CN113571180A - C-peptide layering and organ function-based type 2 diabetes artificial intelligent diagnosis and treatment management system - Google Patents
C-peptide layering and organ function-based type 2 diabetes artificial intelligent diagnosis and treatment management system Download PDFInfo
- Publication number
- CN113571180A CN113571180A CN202110824477.1A CN202110824477A CN113571180A CN 113571180 A CN113571180 A CN 113571180A CN 202110824477 A CN202110824477 A CN 202110824477A CN 113571180 A CN113571180 A CN 113571180A
- Authority
- CN
- China
- Prior art keywords
- data
- classifier
- treatment
- peptide
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Theoretical Computer Science (AREA)
- Biomedical Technology (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Primary Health Care (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Epidemiology (AREA)
- Artificial Intelligence (AREA)
- Pathology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention relates to a C peptide layering and organ function based type 2 diabetes artificial intelligent diagnosis and treatment management system which comprises a data acquisition part, an auxiliary diagnosis part, an auxiliary treatment part, an audit management part, an early warning system part, a referral part and a follow-up part. And the data acquisition module acquires the electronic case data of the diabetic from the hospital case database according to the inclusion standard and the exclusion standard. And (3) data standardization preprocessing, namely performing assignment conversion on non-numerical data in the sample data, designating a non-numerical data conversion rule, performing assignment conversion according to the conversion rule, and performing abnormal value processing on the numerical data in the sample data to obtain standardized sample data. Model + data artificial intelligent diagnosis and treatment of diabetes and complications. Aiming at the defects in the prior art, the invention predicts the physical function condition of the diabetic patient by adopting C peptide layering and organ functions and utilizing machine learning and medical data, and provides an intelligent treatment scheme for doctors.
Description
Technical Field
The invention relates to the technical field of artificial intelligent medical treatment, in particular to an artificial intelligent diagnosis and treatment and intelligent management system for diabetes and complications.
Background
According to the four nationwide flow regulation data, the prevalence rate of adult diabetes in China is in a rising trend, 9.7% in 2010, 11.6% in 2013, 10.9% in 2017 and 12.8% in 2020, and 1.298 million diabetics exist in China at present. Diabetes mellitus presents a growing trend towards younger, special types and other types of diabetes mellitus. But the awareness rate of people is low, the standard reaching rate is low, the treatment management rate is low, complications exist in half of patients when diabetes is diagnosed, and the diabetes can cause serious complications of a large vascular system, kidney, retina and nervous system. The complications caused by diabetes cause the rise of medical expenses, and generate heavy economic and social burden
The main features of patients with T2DM are obesity and insulin resistance, but the reserve function of islet beta cells decreases gradually with the course of diabetes. Beta cell dysfunction is closely related to the occurrence of blood glucose failure and complications. Glucose excursions and poor glucose control due to beta cell dysfunction contribute to an increased risk of developing diabetic complications. C peptide is secreted equimolar to insulin, and has longer half-life, although C peptide does not directly affect blood sugar, it is much more accurate than directly detecting insulin in evaluating endogenous secretion of insulin, and is a reliable index for reflecting pancreatic beta cell function. In current clinical diagnosis and treatment, a clinician usually uses the C peptide as an index reflecting the function of the pancreatic islet. However, as a bioactive peptide with endocrine function, C peptide can be used as an important index for selecting hypoglycemic drugs and a prediction index for diabetic complications. After more than 10 years of clinical practice, the diabetes mellitus layered treatment according to the empty C peptide and the peak C peptide has high standard reaching rate, the chronic complications of the diabetes mellitus are obviously delayed or avoided, and patients obviously benefit, and relevant researches are published in foreign periodicals.
In the past diagnosis and treatment guidelines, the selection of hypoglycemic drugs is mainly based on fasting blood glucose and HbA1c, an individual layered diagnosis and treatment system aiming at patients is lacked, and the selection of drugs greatly depends on clinicians. In order to promote the diabetic patients to reach the blood sugar standard, delay the occurrence of complications, reduce the economic burden of the patients and the society and improve the basic medical diagnosis and treatment level, the system provides an individualized and standardized diagnosis and treatment decision scheme system.
The diabetes patients lack understanding of the diseases, are unaware of the harm of the diabetes, have poor treatment compliance, low rate of return visits, poor self-management and no control of reasonable diet and exercise.
The community hospitals lack a diabetes comprehensive management team, the initial diagnosis of the diabetes patients is under-regulated, the screening rate of the diabetes complications is low, the management and control of risk points are not in place, and the reversal opportunity is missed.
The initial diagnosis of diabetes in the third hospital is less, a large number of high-risk diabetic patients are not well controlled, and the reversal of various complications for treating diabetes is difficult.
Disclosure of Invention
Aiming at the defects in the prior art, the invention predicts the physical function condition of the diabetic by adopting C peptide layering and organ functions and utilizing machine learning and medical data, provides an intelligent treatment scheme for doctors, and provides cooperative referral and follow-up for patients in a Hospital and a community hospital.
