CN115019935A - Nutritional decision method and system for sepsis patient - Google Patents

Abstract

The invention relates to a nutrition decision method and a system for sepsis patients, wherein the method comprises the following steps: s1: acquiring basic information and inpatient period indexes of a patient, and constructing a patient information record; s2: inputting data in the information record into an auxiliary judgment model, carrying out feeding feasibility evaluation on the patient, and switching to S3 if the patient can carry out nutrition feeding, otherwise, entering an observation period; s3: judging the gastrointestinal damage degree of the patient, and determining a nutrition supply mode: recommending an parenteral nutrition mode if severe injury occurs, and recommending an enteral nutrition mode if mild or moderate injury occurs; s4: inputting data in the information record into an auxiliary recommendation model, taking the output of the model as an auxiliary energy value and combining related data in the information record to obtain a final energy metabolism value; and (5) obtaining an auxiliary nutrition recommended scheme according to the nutrition supply mode in the S3. The method provided by the invention can intelligently provide a personalized nutrition decision scheme for patients.

Description

A nutritional decision-making method and system for sepsis patients

technical field

The invention relates to the field of computer technology, in particular to a nutritional decision-making method and system for sepsis patients.

Background technique

Sepsis is a systemic inflammatory response syndrome caused by the invasion of pathogenic microorganisms such as bacteria into the body. It is one of the serious complications of burns, trauma, shock, infection, major surgery and other clinically critically ill patients. It is an important factor that induces septic shock and multiple organ dysfunction syndrome. The pathological mechanism of sepsis is complex, involving pathogen invasion, inflammatory factor release, coagulation mechanism disorder and microcirculation dysfunction leading to tissue metabolic disorder and organ failure. Cellular energy depletion is a potential reason for the failure of many sepsis treatments. Nutritional interventions for sepsis patients can effectively improve the prognosis of sepsis patients and reduce the mortality rate.

At present, without the use of indirect calorimetry, doctors mostly use the formula method to calculate the resting metabolic value of the patient. Multiply by the stress factor. However, different diseases, different stress factors, and the application of drugs will all have unpredictable effects on energy consumption. The use of stressors is often judged by clinicians, resulting in subjective factors that affect the final outcome. Formula prediction is difficult to accurately assess resting energy metabolism value, monitor the changing law of resting energy metabolism value, and guide individualized nutrition.

Indirect calorimetry, i.e. IC method, is the gold standard for calculating the energy metabolism of patients. It can accurately measure the resting energy metabolism value of patients in a certain period of time, so as to grasp the changing trend of the resting metabolic value of critically ill patients. Different situations specify individualized nutrition supply strategies.

The application of computer technology in the field of clinical medicine not only improves efficiency, but also has better accuracy and timeliness than manual labor. At present, nutrition decision-making systems in the medical field are mainly divided into two categories: clinical nutrition decision-making systems based on data modeling and clinical nutrition decision-making systems based on data modeling. The clinical nutrition decision-making system of clinical guidelines, based on knowledge, usually takes clinical guidelines and best practices as knowledge sources, and manually or automatically converts it into rule-based expert knowledge (IF-THEN rules) for inference decision support. The main disadvantage of this method is that the rules cannot fully enumerate all clinical situations, and the rules have limited expressive power for non-deterministic things. On the other hand, data modeling methods use machine learning methods, which have developed rapidly due to high accuracy and relatively easy construction. However, a key disadvantage of such methods is the limited interpretability of the results produced (black-box behavior), which leads to low acceptance by health professionals.

SUMMARY OF THE INVENTION

In order to solve the above technical problems, the present invention provides a nutritional decision-making method and system for sepsis patients.

The technical solution of the present invention is: a nutritional decision-making method for sepsis patients, comprising:

Step S1: obtain the patient's basic information and indicators during hospitalization, and construct the patient's information record;

Step S2: Input the data in the information record into the auxiliary judgment model, evaluate the feeding feasibility of the patient, and transfer to S3 if nutritional feeding can be performed, otherwise the patient enters the observation period;

Step S3: Judging the degree of gastrointestinal injury of the patient according to the data in the information record, and determining the nutrition supply method: if the injury is severe, the parenteral nutrition method is recommended, and if the injury is mild or moderate, the enteral nutrition method is recommended;

Step S4: Input the data in the information record into the auxiliary recommendation model to obtain auxiliary data of energy metabolism value, and determine the final energy metabolism value in combination with the basic information and the indicators during hospitalization; The numerical value is combined with the nutrition supply method to obtain the final recommended supplementary nutrition scheme.

Compared with the prior art, the present invention has the following advantages:

The invention discloses a nutrition decision-making method for patients with sepsis. According to the accurate measurement of the resting energy metabolism value of the patients with sepsis, a personalized auxiliary nutrition decision-making scheme is intelligently provided for the patients, and the nutrition treatment process is realized. To improve the efficiency of clinical nutrition treatment in the whole hospital.

