information of the people who are more similar to the user in the big data and the actual predisposing information of the user himself, so that the actual induction probability corresponding to the current user is more accurate.
[0109] In step S19, probability parameters corresponding to multiple first inducement categories to be analyzed and adjustment weight values associated with each probability parameter are obtained, and the probability parameters of the first inducement category to be analyzed are weighted products with the adjustment weight values, so that the probability ratios of multiple inducement categories closely connected with the current user's own situation can be obtained. Furthermore, by matching and integrating the first actual inducement information in the second inducement group to be analyzed with the probability ratios of the inducement categories, the final actual inducing probability can be obtained. It should be noted that the matching and integration form of the first actual inducement information and the probability ratios of the inducement categories can be set according to actual needs. For example, the first actual inducement information can be converted into a vector and integrated with the probability ratios of the inducement categories in the form of vector products. The actual inducing probability finally obtained can be reflected by calculating the vector modulus of the final vector obtained by the cumulative sum of multiple vector products. This embodiment is not limited to this.
[0110] Furthermore, in step S19, after obtaining the actual induction probability, the first preset time period in the next sampling cycle can be adjusted. For example, when the actual induction probability is high, it indicates that the current user has a high probability of colorectal cancer and needs to be tracked more carefully, then the first preset time period can be extended from the existing 3 months to 6 months, or the first preset time period can be shortened from the existing 3 months to 2 months.
[0111] In step S19, the actual induction probability will also generate a colorectal cancer evaluation report. Specifically, the colorectal cancer evaluation report can apply the first actual inducement information in the second inducement group to be analyzed to make reasonable suggestions for the first inducement category to be analyzed. For example, if the actual induction probability is high because the adjustment weight value of the inducement category of dietary factors in step S18 is high, it means that there are unfavorable factors in the current user's eating habits. The unfavorable factors may be the user's change from a light diet to a diet that tends to be spicy and numb. The output colorectal cancer evaluation report may include suggestions for adjusting the user's eating habits; for example, in the exercise factor, because the user begins to reduce the frequency of exercise, the obesity level increases, then the output colorectal cancer evaluation report can make suggestions for adjusting exercise habits.
[0112] In some optional embodiments, the first neural network model can be trained and obtained by the following steps:
[0113] Construct a first basic neural network model, which may be a convolutional neural network model; the input sample of the first basic neural network model is the first sample inducement group in the first sample database, the first sample inducement group includes the first sample inducement category and the first sample inducement information filled in by the first sample population; the output result of the first basic neural network model is the first sample probability parameter. By repeatedly training the first basic neural network model, so that its accuracy reaches a preset first accuracy threshold, it means that the first basic neural network model has been trained, and the trained first basic neural network model is recorded as the first neural network model.
[0114] In some other optional embodiments, the second neural network model can be trained and obtained by the following steps:
[0115] Construct a second basic neural network model, which can be a convolutional neural network model; the input sample of the second basic neural network model is the second sample inducement group in the second sample database, and the second sample inducement group includes the second sample inducement category and the second sample inducement information filled in by the second sample population; the output result of the second basic neural network model is the second sample adjustment weight value. By repeatedly training the second basic neural network model, so that its accuracy reaches the preset second accuracy threshold, it means that the second basic neural network model has been trained, and the trained second basic neural network model is recorded as the second neural network model.
[0116] This embodiment tracks the data generated by the user's living and eating habits by collecting user information and the first actual inducement data, and reasonably uses the data of confirmed patients in the database to analyze the first actual inducement data generated by the current user to obtain the actual inducement probability of the current user. A reference colorectal cancer evaluation report can be output to the current user, which is convenient for the user to determine whether to seek medical treatment, the time of medical treatment, adjustment of eating habits, etc. according to the actual situation, and timely intervene in colorectal cancer prevention operations in the early stage to reduce the probability of colorectal cancer.
[0117] In some embodiments, generating a first preset screening condition according to user information, and screening the database according to the first preset screening condition to obtain a first analysis data set includes:
[0118] Generate a first screening condition according to the age information, screen out the initial inducement data satisfying the first screening condition in the database, record them as the first initial inducement data, and obtain a first initial inducement data set;
[0119] Generating a second screening condition according to the resident geographic location information, screening out second initial inducement data satisfying the second screening condition from the first initial inducement data set, and recording the second initial inducement data set;
[0120] A third screening condition is generated according to the dietary information, and the third initial inducement data satisfying the third screening condition is screened out from the second initial inducement data set to obtain a first analysis data set.
[0121] In this embodiment, the first preset screening condition includes age information, permanent geographical location information and dietary information. In this embodiment, age information is used as the first screening condition, and the first screening condition is used for screening in the database. It should be noted that the age information is the user's age information, and the first screening condition is generated based on the age information. The specific steps may be to pre-classify the age range of 0-100 years old according to growth characteristics, and divide it into five age categories: childhood, adolescence, youth, young and middle-aged, and old age, and determine the age category to which the user belongs based on the age information. At the same time, select the previous age category and the next age category of the user's age category as backup age categories to generate the first screening condition. For example, if the age category described by the user is young and middle-aged, then young and middle-aged, youth and old age are selected as the first screening condition, and screening is performed in the database to obtain the first initial inducement data.
