JP2023155917A - Examination data comparison device - Google Patents

Abstract

To predict future examination data (disease state) on the basis of the correlation between treatment information and a variety of examination data and make the result useful for the treatment of the disease.SOLUTION: The present invention includes: treatment storage means for storing treatment information of a treatment target person; public information storage means for storing public information derived from a treatment; and examination data storage means for storing a variety of examination data of the treatment target person. The public information stored in the public information storage means and the examination data stored in the examination data storage means are compared and future examination data is predicted.SELECTED DRAWING: Figure 1

Description

The present invention records various test data in association with the treatment details of the patient, predicts how the test data will change when the same treatment continues, and in some cases , the aim is to motivate the person to be treated to change (improve) the treatment content in order to change (improve) the test data.
In the present invention, treatment refers to the act of exerting physical, chemical, biological, or psychological effects on the body of a person to be treated based on a doctor's instructions to improve or cure an illness or injury, i.e., medication, This includes surgery, radiation, rehabilitation, conservative treatment, etc., but does not include lifestyle habits that are not based on a doctor's instructions.

For example, if a patient continues treatment without changing the current treatment content, the trend of test results can be predicted, and the patient can be encouraged to understand the need for treatment and take treatment, thereby improving the patient's condition. The aim is to lead to improved actions.
The purpose is to predict the future value of test data if the current treatment is continued or not, and to make the patient aware of the importance of changing or continuing the treatment.

Appropriate medical treatments have long been loudly advocated from the perspective of reducing medical costs, and the causal relationship between treatment details and various diseases has been published in numerous medical papers and is well known.
For example, the increase in the number of dialysis patients due to the interruption of treatment for early stage diabetes, which does not exhibit any subjective symptoms, and the resulting aggravation of the disease, has become a social problem. Furthermore, although a wide variety of therapeutic drugs are available, it is difficult to match the drug to each individual patient, and many publicly known papers describe various drug administration methods.

However, recently it has been known that diseases should be evaluated using multiple test values, but in reality, for example, in the case of diabetes, confirmation is only performed using blood sugar levels and HbA1C. It is not possible to make a comprehensive evaluation based on the results of multiple tests involving multiple organs, decreased visual acuity, blood pressure, lipids, decreased renal function, etc., which become problematic as the disease progresses.

For this reason, when postprandial blood sugar increases are suppressed after medication, HbA1C becomes the standard value, and the treatment is evaluated to be effective, side effects and related diseases caused by the treatment may be overlooked or become more serious.

Patent No. 6473988

Patent Document 1 mentioned above records a subject's lifestyle habits and various test data in association with each other, and predicts how the test data will change if the subject continues to have the same lifestyle habits. In some cases, inventions are described that attempt to motivate subjects to change (improve) their lifestyle habits in order to change (improve) test data.
However, in this invention, there is no concept of correlating and recording the treatment contents during treatment and test data.
Therefore, it has not been possible to predict the future value of test data when the current treatment is continued or not, and to make the patient aware of the importance of changing or continuing the treatment.

In this regard, the present invention aims to associate treatment details with various test results, and provide the person being treated with guidelines for reviewing (or not revising, as the case may be) future treatment details based on the test results.

(1) The present invention provides an apparatus that compares treatment information and various test data of a treatment target with various known test data derived from the treatment and predicts future test data when the treatment continues. , a treatment storage means for storing treatment information of the treatment subject; a known information storage means for storing known information regarding treatment results derived from various treatments; and an examination data storage for storing various test data of the treatment subject. Means is provided, and by periodically inputting information regarding the treatment information and the test data into the test data storage means, the test data of the person to be treated is accumulated, and the data and the known information storage means are stored. The stored known information is collated with artificial intelligence, an initial prediction function is generated using artificial intelligence from the known information, and the prediction function is combined with test data accumulated in the past multiple times. This test data prediction device is characterized in that it adds treatment information, performs additional learning on an artificial intelligence prediction formula, and predicts future test data as the analysis result.
According to the present invention, treatment information (for example, data on the administration of a specific drug to a treatment target) and data described in known medical papers (data regarding the relationship between blood sugar levels, HbA1C, and specific drugs) are combined. In comparison, it may be predicted that the current symptoms (blood sugar level and HbA1C data) will not decrease simply by continuing to administer a specific drug in this state.
In such cases, it is possible for the person being treated to choose whether the increase in blood sugar levels and HbA1C can be suppressed or lowered by applying another drug or treatment method. becomes.