In order to achieve the purpose, the invention adopts the technical scheme that:
an artificial intelligent diagnosis and treatment management system for type 2 diabetes based on C peptide layering and organ functions comprises a data acquisition part, an auxiliary diagnosis part, an auxiliary treatment part, an audit management part, an early warning system part, a referral part and a follow-up part.
The data acquisition module acquires electronic case data of the diabetic patient from a hospital case database according to the inclusion standard and the exclusion standard, wherein the electronic case data comprises information such as basic information of the diabetic patient, a patient case, diagnosis and treatment data, a medical image, an examination report and the like, and sample data is obtained.
And (3) data standardization preprocessing, namely performing assignment conversion on non-numerical data in the sample data, designating a non-numerical data conversion rule, performing assignment conversion according to the conversion rule, and performing abnormal value processing on the numerical data in the sample data to obtain standardized sample data.
Model + data artificial intelligent diagnosis and treatment of diabetes and complications. And (3) building a diabetes knowledge base based on the latest medical knowledge, clinical research and scientific research results.
An artificial intelligent diagnosis and treatment system model based on AI technology and Adaboost iterative algorithm. The core idea is to train different classifiers (weak classifiers) aiming at the same training set, and then to assemble the weak classifiers to form a stronger final classifier (strong classifier). The algorithm is realized by changing data distribution, and determines the weight of each sample according to whether the classification of each sample in each training set is correct and the accuracy of the last overall classification. And (4) sending the new data set with the modified weight value to a lower-layer classifier for training, and finally fusing the classifiers obtained by each training as a final decision classifier. The use of the adaboost classifier may eliminate some unnecessary training data features and place the key on top of the key training data.
The algorithm is a weak classification algorithm promotion process, and the classification capability of the data can be improved through continuous training in the process. The whole process is as follows:
(1) firstly, learning a C peptide data sample to obtain a first weak classifier;
(2) forming a new N training samples by the samples and other new data, and obtaining a second weak classifier by learning the samples;
(3) adding other new samples into the samples to form another new N training samples, and learning the samples to obtain a third weak classifier;
(4) and finally, the strong classifiers are lifted. I.e. into which class a certain data is to be passed.
The Adaboost algorithm flow is specifically as follows:
given a training data set T = { (x1, y1), (x2, y2) … (xN, yN) }, where examples, and example space, yi, belong to the label set { -1, +1}, the purpose of Adaboost is to learn a series of weak classifiers, or basic classifiers, from the training data and then combine these weak classifiers into a strong classifier.
The algorithm flow of Adaboost is as follows:
Learning with a training data set with weight distribution Dm to obtain a basic classifier (selecting a threshold value with the lowest error rate to design the basic classifier):
b. calculating a Classification error Rate of Gm (x) on a training data set
As can be seen from the above equation, the error rate em of Gm (x) on the training data set is the sum of the weights of the misclassified samples by Gm (x).
c. The coefficient of Gm (x) is calculated, am representing the importance of Gm (x) in the final classifier (objective: obtaining the basic classifier)The weight occupied in the final classifier. Note: this formula is more accurate when written as am =1/2ln ((1-em)/em) because The base number isThe natural logarithm e, so Inlog is easily mistaken for a base of 2 or another base, as follows):
from the above equation, when em < =1/2, am > = 0, and am increases with decreasing em, meaning that the basic classifier with smaller classification error rate has a greater role in the final classifier.
d. The weight distribution of the training data set is updated (in order to obtain a new weight distribution of the samples) for the next iteration
So that the weight of the misclassified samples by the basic classifier gm (x) is increased and the weight of the correctly classified samples is decreased. In this manner, the AdaBoost method can "focus" or "focus on" those samples that are less readily separable.
Where Zm is a normalization factor, making Dm +1 a probability distribution:
The final classifier is thus obtained as follows:
auxiliary diagnosis, combined with clinical experience and results of cross-sectional studies, was first stratified based on fasting C-peptide and C2/C0 levels. The layering basis is as follows: c peptide is less than 1ng/mL, C peptide is less than or equal to 1ng/mL and less than 2 ng/mL, C peptide is less than or equal to 2 ng/mL and less than 3ng/mL, and C peptide is more than or equal to 3 ng/mL; C2/C0 is less than 1, C2/C0 is less than or equal to 1 and less than or equal to 2, C2/C0 is less than or equal to 2 and less than or equal to 3, and C2/C0 is more than or equal to 3. Combining the examination and test indexes of the patient, the method comprises the following steps: 1. no complications, 2 cardiovascular complications, 3 renal complications, 4 diabetic peripheral neuropathy, 5 complications of obesity or fatty liver, 6 complications of gout, 7 complications of gastric and duodenal ulcers and 8 complications of hyperlipidemia.