[0122] Furthermore, a second filtering condition is generated based on the permanent geographic location information. The second filtering condition can be generated by: obtaining first administrative division information to which the permanent geographic location belongs, obtaining second administrative division information of the area surrounding the permanent geographic location based on the first administrative division data, and filtering out initial incentive data that is consistent with the second administrative division information and the first administrative division information from the first initial incentive data set, and recording them as the second initial incentive data set.
[0123] Further, the third screening condition is generated according to the dietary information. The specific generation steps may be: constructing a dietary similarity model, inputting the dietary information into the dietary similarity model, obtaining a dietary taste category with a high similarity to the current dietary information, using the dietary taste category as the third screening condition, and screening in the second initial inducement data set. The dietary similarity model can be evolved based on the Jaccard similarity model. Specifically, the existing dietary data is collected and integrated to form a dietary database, and the dietary database can be divided according to dietary taste to obtain multiple first dietary sets; the user's dietary information is converted into a second dietary set, and the intersection of the first dietary set and the second dietary set is calculated one by one, recorded as the first similarity factor, and the union of the first dietary set and the second dietary set is calculated one by one, recorded as the second similarity factor, and the first similarity factor is divided by the second similarity factor one by one to obtain multiple Jaccard similarity coefficients, and the multiple Jaccard similarity coefficients are arranged according to size, and the first dietary sets corresponding to the several Jaccard similarity coefficients ranked in front are selected, and the corresponding dietary taste categories are used as the third screening condition.
[0124] By screening the initial inducement data in the database in three dimensions according to age information, permanent geographic location information, and dietary information, a first analysis data set centered on the user information of the current user can be defined to facilitate the subsequent calculation of probability parameters.
[0125] See also Figure 2 In some embodiments, the method further comprises:
[0126] S21, judging whether there is a first preset historical time period associated with the current user information according to the user information, where the first preset historical time period is the first preset time period of the current user in the previous sampling period;
[0127] S22: If yes, use the historical first preset time period as the first preset time period corresponding to the current sampling period, and obtain the incentive data of the current user information in the first preset time period corresponding to the current sampling period up to now as the first actual incentive data;
[0128] S23, if not, sorting the initial cause information of the first analysis data set according to the severity of colorectal cancer symptoms of historically diagnosed patients, obtaining a first preset time period corresponding to the sampling period of historically diagnosed patients within the first preset disease threshold range, and recording it as a reference first preset time period;
[0129] S24, performing mean calculation on multiple reference first preset time periods, and using the calculation result as the first preset time period corresponding to the current user in the current sampling period, and obtaining the incentive data within the first preset time period corresponding to the current user information in the current sampling period up to now as the first actual incentive data;
[0130] The user information also includes genetic disease history and past medical history. The first preset disease threshold is obtained by the following steps:
[0131] The threshold adjustment factor is obtained based on the genetic history and past medical history;
[0132] Obtaining a first basic disease threshold of the current user according to age information;
[0133] The first preset disease threshold is calculated based on the first basic disease threshold and the threshold adjustment factor.
[0134] In step S21, it is determined whether the user information exists in the first historical preset time period associated with the current user information. It can be understood that when each user creates an account, his / her own data and sampling period are synchronously updated to the database, thereby realizing the collection and update of real-time data in the database. When obtaining user information, step S21 specifically also includes: matching the user information in the database to obtain the actual incentive data associated with the current user information, and the actual incentive data of the user includes the set duration of the first preset time period. For the convenience of expression, the first preset time period that has been set is recorded as the first historical preset time period.
[0135] In step S22, when the user already has a sampling cycle, that is, already has a setting record of the first preset time period, this historical first preset time period is used as the first preset time period corresponding to the current sampling cycle. It should be noted that the historical first preset time period refers to the first preset time period in the previous sampling cycle, and the user's previous sampling cycle and the current sampling cycle may not be continuous in actual time. For example, if the user has the first sampling cycle in 2021 and the second sampling cycle in 2023, there is no sampling cycle in 2022. Then, when the user sets the first preset time period in 2023, the setting duration of the first preset time period in the first sampling cycle of 2021 will be used.
[0136] It should be noted that in step S19, the first preset time period is adjusted using the actual induced probability, which is an adjustment when the sampling period is continuous, and the steps shown in this embodiment can be an adjustment when the sampling period is continuous, and can also be applied to the setting of the first preset time period when the sampling period is discontinuous. The adjustment method of the first preset time period can be set according to actual needs.