(2) The present invention provides an apparatus for associating treatment information and test data of a treatment target and predicting future test data based on past test data when a certain treatment is continued. A treatment storage means for storing information, a known information storage means for storing known information regarding treatment results derived from various treatments, and an examination data storage means for storing various examination data of the treatment subject, Test data under the same treatment is stored multiple times at intervals of time, past test data conducted each time is accumulated, and the data and the publicly known information stored in the publicly known information storage means are stored using artificial intelligence. An initial prediction function is generated using artificial intelligence from the publicly known information, and test data and treatment information accumulated in the past are added to the prediction function to create an artificial intelligence prediction formula. This is an inspection data prediction device that performs additional learning and predicts future inspection data as an analysis result.

According to the present invention, for example, administration of a specific antihypertensive drug and blood pressure data to a patient to be treated are memorized multiple times at intervals, and past test data conducted each time is stored, thereby allowing It is predicted that if a specific antihypertensive drug is continued to be administered, the increase in blood pressure will be suppressed.

In such a case, the person to be treated should use past administration of a specific antihypertensive drug and blood pressure data to determine how long they should continue administering that specific antihypertensive drug in the future to prevent an increase in blood pressure or to reduce it. This means that it is possible to predict what will happen.

(3) It is desirable that the treatment information includes medication, surgery, radiation, rehabilitation, or conservative treatment.
Further, the treatment information may include information on relatives other than the patient himself/herself.

(4) It is desirable that the treatment information includes genetic information and/or medical history of the person to be treated.

(5) It is desirable that the test data include either various sample tests or physical information.
The sample test includes test data on sugar metabolism, lipid metabolism, liver function, anemia, and kidney function, and the physical information may include age, sex, height, weight, waist circumference, and blood pressure.

(6) Future test data may be predicted when the treatment content is changed.
For example, based on blood pressure data of a patient who has changed their past treatment and increased the dosage of a specific drug, we can determine what dosage of that specific drug should be administered in the future to prevent a rise in blood pressure. It will be possible to predict whether it can be prevented or whether it can be reduced.

(7) The present invention provides a treatment memory for storing treatment information of the treatment target in an apparatus that predicts various test data derived from treatment information of the treatment target and compares the predicted data with actual test data. means, a known information storage means for storing known information derived from the treatment, a prediction means for predicting various test data derived from the treatment in association with the known information, and storing various test data of the person to be treated. and a prediction data output means for outputting predicted data from the prediction means according to the treatment information stored in the treatment storage means of the treatment subject, and comparing the test data and the predicted data. , a data difference storage means for storing the difference, the difference between the inspection data and the prediction data stored in the data difference storage means is collated by artificial intelligence, and the artificial intelligence is calculated from the known information. This is used to generate an initial prediction function, and by adding test data and treatment information accumulated in the past to the prediction function, the artificial intelligence prediction formula is additionally trained, and as the analysis result, the treatment target is It is characterized by being identified as a person's constitution.

According to the present invention, for example, administration of a specific antihypertensive drug to a treatment subject and blood pressure data can be stored multiple times at intervals, and past test data performed each time can be recorded in association with treatment information. can.
Therefore, it is possible to store the administration of a specific antihypertensive drug and blood pressure data to a person to be treated multiple times at intervals, and record the past test data performed each time.

(8) The same treatment information and test data may be stored multiple times at different times, and the differences between the test data and predicted data performed each time may be accumulated.