And (5) assisting treatment and establishing a medicine database. And (4) calculating information such as a preferred treatment scheme, an optional treatment scheme, a cautious scheme and the like according to the C peptide data sample hierarchical data to give a medication scheme. Such as C peptide less than 1ng/ml, preferably long-acting insulin, quick-acting insulin, metformin, DPP-4 inhibitor, SGLT-2 inhibitor; and (3) optional: alpha-glucosidase inhibitors, GLP-1 agonists, thiazolidinediones; use with cautions: meglitinides, sulfonylureas.
The early warning system:
and monitoring various indexes of the patient regularly, and notifying and early warning the patient and a doctor when a certain index exceeds a conventional range.
1) Blood glucose fasting glucose was found for the first time >7mmol/l or random glucose >11.1mmol/l, and age <25 years.
2) Abnormal blood sugar in pregnant and lactating women.
3) The random blood sugar is more than or equal to 16.7 mmol/L, and the HbA1c is more than 9%.
4) The routine of urine: ketone bodies (+); urine protein (+).
5) Newly appeared cardiovascular complications (cardiology department), kidney complications (urinary microalbumin/creatinine >30mg/g or eGFR <60 ml.min-1. (1.73 m 2) -1) or the appearance of glutamic-pyruvic transaminase >120U/L and glutamic-oxaloacetic transaminase >120U/L (gastroenterology department), and patients are advised to visit the department in cardiology and gastroenterology departments.
6) Fasting glucose of <3.9mmol/L occurs more than twice, or severe hypoglycemia occurs once (severe impairment of consciousness and behavior with or without <3.9mmol/L of glucose, requiring assistance from others to replenish carbohydrates, glucagon or take other measures to help recovery).
7) The follow-up visits are not up to standard for two consecutive times.
8) When the patient eGFR <60ml min-1. (1.73 m 2) -1, the follow-up eGFR declined more than the previous one (nephrology department); the glutamic-pyruvic transaminase is more than 120U/L, and the glutamic-oxaloacetic transaminase is more than 120U/L, so that the glutamic-pyruvic transaminase/glutamic-oxaloacetic transaminase is increased earlier (in the department of gastroenterology), and the patients are reminded to visit the department of nephrology or the department of gastroenterology.
9) eGFR <15 ml. min-1. (1.73 m 2) -1, and the special department of nephrology visits the clinic immediately.
10) Patients report patients with severe adverse reactions of hypoglycemic drugs.
11) When other diseases occur, doctors are required to guide the special department to see the doctor.
12) The primary health care facility personnel determine the patients who are deemed to need referral to the superior hospital.
13) Patients in a calm state or after activities have chest distress, suffocation and precordial pain, and the symptoms appear when walking less than 200 meters on flat ground or climbing 1 floor; or dyspnea with or without cough, expectoration and hemoptysis, physical strength decrease, asthenia, etc. Needs to be diagnosed immediately in emergency internal medicine or cardiology department, and perfects the examination of electrocardiogram and the like.
14) Patients have dizziness, tinnitus, cough due to drinking water, difficulty in moving limbs, facial distortion, slurred speech or syncope and the like, and need to be treated by emergency department neurology.
15) The body temperature of the patient is over 37.3 ℃, with or without cough and expectoration (respiratory medicine); frequent, urgent and painful urination (nephrology department); abdominal pain, diarrhea, etc. (infectious department), need special medical treatment, perfecting blood routine, secretion culture, etc. examination, if necessary antibiotic treatment.
16) Patients with visual deterioration, blurred vision, visual field defect, eye pain and the like need to be treated by an eye clinic.
17) The patients have pain, red swelling and limited movement of toes and need outpatient service in special department of endocrinology.
18) Patients have intermittent claudication, pain and numbness of lower limbs after activity, or skin ulceration of toes and lower limbs, and need to be treated as soon as possible by special outpatient service of endocrinology department.
And the auditing module is used for rechecking the scheme of the manual diagnosis and treatment system by the doctor, modifying the scheme, issuing an order and requiring an electronic label.
Referral, basic health institution: when a diabetic patient is newly diagnosed, abnormal blood sugar, uncertain etiology and classification, abnormal blood sugar less than 25 years old, during pregnancy and lactation, acute complications or severe chronic complications are discovered, after repeated hypoglycemia or one-time severe hypoglycemia, screening and treatment of the chronic complications are difficult, after standard treatment, the blood sugar, blood pressure and blood fat are not controlled to reach the standard, blood sugar fluctuation is large or insulin is difficult to adjust, the diabetic patient is transferred to a superior hospital, diagnosis of the diabetic patient is definite, a treatment scheme is determined, after treatment of the acute and chronic complications is stable, the treatment scheme is adjusted, and control of the blood sugar, the blood pressure and the blood fat reaches the standard and is transferred back to a basic medical health institution.