[0137] In step S23, if there is no historical first preset time period, that is, the current user is registering for the first time, and there is no collection record of actual cause data for a complete sampling cycle. Then, the initial cause information of the first analysis data set can be sorted, and the sorting basis is the severity of the colorectal cancer symptoms of the historically diagnosed patients, and the first preset time period corresponding to the sampling period of the historically diagnosed patients within the first preset disease threshold range is obtained, which is recorded as the reference first preset time period. There are multiple historically diagnosed patients within the first preset disease threshold range, that is, there are multiple reference first preset time periods.
[0138] In step S24, the average of multiple reference first preset time periods is calculated, and the calculation result is used as the first preset time period corresponding to the current user in the current sampling cycle. Then, the incentive data of the current user information in the first preset time period up to the present in the current sampling cycle is obtained as the first actual incentive data.
[0139] It should be noted that, in this embodiment, the first preset disease threshold is calculated by genetic history and past medical history. Specifically, the threshold adjustment factor is obtained according to the genetic history and past medical history. The specific calculation method can be: using the Roche risk assessment algorithm or the Mengelon risk assessment algorithm for the genetic history to obtain the corresponding risk level, such as setting three risk levels of medium risk, high risk, and low risk, and each risk level corresponds to a score or weight; for the past medical history, the logistic regression or decision tree algorithm in the statistical algorithm can be used, or the deep learning algorithm can be used to output a score or weight. Taking weight as an example, the weights of the genetic history and the past medical history are added and processed, and the final weight obtained is used as the threshold adjustment factor.
[0140] Furthermore, the first basic disease threshold of the current user is obtained according to the age information. In this embodiment, considering that colorectal cancer is greatly affected by age factors, the probability of disease significantly increases when the age is over 50 years old. Based on this, a basic disease threshold table is set. The independent variable of the basic disease threshold table is age, and the dependent variable is the probability of disease of colorectal cancer. The dependent variable can be obtained by screening big data. According to the user's age information, the first basic disease threshold can be obtained in the basic disease threshold table.
[0141] Furthermore, the first preset disease threshold is calculated based on the first basic disease threshold and the threshold adjustment factor, which can be specifically: the threshold adjustment factor is converted into a proportion value, and the first basic disease threshold is multiplied by the converted threshold adjustment factor to obtain the first preset disease threshold.
[0142] Then in step S24, the younger the age information recorded in the user information and the lower the first preset disease threshold, the more similar the historically diagnosed patients placed in the first preset disease threshold are to the inducement data of the current user information, and the reference first preset time period of the historically diagnosed patients has a certain reference value, which can be used as the setting duration of the first preset time period of the current user within this sampling period.
[0143] This embodiment reasonably selects and sets the first preset time period based on user information to make the screening of the first actual inducement data more accurate, facilitate the accurate selection of data in the second inducement group to be analyzed, and thus improve the accuracy of the user's actual inducement probability calculation.
[0144] See also Figure 3 In some embodiments, collecting the first actual inducement data of the user within the first preset time period corresponding to the current sampling period to obtain the second analysis data set includes:
[0145] S31, dividing the first preset time period into multiple second preset time periods;
[0146] S32, allocating the first actual inducement data according to a second preset time period to obtain a plurality of second actual inducement data;
[0147] S33, generating an inducement score according to the second actual inducement data, each second actual inducement data corresponds to an inducement score;
[0148] S34. Sort the plurality of second actual inducement data according to the inducement scores, and remove the second actual inducement data within a preset inducement threshold range to obtain a second analysis data set.
[0149] In step S31, the first preset time period is divided to obtain a plurality of second preset time periods, and the second preset time periods in the same first preset time period may be different or the same. For example, taking the first preset time period being divided equally as an example, when the first preset time period is 3 months, the first preset time period may be divided into 10-day segmentation intervals to obtain 6 second preset time periods, each of which is 10 days. Taking the unequal segmentation of the first preset time period as an example, when the first preset time period is 12 months a year, if the current user has an obvious movement trend distinction, that is, the user has less movement trends from January to March, a higher movement trend from April to August, a medium movement trend from September to October, and no movement in November and December, the first preset time period may be divided unequally according to the movement trend to obtain 4 second preset time periods, the first second preset time period being January to March, the second second preset time period being April to August, the third second preset time period being September to October, and the fourth second preset time period being November to December.
[0150] In step S32, the first actual incentive data is distributed according to the second preset time period to obtain a plurality of second actual incentive data.
[0151] In step S33, an incentive score is generated according to the second actual incentive data. It should be noted that the distribution of the incentive score can be preset or obtained by statistically analyzing the actual incentive data of users in the database.
[0152] In step S34, multiple second actual inducement data are sorted according to the inducement score, and the second actual inducement data placed within the preset inducement threshold range are proposed to obtain a second analysis data set. In this embodiment, the determination of the preset inducement threshold range can also be obtained in step S33, that is, in the process of generating the inducement score, the preset inducement threshold range can be set. When the preset inducement threshold range is exceeded, it means that the second actual inducement data in the inducement score has a greater impact on the user's own colorectal cancer probability. The second actual inducement data exceeding the preset inducement threshold range are integrated to obtain the second analysis data set. This method can reduce the processing volume of a large amount of first actual inducement data generated when the first preset time period is too long. After selectively screening the first actual inducement data, it can still maintain the tracking and selection of valid data in the second actual inducement data, thereby improving the speed and efficiency of generating the second analysis data set from the entire first actual inducement data.