(9) The present invention compares various test data of a person to be treated with various known test data derived from the treatment, and identifies diseases that are predicted to develop in the future if the treatment is continued and the disease. In an apparatus for presenting improvement contents of treatment for preventing cancer, a treatment storage means for storing treatment information of the treatment subject, a known information storage means for storing known information derived from the treatment, and a treatment storage means for storing treatment information of the treatment subject; test data storage means for storing various test data, and compares the known information stored in the known information storage means with the test data stored in the test data storage means, and compares the test data with the treatment information. By periodically inputting information regarding the test data into the test data storage means, the test data of the treatment target is accumulated, and the data and the known information stored in the public information storage means are stored in the artificial intelligence. An initial prediction function is generated using artificial intelligence from the publicly known information, and test data and treatment information accumulated in the past are added to the prediction function to create an artificial intelligence prediction formula. Additional learning is performed, and the results are supplied to the selection of the body's constitution, prediction of post-examination data, and furthermore, recommended treatment, to improve the accuracy and probability of each constitution selection, prediction, and selection of recommended treatment details. It is characterized by improving the degree of

According to the present invention, for example, if a test result is obtained that shows that blood pressure does not decrease even if a person to be treated takes a specific antihypertensive drug, the test result and information described in a known document, i.e., the relevant By comparing this with the information that blood pressure decreases when taking a specific antihypertensive drug, it is determined that the person to be treated has an unusual constitution (body type).

Therefore, if the subject to be treated has hypertension, consideration should be given to changing the treatment method, such as changing the drug.

(10) It is desirable that the treatment improvement recommendation information includes test items to be improved and test value targets for avoiding the estimated future risk of aggravation.
(11) It is desirable to predict multiple treatment types and the corresponding risks of future improvement or aggravation, and to present the results to the patient so that he or she can select the most appropriate treatment from the multiple treatment types.
For example, there are multiple treatment methods to lower HbA1c levels, and depending on the patient, there are cases where it is better to lower HbA1c levels rapidly and cases where it is better to lower them gradually. If changes in HbA1c values can be predicted and provided to patients, more appropriate treatment can be provided.

According to the present invention, it is possible to associate treatment contents with various test results, and to provide the person receiving treatment with guidelines for reviewing (or not revising, in some cases) future treatment contents based on the test results, which is a great measure to prevent the worsening of the disease. You will be able to execute this.

FIG. 2 is a control block diagram of an embodiment of the present invention. FIG. 2 is a block diagram showing the relationship between various test data and treatment information according to an embodiment of the present invention. FIG. 3 is a diagram showing a process for predicting a future disease state from various test data and treatment information according to an embodiment of the present invention. FIG. 3 is a diagram illustrating a process for inputting personal attributes, medication information, etc., blood sugar level, etc., and outputting future medication information, etc. using the next test value prediction engine, according to an embodiment of the present invention. FIG. 3 is a relationship diagram showing the relationship between various test data and treatment information according to an embodiment of the present invention. FIG. 3 is a diagram showing the contents of various types of medication information (treatment information) according to an embodiment of the present invention. It is a line graph showing prediction of inspection data according to an embodiment of the present invention. FIG. 2 is a diagram illustrating guidelines for maintaining the status quo and improving targets in an embodiment of the present invention. It is a table showing the relationship between lifestyle habits and test values according to an embodiment of the present invention. FIG. 2 is a diagram showing an explanatory text of an embodiment of the present invention. FIG. 9 is a diagram following FIG. 9 of an embodiment of the present invention. 3 shows the therapeutic effect of artificial intelligence according to an embodiment of the present invention. 12 is a graph of the contents of FIG. 12 according to an embodiment of the present invention.

FIG. 1 is a control block diagram of the entire device. A central processing unit (CPU) 1 includes, via a bus line 2, a treatment information memory 3 for storing treatment information of a person to be treated; A test data memory 4 that stores test data, a public information memory 5 that stores public information published in medical papers, etc., and a difference storage memory 9 that stores differences between predicted data and test data. Various types of memory are connected.