And during follow-up visit, a doctor can call the information of the electronic case system through the follow-up visit system to generate a follow-up visit queue, set follow-up visit time, send follow-up visit reminders to patients through short messages or micro-information small programs, monitor the blood sugar condition through follow-up visit every month, and start a new treatment scheme if the blood sugar does not reach the standard after 3 months. Follow-up monitoring is carried out regularly after treatment, the diabetic is recommended to carry out physical examination every year, and complication screening is carried out in 5 years and 10 years, so that early detection and early intervention are realized.
And (4) health management, namely performing individual health education according to the diabetes stratification and complication information of the patient. The popular science knowledge, the diet knowledge and the exercise knowledge are pushed in a targeted manner. Improve the consciousness and the compliance of the patients and delay the progress of complications.
The invention has the beneficial effects that: aiming at the defects in the prior art, the invention predicts the physical function condition of the diabetic by adopting C peptide layering and organ functions and utilizing machine learning and medical data, provides an intelligent treatment scheme for doctors, realizes patient referral and follow-up of cooperative patients in the third hospital and the community hospital, has convenient data acquisition and quick flow processing, and greatly improves the diagnosis and treatment management effect of the patients. The invention provides an artificial intelligent diagnosis and treatment management system platform for Type 2 diabetes mellitus (T2 DM) based on C peptide layering and organ functions, provides an artificial intelligent diagnosis and treatment scheme, a medication scheme and a diet and exercise management system thereof for diabetics, assists doctors in the hospitals and community hospitals in accurately treating the T2DM patients and enabling the patients to follow up for a long time, and improves the long-term standard of blood sugar, blood fat and blood pressure of the T2DM patients. The system comprises a data acquisition module, an artificial diagnosis and treatment model (a screening method and a model for diabetes and complications), a medication model, an early warning system, an auditing module, a follow-up module, a referral module and a health management module. The system has the innovation points that a new strategy for layered treatment of a T2DM patient is guided by the basal secretion function, the peak secretion function and the visceral organ function of the insulin C peptide, a comprehensive treatment scheme and a medication scheme are formulated by collecting relevant index data of diabetes and complications of the patient and receiving and analyzing relevant parameters of an artificial diagnosis and treatment model, so that life style guidance is provided, a doctor-seeing prompt and the education science popularization of the diabetes patient are regularly pushed, the cognition and treatment compliance of the diabetes patient are improved, the insulin function and the visceral organ function are protected in an early stage, the occurrence and development of the chronic complications of the diabetes are prevented, and the life quality of the patient is improved.
Drawings
FIG. 1 is a graph of data analysis of peptide C and blood glucose standard deviation, diabetic complications and fatty liver (corrected for gender, age, course, BMI).
FIG. 2 is a graph of data analysis of C2/C0 ratio versus glycemic compliance rate, diabetic complications, and fatty liver (corrected for gender, age, course, BMI) when grouped according to C peptide.
Fig. 3 is a block diagram of the system.
FIG. 4 is a schematic diagram of the flow of the artificial intelligence Adaboost algorithm of the system.
Fig. 5 is a schematic view of the follow-up procedure of the system.
Fig. 6 is a block diagram of the functions and network security modules of the system.
Detailed Description
For better understanding and implementation, the technical solutions will be described clearly and completely with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
The system of the embodiment integrally adopts a framework of front-end and back-end separation and micro-service, the front end is realized through vue, and the back end provides service data for the front end based on the micro-service constructed by spring group.
The system comprises a cloud server, a database, a medical management end, a patient end and a data acquisition terminal.
Data acquisition of the system of the embodiment is realized by docking a hospital HIS system through an open interface, and information of medical diabetic patients, including basic information of the patients, cases of the patients, diagnosis and treatment data, medical images and examination reports, is collected. Data collection strategy: patient data was measured by date 8 am: 00 newly increased in the last day of grabbing, and single grabbing is performed by a doctor when the doctor diagnoses the patient according to the related data of the state of an illness of the patient.
And the patient side receives follow-up visit reminding through a WeChat applet or a short message and checks the popular science knowledge, the diet knowledge and the exercise knowledge in the health management column. The patient ID is the same as the ID distributed by the hospital HIS system, and the case record is also synchronized with the HIS
And the database adopts a MySql database to establish a patient basic information base which comprises personal files, case records and treatment records of the patient. A medicine database, a medicine retrieval database is established by the classification and the function of the medicines,
and the medical management terminal establishes a multi-level authority account, and a doctor manages the patient through logging in the platform through the account.
Patient list: 1, list show item: name, gender, identification number, date of birth, age, organization, address, phone and contact phone, follow-up status, visit list (clickable), follow-up list (clickable). 2. Patients are grouped for ease of administration (e.g., present unit patient, present day follow-up). 2, a retrieval function: the patient's name (patient identification number) is quickly looked up. And 3, inputting a function by the new patient, adding a [ add patient ] button on the page, and clicking an add patient button.