[0153] In some embodiments, the method further comprises:
[0154] Obtaining medical information, which includes the symptoms and test reports of the current user under the colorectal cancer category, and the test reports include the values of multiple test indicators;
[0155] Generate initial diagnosis information based on the medical information and display it on the doctor's side, obtain the modified content on the doctor's side, and generate final diagnosis information based on the modified content and the initial diagnosis information;
[0156] According to the final diagnosis information, the treatment information and recovery information of the user under the same final diagnosis information are screened in the database, where the recovery information includes the recovery diet information and / or the recovery exercise information;
[0157] Generate and display recovery strategy recommendations within the first preset treatment cycle based on treatment information, post-recovery information and user information.
[0158] In this embodiment, the medical information can be obtained through the database in the hospital. The medical information is the symptoms of the current user's visit under the colorectal cancer category and the corresponding test report. The test report includes the values of multiple test indicators. This medical information generates initial diagnosis information and displays it on the doctor's side. The doctor can modify the initial diagnosis information according to the actual examination and medical situation. The modified part is recorded as the modification content, and the final diagnosis information is generated according to the modification content and the initial diagnosis information. It should be noted that existing hospitals have their own complete medical treatment and diagnosis system. The steps shown in this embodiment can also be completed in the hospital's own system, and then the final diagnosis information is updated in the form of information interaction to the database containing the initial inducement information in this embodiment.
[0159] According to the final diagnosis information, the treatment information and post-recovery information of the user under the same final diagnosis information are filtered in the database, and the post-recovery information includes post-recovery diet information and / or post-recovery exercise information. According to the treatment information, post-recovery information and user information, a recovery strategy recommendation within the first preset treatment cycle is generated and displayed. In this embodiment, after the diagnosis is confirmed, the first preset treatment cycle can be given by the doctor's side, and the treatment information includes the prescription information issued by the doctor's side and the healing information of the equipment and instruments, which can be used for reference by the user, so that the user can choose the appropriate treatment method and medicine according to the actual needs. Furthermore, the post-recovery information can provide a reference for the user's diet and exercise after the diagnosis, so that the user can consciously intervene in his diet and exercise at home.
[0160] This embodiment enables the user to adjust and select the recovery strategy within the first preset treatment cycle in a timely manner according to his or her own situation after diagnosis, thereby improving the user's healing efficiency.
[0161] In some embodiments, the final diagnosis information includes disease image information, and the method further includes:
[0162] Acquiring lesion distribution information in the disease image information, the lesion distribution information including at least one of the canceration point of colon cancer, the degree of colon canceration, the canceration point of rectal cancer, and the degree of rectal canceration;
[0163] generating first predicted inducement data within a first preset treatment cycle according to the first actual inducement data;
[0164] Generate colorectal cancer simulation evolution information of the current user according to the first predicted cause data and the symptom image information, and dynamically display the simulation evolution information, wherein the colorectal cancer simulation evolution information includes a predicted evolution animation image of the cancerous symptom degree of the current user under the first predicted cause data;
[0165] The method also includes:
[0166] Obtain the medical information associated with the current user based on the user information. If the current user has visited the doctor multiple times, obtain the historical test report of the current user in the previous medical record and the real-time test report of the current user under the current medical record based on the user information;
[0167] Compare historical inspection reports and real-time inspection reports to obtain the numerical change trends of multiple inspection indicators;
[0168] Determine one by one whether the numerical change trend of the inspection index is within the preset change trend range. If not, mark the inspection index that is not within the preset change trend as an abnormal index, and record the numerical change trend of the abnormal index as an abnormal change trend;
[0169] The simulated trend of abnormal indicators is generated according to the abnormal change trend, and the disease prediction results are generated and displayed according to the simulated trend of abnormal indicators.
[0170] In this embodiment, the final confirmed information includes symptom image information. Specifically, the symptom image information can obtain lesion distribution information through an image recognition algorithm, such as the image recognition algorithm model of yolov8. The lesion distribution information can specifically include at least one of the cancerous point of colon cancer, the degree of colon cancer, the cancerous point of rectal cancer, and the degree of rectal cancer.
[0171] The first predicted inducement data within the first preset treatment cycle is generated according to the first actual inducement data. Specifically, the first actual inducement data can be used as the dependent variable and the time data can be used as the independent variable through a hyperbola fitting model to obtain a two-dimensional coordinate distribution diagram with the first actual inducement data as discrete data points. Based on this, a first inducement prediction curve corresponding to the first actual inducement data is generated. In this first predicted inducement curve, a fitting arc segment within the first preset treatment cycle can be selected to convert it into the first predicted inducement data.