The bus line 2 also includes a predicted data display section 6 that displays test data predicted by inputting treatment information of a person to be treated, and a test data comparison section 7 that compares past test data and current test data. , a test data comparing section 8 that compares the test data with the known information, and a recommended treatment display section 10 that teaches the treatment to be recommended to the person to be treated based on the test data.
Further, connected to the bus line 2 are an examination data input section 11 for inputting various examination data, and a treatment information input section 12 for inputting treatment information.

To predict test data using this device, first, information regarding the treatment of the patient as shown in FIG. 6 is input.

Specifically, the drugs currently being taken by the person to be treated, such as diabetes drugs (oral drugs and insulin) as shown in Figure 6, high blood pressure drugs, anticoagulants, heart disease drugs, drugs that lower cholesterol, etc. Participants were asked to select the drug they are currently taking from among the following: , gastrointestinal drugs, gout drugs, tranquilizers, adrenocortical hormones, and other anticancer drugs.
In addition, you will be asked to enter other treatment information such as your surgical history and radiation treatment history.
Further, treatment information includes the family history of the person to be treated (blood pressure, blood sugar level, lipids, liver function, treatment experience, weight at age 20), etc.

Next, information regarding various inspection data as shown in FIG. 5 is input.
The test items include physique (abdominal circumference, weight, BMI), blood pressure (systolic blood pressure, diastolic blood pressure), glucose metabolism (fasting blood sugar level, estimated average blood sugar level, HbA1c value), lipid metabolism (HDL cholesterol, LDL cholesterol, Examples include neutral fat and liver function (AST, ALT, γ-GTP).

As shown in Figures 2 and 4, by periodically inputting information regarding treatment information and test data as described above, data on treatment recipients is accumulated, and the data is introduced in medical papers, etc. An artificial intelligence (AI) engine collates various publicly known information, and the analysis results include future test value prediction (deterioration or improvement), genetic test value prediction, deterioration risk determination, disease condition deterioration factor analysis, and disease condition. Improvement factor analysis (sleep) is output.

From these analysis results, treatment evaluation (future test values, treatment details), test value prediction graph, test value targets for improving and curing the condition, risk of aggravation and development of complications, action plan for recovery and priority ranking are obtained. , evaluates the relationship between eating habits and diseases for each test item, displays recommended treatments as a result, and outputs medication information that can achieve treatment goals.
In addition to medication information, recommended treatment information such as surgery and radiation therapy is also output.

Those receiving treatment can understand that their physical condition will improve if they carry out the recommended treatment and have regular examinations.
On the other hand, it is understandable that even if you do not carry out the recommended treatment, your physical condition will worsen if you conduct regular examinations.
Therefore, by using this device, the person to be treated can better manage their physical condition.

Figure 7 is a line graph showing how future test values will change if a specific treatment target continues with the current treatment. The current value is predicted to be 6.5%, 6.6% in 3 years, 6.8% in 6 years, and 7.3% in 9 years.
At the bottom is a radar chart that predicts the risk of complications based on treatment methods, lifestyle habits, and test results.

Next, according to FIG. 5, the relationship between various test data and treatment information will be explained.
In this figure, one vertical column on the left side represents a state in which test data and treatment information are input and finally test data and treatment information are accumulated.

That is, by first inputting test data and treatment information and then selecting the type of current pathological condition, subsequent test data is predicted. For example, if a person to be treated has a constitution that does not affect their blood pressure even if they take a specific antihypertensive drug, posttest test data will be predicted based on this assumption. Also, if you have a constitution that lowers your blood pressure when you take the antihypertensive drug, test data will be predicted accordingly.

Next, a treatment recommended for improving the test data is selected based on the predicted value of the test data. For example, for a patient whose blood pressure does not drop even after taking a specific antihypertensive drug, a recommendation would be to prescribe another antihypertensive drug.
The system then accumulates multiple test data and treatment information to provide treatment targets with guidelines for future treatment.