Personal details of the patient: 1, list show item: name, gender, date of birth, time of visit, hospital visit, department of visit, doctor, diagnosis, card number of visit, patient type, diagnostic information (assay index, examination, consultation, admission record, case, drug allergy, etc.) 2, sorting: the diagnosis time is reversed.
Diagnosis and treatment scheme details: 1, click [ add/edit treatment plan ] medication regimens can be added or edited, and the type, single dose unit, frequency, usage, quantity unit, whether to perform skin test, results of skin test, notes, etc. of the drug can be added or modified. 2, submitting to be checked and sending to the patient after being checked
Follow-up, non-submitted treatment protocol: 1, a presentation item: gender, age, complaints, systolic blood pressure, diastolic blood pressure, BMI, waist circumference, hip circumference, fasting plasma glucose, 120 minute plasma glucose, HbA1C, fasting C-peptide, 120 minute C-peptide, fasting insulin, urinary microalbumin/creatinine, complications currently associated with patients. And 2, if the patient is a follow-up patient, displaying the original treatment scheme. And 3, judging whether the follow-up visit of the patient reaches the standard or not, and if not, carrying out special marking. 4, whether the patient has new complications or not, if so, special marking is needed. 5, clickable button: the button-this treatment scheme of the patient is submitted to be examined and checked, and then the treatment scheme is sent to the patient after being examined and checked by adopting an electronic tag. 6, the allergy history or adverse reaction record of the medicine needs to be displayed, a doctor can input the medicine which is allergic or has adverse reaction for a patient, and all treatment schemes of the patient need to remove the medicine. Reminding needs to be set according to a calculation formula of early warning of each check item.
Follow-up list: 1, list show item: name, gender, date of birth, diagnosis, chief complaints, follow-up time, follow-up item, follow-up status, follow-up details (clickable access) 2, add or modify follow-up date
Health management: classification of patients Using Multi-tag Classification, non-patients providing tailored health knowledge push Table 1 general conditions of patients grouped according to C-peptide
| Q1（n = 346） | Q2（n = 347） | Q3（n = 341） | Q4（n = 343） | PValue of | |
| Gender (female) | 48.00% | 52.20% | 46.30% | 47.80% | 0.457 |
| Age (year of old) | 63.65 ± 9.509 | 65.15 ± 9.624 | 64.09 ± 10.158 | 64.98 ± 10.563 | 0.151 |
| Course of diabetes mellitus (year) | 16.42 ± 6.942 | 15.90 ± 6.435 | 14.07 ± 6.403 | 13.59 ± 6.331 | < 0.05 |
| BMI（kg/m2） | 24.74 ± 2.990 | 25.53 ± 3.014 | 25.98 ± 3.388 | 27.31 ± 3.973 | < 0.05 |
| Systolic pressure (mmHg) | 142.54 ± 22.091 | 141.29 ± 17.676 | 141.21 ± 19.226 | 143.20 ± 21.916 | 0.499 |
| Diastolic blood pressure (mmHg) | 76.75 ± 11.606 | 76.63 ± 10.969 | 78.01 ± 11.104 | 79.18 ± 12.011 | < 0.05 |
| History of smoking (%) | 29.00% | 23.50% | 28.50% | 28.90% | 0.293 |
| History of drinking (%) | 26.90% | 25.10% | 26.60% | 29.90% | 0.573 |
| Fasting blood glucose (mmol/L) | 6.09 ± 2.652 | 7.13 ± 2.634 | 7.53 ± 2.464 | 8.08 ± 2.921 | < 0.05 |
| HbA1c（%） | 8.74 ± 1.910 | 8.37 ± 1.757 | 8.13 ± 1.728 | 8.10 ± 1.554 | < 0.05 |
| Fasting insulin (mU/L) | 5.70 ± 11.785 | 10.71 ± 25.404 | 11.73 ± 16.485 | 21.76 ± 49.524 | < 0.05 |
| C2/C0 | 2.71 ± 2.133 | 2.01 ± 1.072 | 1.93 ± 0.903 | 1.56 ± 0.707 | < 0.05 |