[0172] The colorectal cancer simulation evolution information of the current user is generated according to the first predicted predisposing data and the symptom image information, and then the simulation evolution information is dynamically displayed. Specifically, the symptom image information contained in the final confirmed information of the historically confirmed patients consistent with the first predicted predisposing data can be matched in the database according to the first predicted predisposing data, and for the convenience of distinction, it is recorded as the reference symptom image information, and the reference symptom image information whose lesion distribution information is consistent with the symptom image information of the current user is screened out from the reference symptom image information, and after arranging them in order according to the severity of the first predicted predisposing data, each of the screened reference symptom image information is used as a frame to synthesize the predicted evolution dynamic image and display it to the user, so as to deepen the user's concept of the current degree of colorectal cancer, and at the same time, it can have a warning effect on the user, and improve the user's treatment cooperation and consciousness in the next preset treatment cycle, so as to achieve a better healing effect.
[0173] In this embodiment, the medical information associated with the current user is obtained based on the user information, that is, the historical medical records of the current user are stored in the database, and each medical record corresponds to a medical information. If the current user has multiple medical visits, there are multiple medical information, and the historical test report of the user in the previous medical record and the real-time test report under the current medical visit are obtained.
[0174] Compare the data of historical test reports with those of implementation test reports to obtain the numerical change trends of multiple test indicators, and judge whether the numerical change trends of the test indicators are within the preset change trend range one by one. If not, the test indicators that are not within the preset change trend are marked as abnormal indicators, and the numerical change trends of the abnormal indicators are recorded as abnormal change trends; generate simulated trends of abnormal indicators based on the abnormal change trends, and generate and display disease prediction results based on the simulated trends of abnormal indicators. This method can predict the user's disease condition when there is no disease image information, and generate disease prediction results based on the simulated trends of abnormal indicators, so that users can deepen their understanding of the current degree of colorectal cancer, and at the same time, it can have a warning effect on users, improve the user's treatment cooperation and consciousness in the next preset treatment cycle, so as to achieve better healing effects.
[0175] See also Figure 4 In some embodiments, the method further comprises:
[0176] According to the first actual inducement data, a trigger condition for triggering the update of the adjustment weight value is obtained, wherein the trigger condition includes a geographical trigger condition, a time trigger condition and a pathological trigger condition, and the geographical trigger condition is obtained by the following steps:
[0177] S41, obtaining first geographical location information in the first actual inducement data, and arranging the first geographical location information in the first actual inducement data in chronological order;
[0178] S42, determining whether the plurality of first geographical location information are consistent, and if not, obtaining a change record of the first geographical location information;
[0179] S43. Determine whether the duration of the change record exceeds a preset time period. If so, generate a geographic trigger condition, which is used to update the adjustment weight value and update the first geographic location information to the permanent geographic location information in the user information.
[0180] In this embodiment, a trigger condition for updating step S18 is also included. In step S18, an adjustment weight value is generated. In the actual application process, the adjustment weight value will be updated under the influence of some influencing factors. In view of this influencing factor, the concept of trigger condition is introduced in this embodiment. That is, when the trigger condition is met, step S18 will be re-executed to obtain a new adjustment weight value. Specifically, the trigger condition includes a geographical trigger condition, a time trigger condition and a pathological trigger condition, wherein the time trigger condition can be understood as regularly updating the adjustment weight value, and the pathological trigger condition can be understood as the difference between the user's adjustment weight value before diagnosis and the adjustment weight value after diagnosis, so that the adjustment weight value needs to be updated when re-diagnosed. The geographical trigger condition can be understood as when a user changes from one permanent geographical location information to another permanent geographical location information, there will be changes in work and rest and diet, such as when a user moves from the northern region to the southern coastal area due to a job change.
[0181] Specifically, the geographic trigger condition is obtained through step S41 to step S43. In step S41, the first geographic location information in the first actual incentive data is obtained, and the first geographic location information in the first actual incentive data is arranged in chronological order to obtain the migration route of the user in the first actual incentive data.
[0182] In step S42, it is determined whether the plurality of first geographical location information are consistent, if yes, it means that the user has not moved geographically, if no, it means that the user has moved geographically, and it is necessary to obtain the change record of the first geographical location information.
[0183] In step S43, it is determined whether the duration of the change record exceeds the preset time period. The preset time period here can be understood as a pre-set time period, such as three months. When the change record of the first geographic location information exceeds three months, it can be determined that the current user has completed the replacement of the permanent geographic location information, and the first geographic location information is updated to the permanent geographic location information in the user information, and a geographic trigger condition is generated at the same time to update the adjustment weight value.
[0184] This embodiment sets the trigger conditions for updating the adjustment weight value so that the actual induction probability corresponding to the user can be updated and adjusted in time, thereby realizing dynamic adjustment of the actual induction probability of colorectal cancer for the user, which is more in line with the actual usage scenario.