In selecting the constitution of the person to be treated, the modification of the person to be treated is selected based on the constitution selection logic according to several types disclosed in public information such as papers.
In addition, an initial prediction function is generated using AI (artificial intelligence) from public information such as papers, and test data and treatment information accumulated in the past are added to the prediction function to create an AI prediction formula. Have them learn additionally.
The results are then supplied to the selection of the body's constitution, prediction of post-test data, and recommended treatment information, improving the accuracy and probability of each constitution selection, prediction, and recommended treatment selection. Plan.

Next, referring to FIG. 3, a description will be given of a mode in which the risk of aggravation of the current disease is suppressed if the treatment is improved so that the treatment target meets the improvement recommendation information.
First, we track the medical examinations of the person being treated over the past six years or more, estimate the future type of disease that could become more severe if the treatment continues as it is, and then estimate the severity of the estimated future disease. By setting test value targets for avoidance and informing the treatment target of the test value targets, the possibility of suppressing the onset of symptoms increases.

That is, as shown in the lower part of Figure 3, if the physical condition and treatment are left as is, the risk of future deterioration (severe illness) is predicted from test value trends and disease types. To avoid this risk, it is necessary to set test value targets to avoid future diseases, and to set treatment improvement targets to achieve those goals. The aim is to prevent the current disease from becoming more severe.

Specifically, for the diabetic patients shown in Figure 8, we set the goal of maintaining/improving the condition as follows: ``Drug therapy alone will not control blood sugar levels unless calorie intake is suppressed and basal metabolic energy is increased.'' It is predicted that this will become difficult. Please make the following behavioral changes and avoid taking additional medications or increasing the dosage."
More specifically, we offer the following suggestions as initiatives aimed at reducing dosage and discontinuing medication. They are listed from top to bottom in order of effectiveness (the display is small in the figure, so they are listed below).

・The title ``Continue'' suggests ``taking medicine correctly.''
More specifically, ``Dietary therapy and exercise therapy are the basic treatments for diabetes. "If good blood sugar control cannot be achieved with physical therapy and exercise therapy, drug therapy is used to suppress the occurrence and progression of complications."
・The first title, ``Increase,'' suggests ``taking 8,000 steps in your daily life, of which 20 minutes are moderate activities.''
More specifically, he says, ``If you can't walk 8,000 steps for 20 minutes, first make it a habit of 4,000 steps for 5 minutes over two months, then 6,000 steps for 10 minutes over two months...'' Make suggestions.
・The first title, ``Make it a habit,'' suggests ``Eating vegetables and mushrooms, then meat or fish, then rice and bread.''
・The second title, ``Make it a habit,'' suggests ``Eating three hours before going to bed.''
・The third item, titled ``Make it a habit,'' suggests ``chew each bite slowly and thoroughly 20 to 30 times.''
More specifically, he makes suggestions such as eating more vegetables, avoiding pour-over-eating, and putting food in your mouth until you have chewed it a set number of times.
- The first title, ``Reduce,'' suggests ``Avoid single-grain dishes such as noodles, fried rice, rice bowls, sweet breads, and rice balls.''
・The second title, ``Reduce,'' suggests ``Reduce eating foods that use a lot of oil, such as fried foods, Chinese food, and Western-style food.''
・The third title, ``Reduce,'' suggests ``Get used to lighter flavors by using seasonings other than salt.''
More specifically, he suggests adding ``accents to the taste by using vinegar, citrus juice, flavors such as ginger and garlic, spices such as wasabi and Japanese pepper, and herbs.''
・The third item, titled ``Increase,'' suggests ``Eating blue fish dishes that contain high-quality fat and protein.''
More specifically, he makes suggestions such as purchasing fillets, having them prepared at the store, using a frying pan or microwave, and using canned food to make cooking easier.

As shown in Figure 9, the vertical axis lists test values (test contents) such as blood pressure, blood sugar, and triglycerides, and the horizontal axis List the above-mentioned proposal contents in the table below, and link "blood pressure" on the vertical axis to "don't drink too much alcohol" on the horizontal axis as items that are given high priority in the artificial intelligence analysis, and put them in "△". It is represented by a mark with "!".