| HOMA-IR | 2.08 ± 0.369 | 2.72 ± 0.477 | 3.37 ± 0.643 | 4.98 ± 2.050 | < 0.05 |
| Triglyceride (mmol/L) | 1.29 ± 1.182 | 1.65 ± 1.596 | 1.90 ± 1.627 | 2.53 ± 2.180 | < 0.05 |
| Total cholesterol (mmol/L) | 4.70 ± 1.422 | 4.55 ± 1.346 | 4.50 ± 1.345 | 4.49 ± 1.242 | 0.163 |
| LDL-C（mmol / L） | 2.74 ± 1.141 | 2.69 ± 1.090 | 2.67 ± 1.021 | 2.66 ± 0.951 | 0.73 |
| HDL-C（mmol / L） | 1.35 ± 0.342 | 1.26 ± 0.323 | 1.20 ± 0.296 | 1.11 ± 0.245 | < 0.05 |
| Microalbumin urine/creatinine (mg/g) | 123.06 ± 317.400 | 70.77 ± 173.126 | 93.07 ± 264.876 | 108.95 ± 229.465 | 0.092 |
| eGFR（mL·min-1·（1.73 m2）-1） | 116.51 ± 36.412 | 119.79 ± 37.228 | 118.34 ± 33.649 | 100.34 ± 40.925 | < 0.05 |
| Diabetic complications | |||||
| Diabetic nephropathy (%) | 34.40% | 27.40% | 21.70% | 36.80% | <0.05 |
| Diabetic retinopathy (%) | 47.40% | 44.50% | 35.80% | 38.50% | <0.05 |
| Diabetic complications | |||||
| Fatty liver (%) | 13.50% | 22.40% | 26.60% | 36.40% | <0.05 |
| Cerebral infarction (%) | 8.70% | 11.00% | 8.50% | 9.40% | 0.677 |
| Drug use cases | |||||
| Metformin (%) | 56.50% | 68.60% | 75.70% | 74.80% | <0.05 |
| Alpha glucosidase inhibitor (%) | 69.70% | 72.40% | 65.10% | 72.40% | 0.125 |
| Secretagogues (%) | 3.50% | 11.30% | 17.50% | 18.80% | <0.05 |
| Thiazolidinediones (%) | 12.60% | 18.30% | 21.90% | 27.30% | <0.05 |
| DPP-4i（%） | 55.90% | 58.70% | 52.70% | 57.00% | 0.442 |
| GLP-1RA（%） | 2.70% | 4.90% | 7.40% | 10.90% | <0.05 |
| Insulin (%) | 95.60% | 82.80% | 60.40% | 52.10% | <0.05 |
| Statins (%) | 79.70% | 80.50% | 83.70% | 79.70% | 0.493 |
| ACEI/ARB（%） | 41.80% | 42.70% | 49.40% | 53.20% | <0.05 |
| Blood glucose standard reaching rate (%) | 19.20% | 23.30% | 29.50% | 24.20% | <0.05 |
Table 1 is a table of data for the general condition of patients grouped according to C-peptide, and FIG. 1 is a graph of data for C-peptide and glycemic compliance, diabetic complications and fatty liver (corrected for gender, age, course, BMI). In the figure, after adjusting age, sex, diabetes course and BMI, when the C peptide in the empty abdomen is more than or equal to 1.71 ng/mL, the risk of blood sugar not reaching the standard is reduced along with the increase of the C peptide compared with the Q1 group. Only when fasting C-peptide is 1.71 ≦ 2.51 ng/mL, the risk of DKD and DR development decreases with increasing fasting C-peptide. In the Q2, Q3, Q4 groups, the risk of fatty liver development increased with increasing fasting C-peptide. FIG. 2 is a graph showing the data of C2/C0 ratio and blood glucose standard-reaching rate, diabetic complications and fatty liver (correcting sex, age, course of disease, BMI) when C peptides are grouped, and further analyzing the relationship between C2/C0 and blood glucose standard-reaching rate, diabetic nephropathy, diabetic retinopathy and fatty liver in four groups. With increasing C2/C0, the risk of blood glucose not reaching standards is reduced. In the Q2 and Q3 groups, the risk of DKD development decreased with increasing C2/C0, but this relationship was not observed in the Q1, Q4 groups. Furthermore, no significant relationship was observed between DR and fatty liver and C2/C0. Fig. 3 is a block diagram of the system. FIG. 4 is a schematic diagram of the flow of the artificial intelligence Adaboost algorithm of the system. Fig. 5 is a schematic view of the follow-up procedure of the system. Fig. 6 is a block diagram of the functions and network security modules of the system.