[0185] In some embodiments, extracting an inducement group having the same similarity with the current user information from multiple inducement groups according to the user information includes:
[0186] Performing vector transformation on each inducement group to obtain a group of multidimensional vectors corresponding to each inducement group, and recording the group of multidimensional vectors corresponding to each inducement group as a reference vector group to obtain multiple reference vector groups;
[0187] Perform vector conversion on the user information to obtain a set of multi-dimensional vectors corresponding to the user information, and record it as a user vector group;
[0188] Perform inner product calculations on the reference vector group and the user vector group one by one to obtain multiple inner product calculation results, each inner product calculation result corresponding to a reference vector group;
[0189] Calculate the modulus of the reference vector group one by one to obtain multiple reference vector moduli, calculate the modulus of the user vector group to obtain the user vector modulus;
[0190] Perform product operations on the reference vector modulus and the user vector modulus one by one to obtain multiple product calculation results, each product calculation result corresponds to a reference vector group;
[0191] Pairing the inner product calculation results with the product calculation results according to the reference vector group to obtain multiple similarity calculation groups;
[0192] Divide the inner product calculation result and the product calculation result in the same similarity calculation group to obtain a similarity value, and each reference vector group corresponds to a similarity value;
[0193] Arrange the multiple similarity values in order of their values, select the reference vector group corresponding to the similarity value ranked first, and record it as the first reference vector group to be analyzed;
[0194] The inducement group corresponding to the first reference vector group to be analyzed is obtained, and recorded as the first inducement group to be analyzed.
[0195] In this embodiment, the cosine similarity algorithm model is used to determine whether the similarity between multiple incentive groups and the current user information is consistent. Specifically, each incentive group is converted into a vector. The incentive group contains multiple incentive categories. Each incentive category has initial incentive information. The initial incentive information of an incentive category is converted into a vector in one direction. The initial incentive information is also the length of this vector, which is recorded as a vector of one dimension. Then multiple incentive categories correspond to vectors of multiple dimensions, and one incentive group corresponds to a group of multidimensional vectors. For ease of distinction, the incentive group converted into a vector embodiment is recorded as a reference vector group. Similarly, the user information is converted into a group of multidimensional vectors, recorded as a user vector group. After performing inner product calculations and multiplication calculations on the reference vector group and the user vector group, multiple similarity values are obtained. Each reference vector group corresponds to a similarity value, that is, each incentive group corresponds to a similarity value. Based on the numerical arrangement of the similarity values, the similarity value ranked first is selected, and the inducement group associated with the reference vector group corresponding to it, that is, the inducement group with the highest similarity to the user information, is recorded as the first inducement group to be analyzed.
[0196] This embodiment shows the specific calculation process of obtaining the first incentive group to be analyzed. By using the cosine similarity algorithm model, the incentive group with the closest similarity to the current user can be obtained, which is convenient for subsequent operation steps.
[0197] In a second aspect, this embodiment further provides a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described in the first aspect is implemented.
[0198] See also Figure 5 In a third aspect, this embodiment further provides an electronic device 1, comprising a memory 11 and a processor 12, wherein the memory 11 is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor 12 to implement the method described in the first aspect.
[0199] The storage medium / memory 11 includes but is not limited to: RAM, ROM, disk, tape, CD, flash memory, USB disk, mobile hard disk, memory card, memory stick, network server storage, network cloud storage, etc. The processor 12 includes but is not limited to CPU (central processing unit 12), GPU (graphics processor 12), MCU (microprocessor 12), etc.
[0200] In the above technical scheme, user information is obtained, a first preset screening condition is generated according to the user information, and the screening is performed in the database to obtain a first analysis data set, and then the first analysis data set is clustered and analyzed according to the preset classification condition to obtain multiple predisposing groups, and a predisposing group with consistent similarity with the user information is screened out from the multiple predisposing groups, recorded as the first predisposing group to be analyzed, and the first predisposing group to be analyzed is input into the first neural network model to obtain the probability parameters of inducing colorectal cancer under each first predisposing category to be analyzed in the first predisposing group to be analyzed. At the same time, the first actual predisposing data of the user in the first preset time period corresponding to the current sampling period is collected to obtain a second analysis data set, and the second analysis data set is screened to obtain a second predisposing group to be analyzed, and the second predisposing group to be analyzed is input into the second neural network model to obtain the adjustment weight value of the probability parameter of the second predisposing group to be analyzed inducing colorectal cancer under each first predisposing category to be analyzed, and the probability parameter, the adjustment weight value and the second predisposing group to be analyzed are used to obtain the actual induction probability of the user's colorectal cancer, and a colorectal cancer evaluation report is generated according to the actual induction probability. This technical solution tracks the data generated by the user's living and eating habits by collecting user information and the first actual inducement data, and rationally uses the data of confirmed patients in the database to analyze the first actual inducement data generated by the current user to obtain the actual inducement probability of the current user. A reference colorectal cancer evaluation report can be output to the current user, which is convenient for the user to determine whether to seek medical treatment, the time of treatment, adjustment of eating habits, etc. according to the actual situation, and timely intervene in colorectal cancer prevention operations in the early stage to reduce the probability of colorectal cancer.