Furthermore, in Figure 10, the title ``Explanation'' is ``What is happening inside your body...!? Dangerous deterioration of the condition and development of complications may occur in this way.'' ``What if hyperglycemia (diabetes) is left alone?'' The condition is expressed with an explanatory text and a cross-sectional diagram of the inside of a blood vessel.

Then, as shown in FIG. 11, various countermeasures are displayed with the subtitle "Suppress the rapid rise in blood sugar level by suppressing meal speed."

Figure 12 shows that using artificial intelligence, a patient with a current HbA1c value of 8.4, a fasting blood sugar level of 180, and a 2-hour postprandial blood sugar level of 220 was treated with no treatment and with three types of treatment. The predicted values for three months later are shown in a table.
Here, drug A (biguanide drug) mainly acts on the liver and has the effect of lowering blood sugar levels by suppressing excessive release of glucose from the liver, improving insulin resistance and lowering blood sugar levels. It is something to lower.

Drug B (sulfonylurea (SU) drug) acts on the pancreas to promote insulin secretion, and is effective for the type of diabetes where insulin secretion is weakened, which is common in Japanese people.
In Figure 12, if the patient does not undergo any further treatment, artificial intelligence predicts that the HbA1c value will be 9.1, the fasting blood sugar level will be 200, and the 2-hour postprandial blood sugar level will be 240 after 3 months. are doing.
If the patient performed only exercise and diet, artificial intelligence predicted that the HbA1c value would be 7.8, the fasting blood sugar level would be 160, and the 2-hour postprandial blood sugar level would be 190 after 3 months. ing.

Furthermore, if the patient takes the biguanide drug (A) and also exercises and diets, after 3 months, the HbA1c value is 6.8, the fasting blood sugar is 120, and the 2-hour postprandial blood sugar is 6.8. Artificial intelligence predicts that it will reach 160.
Furthermore, if the patient takes the sulfonylurea drug (B) and also exercises and diets, the HbA1c value will be 6.1, the fasting blood sugar will be 90, and the 2-hour postprandial blood sugar will decrease after 3 months. Artificial intelligence predicts that the number will rise to 110.

FIG. 13 is a graph of the predicted HbA1c value of the patient 3 months later.
Depending on the patient, it is not always advisable to lower the HbA1c value rapidly, but it may be recommended to lower it gradually.
In the case of the patient shown in FIG. 13, it is considered appropriate by the doctor to lower the HbA1c value gradually to some extent, and it is recommended that the patient take a biguanide drug (A) and use exercise and diet therapy together. In the figure, it is surrounded by □.

1 Central processing unit (CPU)
2 Bus line 3 Treatment information memory 4 Test data memory 5 Known information memory 6 Predicted data display section 7 Test data comparison section 8 Test data comparison section 9 Difference storage memory 10 Recommended treatment display section 11 Test data input section 12 Treatment information input section

Claims (2)

In a device that predicts various test data derived from treatment information of a treatment target and compares the predicted data with actual test data,
treatment storage means for storing treatment information of the treatment subject;
a known information storage means for storing known information derived from treatment;
Prediction means for predicting various test data derived from treatment by associating it with the publicly known information;
Test data storage means for storing various test data of the treatment target;
Predicted data output means for outputting predicted data from the prediction means according to treatment information stored in the treatment storage means for the treatment subject;
data difference storage means for comparing the inspection data and the prediction data and storing the difference;
is equipped with
collating the difference between the test data and the predicted data stored in the data difference storage means using artificial intelligence;
Generate an initial prediction function from the known information using artificial intelligence, add test data and treatment information accumulated in the past to the prediction function, and cause the artificial intelligence prediction formula to additionally learn,
The test data comparison device is characterized in that, as a result of the analysis, the constitution of the person to be treated is specified. 2. The test data comparison device according to claim 1, further comprising storage means for storing the same treatment information and test data multiple times at intervals, and storing differences between the test data and predicted data for each time.