Claims (2)
1. The utility model provides a management system is diagnose to type 2 diabetes artificial intelligence based on C peptide layering and visceral organ function which characterized in that: the system comprises a data acquisition part, an auxiliary diagnosis part, an auxiliary treatment part, an audit management part, an early warning system part, a referral part and a follow-up part; the data acquisition module acquires electronic case data of the diabetic patient from a hospital case database according to the inclusion standard and the exclusion standard, wherein the electronic case data comprises information such as basic information of the diabetic patient, a patient case, diagnosis and treatment data, a medical image, an examination report and the like, and sample data is obtained; the method comprises the steps of data standardization preprocessing, namely carrying out assignment conversion on non-numerical data in sample data, designating a non-numerical data conversion rule, carrying out assignment conversion according to the conversion rule, and carrying out abnormal value processing on the numerical data in the sample data to obtain standardized sample data; the artificial intelligent diagnosis and treatment of diabetes and complications of the model and data; auxiliary diagnosis, combined with clinical experience and results of cross-sectional studies, first stratified according to fasting C-peptide and C2/C0 levels; the layering basis is as follows: c peptide is less than 1ng/mL, C peptide is less than or equal to 1ng/mL and less than 2 ng/mL, C peptide is less than or equal to 2 ng/mL and less than 3ng/mL, and C peptide is more than or equal to 3 ng/mL; C2/C0 is less than 1, C2/C0 is less than or equal to 1 and less than or equal to 2, C2/C0 is less than or equal to 2 and less than or equal to 3, and C2/C0 is more than or equal to 3; auxiliary treatment, establishing a medicine database; calculating information such as a preferred treatment scheme, an optional treatment scheme, a cautious scheme and the like according to the C peptide data sample hierarchical data to give a medication scheme; the early warning system: monitoring various indexes of a patient regularly, and notifying and early warning the patient and a doctor when a certain index exceeds a conventional range; the auditing module is used for rechecking the scheme of the manual diagnosis and treatment system by a doctor, modifying the scheme, issuing an order and requiring an electronic label; referral, basic health institution: newly diagnosing a diabetic, finding abnormal blood sugar, unclear etiology and classification, abnormal blood sugar less than 25 years old, in pregnancy and lactation, acute complications or severe chronic complications, after repeated hypoglycemia or one-time severe hypoglycemia, screening and treating chronic complications are difficult, after standard treatment, the blood sugar, blood pressure and blood fat are not controlled to reach the standard, blood sugar fluctuation is large or insulin is difficult to adjust, and the diabetic is transferred to a superior hospital; follow-up visit, a doctor can call the information of the electronic case system through a follow-up visit system to generate a follow-up visit queue, set follow-up visit time, send follow-up visit reminders to patients through short messages or micro-information small programs, monitor blood sugar conditions through follow-up visit every month, and start a new treatment scheme if the blood sugar does not reach the standard after 3 months; follow-up monitoring is carried out regularly after treatment, the diabetic is recommended to carry out physical examination every year, and complication screening is carried out in 5 years and 10 years, so that early detection and early intervention are realized; and (4) health management, namely performing individual health education according to the diabetes stratification and complication information of the patient.
2. The artificial intelligent diagnosis and treatment management system for type 2 diabetes based on C peptide stratification and organ function according to claim 1, characterized in that: the system is based on an artificial intelligent diagnosis and treatment system model of an AI technology and an Adaboost iterative algorithm; the core idea is that different classifiers (weak classifiers) are trained aiming at the same training set, and then the weak classifiers are integrated to form a stronger final classifier (strong classifier); the algorithm is realized by changing data distribution, and the weight of each sample is determined according to whether the classification of each sample in each training set is correct and the accuracy of the last overall classification; sending the new data set with the modified weight value to a lower-layer classifier for training, and finally fusing the classifiers obtained by each training as a final decision classifier; using the adaboost classifier, some unnecessary training data features can be excluded, and the key is placed on the key training data; the algorithm is a weak classification algorithm promotion process, and the classification capability of the data can be improved through continuous training in the process; the whole process is as follows:
(1) firstly, learning a C peptide data sample to obtain a first weak classifier;
(2) forming a new N training samples by the samples and other new data, and obtaining a second weak classifier by learning the samples;
(3) adding other new samples into the samples to form another new N training samples, and learning the samples to obtain a third weak classifier;
(4) a final boosted strong classifier; i.e. into which class a certain data is to be passed;
the Adaboost algorithm flow is specifically as follows:
given a training data set T = { (x1, y1), (x2, y2) … (xN, yN) }, where examples, and example space, yi, belong to the label set { -1, +1}, the purpose of Adaboost is to learn a series of weak classifiers, or basic classifiers, from the training data and then combine these weak classifiers into a strong classifier;
the algorithm flow of Adaboost is as follows:
step 1, firstly, initializing weight distribution of training data; each training sample is initially given the same weight: 1/N;
step 2, performing multiple iterations, and using M =1, 2.. times, M to indicate the first iteration round