[0201] Finally, it should be noted that although the above embodiments have been described in the specification and drawings of the present invention, the patent protection scope of the present invention cannot be limited thereby. All technical solutions generated by replacing or modifying equivalent structures or equivalent processes based on the essential concept of the present invention and using the contents recorded in the specification and drawings of the present invention, as well as directly or indirectly implementing the technical solutions of the above embodiments in other related technical fields, are included in the patent protection scope of the present invention.
Claims
1. A method for colorectal cancer screening based on big data analysis, characterized in that: include: Obtaining user information, including user name, ID number, age, geographic location, and diet information; Generate a first preset screening condition according to the user information, and screen the database according to the first preset screening condition to obtain a first analysis data set, wherein the database includes initial predisposing data of historically diagnosed patients, the initial predisposing data includes initial predisposing information of multiple predisposing categories that induce colorectal cancer, and the predisposing categories include any one of age factors, geographical location factors, dietary factors, exercise factors, and genetic factors; Performing cluster analysis on the first analysis data set according to preset classification conditions, wherein the preset classification conditions include multiple classification levels, obtaining multiple inducement groups, each of which includes classification feature data of multiple inducement categories at the same classification level, and the classification feature data includes initial inducement information and inducement categories; Extracting an inducement group with the same similarity to the current user information from the plurality of inducement groups according to the user information, recording it as a first inducement group to be analyzed, and recording the classified feature data in the first inducement group to be analyzed as first feature data to be analyzed, wherein the first feature data to be analyzed includes first inducement information to be analyzed and a first inducement category to be analyzed; Inputting the first predisposing factor group to be analyzed into the trained first neural network model, the output result of the first neural network model is the probability parameter of inducing colorectal cancer under each of the first predisposing factor categories to be analyzed in the first predisposing factor group to be analyzed; Collecting first actual inducement data of the user within a first preset time period corresponding to a current sampling period to obtain a second analysis data set, wherein the second analysis data set includes first actual inducement information of multiple inducement categories that induce colorectal cancer; Screening the second analysis data set according to the first inducement category to be analyzed to obtain a second inducement group to be analyzed, wherein the second inducement group to be analyzed includes second characteristic data to be analyzed, and the second characteristic data to be analyzed includes second inducement information to be analyzed; Inputting the second predisposing factor group to be analyzed into the trained second neural network model, the output result of the second neural network model is the adjusted weight value of the probability parameter of inducing colorectal cancer under each of the first predisposing factor categories to be analyzed in the second predisposing factor group to be analyzed; Obtaining the actual induction probability of colorectal cancer of the user according to the probability parameter, the adjustment weight value and the second induction group to be analyzed; adjusting the length of the first preset time period corresponding to the next sampling period according to the actual induction probability; Generating a colorectal cancer assessment report according to the actual induction probability; Generating a first preset screening condition according to the user information, and screening the database according to the first preset screening condition to obtain a first analysis data set includes: Generating a first screening condition according to the age information, screening out the initial inducement data satisfying the first screening condition from the database, recording the data as the first initial inducement data, and obtaining a first initial inducement data set; generating a second screening condition according to the permanent geographical location information, and screening out second initial inducement data satisfying the second screening condition from the first initial inducement data set, which is recorded as a second initial inducement data set; generating a third screening condition according to the dietary information, screening out third initial inducement data satisfying the third screening condition from the second initial inducement data set, and obtaining the first analysis data set; The method further comprises: Determine, according to the user information, whether there is a first preset historical time period associated with the current user information, where the first preset historical time period is the first preset time period of the current user in the previous sampling cycle; If yes, the historical first preset time period is used as the first preset time period corresponding to the current sampling period, and the incentive data of the current user information within the first preset time period corresponding to the current sampling period is obtained as the first actual incentive data; If not, the initial cause information of the first analysis data set is sorted according to the severity of the colorectal cancer symptoms of the historically diagnosed patients, and a first preset time period corresponding to the sampling period of the historically diagnosed patients within the first preset disease threshold range is obtained, and recorded as a reference first preset time period; Performing average calculation on the plurality of reference first preset time periods, and using the calculation result as the first preset time period corresponding to the current user in the current sampling period, and obtaining the incentive data within the first preset time period corresponding to the current user information in the current sampling period up to now as the first actual incentive data; The user information also includes genetic disease history and past medical history, and the first preset disease threshold is obtained by the following steps: Obtaining a threshold adjustment factor based on the genetic disease history and past medical history; Obtaining a first basic disease threshold of the current user according to the age information; The first preset disease threshold is calculated according to the first basic disease threshold and the threshold adjustment factor; Collecting the first actual inducement data of the user within the first preset time period corresponding to the current sampling period to obtain the second analysis data set includes: Dividing the first preset time period into multiple second preset time periods; Allocating the first actual inducement data according to a second preset time period to obtain a plurality of second actual inducement data; Generate an inducement score according to the second actual inducement data, each second actual inducement data corresponds to an inducement score; The plurality of second actual inducement data are sorted according to the inducement scores, and the second actual inducement data within a preset inducement threshold range are eliminated to obtain the second analysis data set.