Learning with a training data set with weight distribution Dm to obtain a basic classifier (selecting a threshold value with the lowest error rate to design the basic classifier):
b. calculating a Classification error Rate of Gm (x) on a training data set
As can be seen from the above equation, the error rate em of Gm (x) on the training data set is the sum of the weights of the misclassified samples Gm (x);
c. the coefficient of Gm (x) is calculated, am representing the importance of Gm (x) in the final classifier (objective: obtaining the basic classifier)The weight occupied in the final classifier; note: this formula is written to am =1/2ln ((1-em)/em) more accurately because of the base number Is thatThe natural logarithm e, so writing a log with In is easy for a person to mistakenly assume that the base is 2 or another base, as follows):
from the above equation, when em < =1/2, am > = 0, and am increases with decreasing em, meaning that the basic classifier with smaller classification error rate has greater effect in the final classifier;
d. the weight distribution of the training data set is updated (in order to obtain a new weight distribution of the samples) for the next iteration
So that the weight of the misclassified samples by the basic classifier Gm (x) is increased, and the weight of the correctly classified samples is decreased; in this manner, in this way, the AdaBoost method can "focus" or "focus on" those samples that are less readily separable;
where Zm is a normalization factor, making Dm +1 a probability distribution:
step 3, combining each weak classifier
The final classifier is thus obtained as follows:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110824477.1A CN113571180A (en) | 2021-07-21 | 2021-07-21 | C-peptide layering and organ function-based type 2 diabetes artificial intelligent diagnosis and treatment management system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110824477.1A CN113571180A (en) | 2021-07-21 | 2021-07-21 | C-peptide layering and organ function-based type 2 diabetes artificial intelligent diagnosis and treatment management system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113571180A true CN113571180A (en) | 2021-10-29 |
Family
ID=78166001
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110824477.1A Withdrawn CN113571180A (en) | 2021-07-21 | 2021-07-21 | C-peptide layering and organ function-based type 2 diabetes artificial intelligent diagnosis and treatment management system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113571180A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115083601A (en) * | 2022-07-25 | 2022-09-20 | 四川省医学科学院·四川省人民医院 | Type 2diabetes auxiliary decision making system based on machine learning |
| CN117894424A (en) * | 2024-03-14 | 2024-04-16 | 四川省医学科学院·四川省人民医院 | Recommendation system for constructing T2DM patient drug scheme based on deep learning and reinforcement learning |
-
2021
- 2021-07-21 CN CN202110824477.1A patent/CN113571180A/en not_active Withdrawn
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115083601A (en) * | 2022-07-25 | 2022-09-20 | 四川省医学科学院·四川省人民医院 | Type 2diabetes auxiliary decision making system based on machine learning |
| CN117894424A (en) * | 2024-03-14 | 2024-04-16 | 四川省医学科学院·四川省人民医院 | Recommendation system for constructing T2DM patient drug scheme based on deep learning and reinforcement learning |
| CN117894424B (en) * | 2024-03-14 | 2024-05-14 | 四川省医学科学院·四川省人民医院 | Recommendation system for constructing T2DM patient drug scheme based on deep learning and reinforcement learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hooper et al. | Which frail older people are dehydrated? The UK DRIE study | |
| Happ et al. | Nurse-patient communication interactions in the intensive care unit | |
| Shock | Normal human aging: The Baltimore longitudinal study of aging | |
| Cohen | Plan and operation of the NHANES I Epidemiologic Followup Study, 1982-84 | |
| Dickinson et al. | A comparison of cognitive structure in schizophrenia patients and healthy controls using confirmatory factor analysis | |
| CN107591203A (en) | A kind of disease pre-warning method and system in knowledge based storehouse | |
| CN113571180A (en) | C-peptide layering and organ function-based type 2 diabetes artificial intelligent diagnosis and treatment management system | |
| Khader et al. | Diabetes mellitus and treatment outcomes in Palestine refugees in UNRWA primary health care clinics in Jordan | |
| WO2021073255A1 (en) | Time series clustering-based medication reminder method and related device | |
| CN106919804A (en) | Medicine based on clinical data recommends method, recommendation apparatus and server | |
| Tootooni et al. | CCMapper: An adaptive NLP-based free-text chief complaint mapping algorithm | |
| CN112102956A (en) | Method and system for improving compliance of diabetic patient | |
| Headley et al. | Using the pregnancy perception of risk questionnaire to assess health care literacy gaps in maternal perception of prenatal risk | |
| Bigdeli et al. | Factors affecting self-care in patients with type II diabetes using path analysis | |
| CN107066816A (en) | Medical treatment guidance method, device and server based on clinical data | |
| Rasubala et al. | Digital twin roles in public healthcare | |
| Shapiro et al. | A program to measure the impact of multiphasic health testing on health differentials between poverty and nonpoverty groups | |
| Athar et al. | Impact of general education status on glycemic control in patients of diabetes mellitus | |
| Shapiro | Evaluation of two contrasting types of screening programs | |
| Shita et al. | Predictors of glycemic control among patients with type 2 diabetes in North West Ethiopia: a longitudinal study | |
| Bancks et al. | Association of type 2 diabetes subgroups with cognitive status without modification from lifestyle intervention | |
| Agofure et al. | Knowledge of Diabetes Mellitus and Adherence to treatment among patients with Type-2 diabetes mellitus attending a Tertiary Facility in Southern Nigeria | |
| Benech et al. | PMU124 patients knowledge on microbiota and acceptability of fecal microbiota transplantation in various chronic diseases | |
| Jayasree et al. | Psychiatric Morbidity and its Correlations in Diabetic Patients in a Tertiary Care Hospital | |
| Phaswana-Mafuya et al. | Study on Global AGEing and Adult health (SAGE), Wave 1: South Africa national report |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20211029 |