2. The method for colorectal cancer screening based on big data analysis according to claim 1, characterized in that: The method further comprises: Obtaining medical consultation information, wherein the medical consultation information includes symptoms and test reports of the current user who has been diagnosed with colorectal cancer, wherein the test reports include values of multiple test indicators; Generating initial diagnosis information according to the medical consultation information and displaying it on the doctor's end, obtaining the modification content on the doctor's end, and generating final diagnosis information according to the modification content and the initial diagnosis information; According to the final diagnosis information, the database is filtered for treatment information and recovery information of users with the same final diagnosis information, wherein the recovery information includes recovery diet information and / or recovery exercise information; A recovery strategy recommendation within a first preset treatment cycle is generated and displayed based on the treatment information, post-healing information and user information.
3. The method for colorectal cancer screening based on big data analysis according to claim 2, characterized in that: The final diagnosis information includes symptom image information, and the method further includes: Acquiring lesion distribution information in the disease image information, wherein the lesion distribution information includes at least one of the canceration point of colon cancer, the degree of colon canceration, the canceration point of rectal cancer, and the degree of rectal canceration; generating first predicted inducement data within the first preset treatment cycle according to the first actual inducement data; Generate colorectal cancer simulation evolution information of the current user according to the first predicted cause data and the symptom image information, and dynamically display the simulation evolution information, wherein the colorectal cancer simulation evolution information includes a predicted evolution animation image of the cancerous symptom degree of the current user under the first predicted cause data; The method further comprises: Obtaining medical consultation information associated with the current user based on the user information. If the current user has visited the doctor multiple times, obtaining the historical test report of the current user in the previous medical consultation record and the real-time test report of the current user under the current medical consultation based on the user information; Compare the historical inspection report and the real-time inspection report to obtain the numerical change trend of multiple inspection indicators; Determine one by one whether the numerical variation trend of the inspection index is within the preset variation trend range, if not, mark the inspection index that is not within the preset variation trend range as an abnormal index, and record the numerical variation trend of the abnormal index as an abnormal variation trend; A simulated trend of the abnormal indicator is generated according to the abnormal change trend, and a disease prediction result is generated and displayed according to the simulated trend of the abnormal indicator.
4. The method for colorectal cancer screening based on big data analysis according to claim 1, characterized in that: The method further comprises: According to the first actual inducement data, a triggering condition for triggering the updating of the adjustment weight value is obtained, wherein the triggering condition includes a geographical triggering condition, a time triggering condition and a pathological triggering condition, and the geographical triggering condition is obtained by the following steps: Acquire the first geographical location information in the first actual inducement data, and arrange the first geographical location information in the first actual inducement data in chronological order; Determine whether the plurality of first geographic location information are consistent, and if not, obtain a change record of the first geographic location information; Determine whether the duration of the change record exceeds a preset time period. If so, generate a geographic trigger condition, which is used to update the adjustment weight value and update the first geographic location information to the permanent geographic location information in the user information.
5. The method for colorectal cancer screening based on big data analysis according to claim 1, characterized in that: Extracting an inducement group having the same similarity with the current user information from the plurality of inducement groups according to the user information includes: Performing vector conversion on each inducement group to obtain a group of multidimensional vectors corresponding to each inducement group, and recording the group of multidimensional vectors corresponding to each inducement group as a reference vector group to obtain multiple reference vector groups; Performing vector conversion on the user information to obtain a group of multi-dimensional vectors corresponding to the user information, and recording the group of multi-dimensional vectors as a user vector group; Performing inner product calculations on the reference vector group and the user vector group one by one to obtain a plurality of inner product calculation results, each of the inner product calculation results corresponding to a reference vector group; Calculate the modulus of the reference vector group one by one to obtain a plurality of reference vector moduli, and calculate the modulus of the user vector group to obtain a user vector modulus; Performing product operations on the reference vector modulus and the user vector modulus one by one to obtain a plurality of product calculation results, each of the product calculation results corresponding to a reference vector group; Pairing the inner product calculation result with the product calculation result according to the reference vector group to obtain multiple similarity calculation groups; Performing a division operation on the inner product calculation result and the product calculation result in the same similarity calculation group to obtain a similarity value, wherein each reference vector group corresponds to one similarity value; Arrange the plurality of similarity values in order of value, select a reference vector group corresponding to the similarity value ranked first, and record it as a first reference vector group to be analyzed; The inducement group corresponding to the first reference vector group to be analyzed is obtained, and recorded as the first inducement group to be analyzed.
6. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: The computer program instructions implement the method according to any one of claims 1 to 5 when executed by a processor.
7. An electronic device comprising a memory and a processor, characterized in that: The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method according to any one of claims 1 to 5.