Description of drawings

1 is a flowchart of a nutritional decision-making method for sepsis patients according to Embodiment 1 of the present invention;

2 is a flowchart of a nutritional decision-making method for sepsis patients according to Embodiment 2 of the present invention;

3 is a structural block diagram of a nutrition decision-making system for sepsis patients according to Embodiment 3 of the present invention;

FIG. 4 is a structural block diagram of a nutrition decision-making system for sepsis patients according to Embodiment 4 of the present invention.

Detailed ways

The invention provides a nutrition decision-making method for patients with sepsis, which can intelligently provide patients with a personalized auxiliary nutrition decision-making scheme, realizes the process, standardization and informationization of the nutrition decision-making process, and improves the clinical nutrition treatment in the whole hospital. efficiency.

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below through specific implementation and in conjunction with the accompanying drawings.

Example 1

As shown in FIG. 1 , a nutritional decision-making method for sepsis patients provided by an embodiment of the present invention includes the following steps:

Step S1: obtain the patient's basic information and indicators during hospitalization, and construct the patient's information record;

Step S2: Input the data in the information record into the auxiliary judgment model, and evaluate the feeding feasibility of the patient. If nutritious feeding can be performed, transfer to S3, otherwise the patient enters the observation period;

Step S3: Judging the degree of gastrointestinal injury of the patient according to the data in the information record, and determining the nutrition supply method: if the injury is severe, the parenteral nutrition method is recommended, and if the injury is mild or moderate, the enteral nutrition method is recommended;

Step S4: Input the data in the information record into the auxiliary recommendation model, obtain auxiliary data of energy metabolism value, and determine the final energy metabolism value by combining the basic information and indicators during hospitalization; combine the final energy metabolism value with the nutrition supply method to obtain the final energy metabolism value. supplementary nutrition recommendations.

In one embodiment, the above-mentioned step S1: obtaining the basic information of the patient and the indicators during hospitalization, and constructing the patient information record, which specifically includes:

Step S11: Acquire the basic information of the patient, including: the patient's direct information measured when the patient is admitted to the hospital and the patient's energy metabolism information measured by professional equipment; wherein,

The patient's direct information measured at the time of admission includes: gender, level of consciousness, age, height, weight, BMI, and time of admission;

When the patient is admitted to the hospital, the energy metabolism information measured by professional equipment includes: resting energy metabolism value, respiratory quotient and corresponding measurement time;

Step S12: Obtain the indicators of the patient during hospitalization, including: laboratory inspection data, vital sign data, duration of ICU admission, and duration of nutritional supply, wherein,

Laboratory examination data include: blood routine, blood glucose, liver and kidney function, electrolyte level and coagulation function;

Vital signs data include: heart rate, body temperature, systolic blood pressure, diastolic blood pressure, pulse, pulse oximetry, respiratory rate, mean arterial pressure, fluid balance (water in and out: urine volume, various drainage volumes, intake), body weight Changes, dietary intake, albumin, types of concomitant diseases and nutrient supply;

As well as the length of the patient's admission to the ICU and the length of nutritional supply;

In the embodiment of the present invention, the above-mentioned collected basic information of patients and indicators during hospitalization are stored in the multi-center emergency triage data related database (CETAT database) of the China Emergency Specialist Medical Association.

In one embodiment, the calculation of the resting energy metabolism value in the above step S11 specifically includes:

Step S111: Acquire the oxygen uptake VO ₂ and the carbon dioxide exhalation VCO ₂ of the patient within a specific time;

Step S112: Calculate the resting energy metabolism value Y using formula (1):

Y(kcal/day)=1.44×([VO ₂ (ml/min)×3.94]+[VCO ₂ (ml/min)×1.11]) (1).

In one embodiment, the above step S2: the data in the information record is input into the auxiliary judgment model, the feeding feasibility assessment is performed on the patient, and if nutritious feeding can be performed, it is transferred to S3, otherwise the patient enters the observation period, which specifically includes:

Step S21: performing data cleaning on the data in the information record to obtain a cleaned data set;

The embodiment of the present invention extracts information on patients with sepsis admitted to the First Affiliated Hospital of the University of Science and Technology of China from January 2020 to July 2021 from the multi-center emergency triage data related database (CETAT database) of the China Emergency Specialist Medical Association Records, a total of 48,575 patient information, including vital signs data, laboratory data, demographic data, etc.

First, clean the missing values of the data in the information records: remove the data with low importance and more than 80% missing values in the data set, and fill in the missing data with high importance with the median of the missing data.

Second, clean the format and content of the data in the dataset:

1) The display formats of time, date, numerical value, etc. are inconsistent, and they are all unified in standard format;

2) There are characters that should not exist in the content, such as English letters appearing in the name;

3) The content does not match the expected content of this field, such as spaces in the name, or Chinese characters in the heart rate information.

Finally, the non-essential medical data is cleaned, such as removing information such as ID cards.

After the above data cleaning, a cleaned data set containing valid data is obtained, which is used for subsequent model training.

Step S22: Select the auxiliary judgment feature value from the data set according to the NRS-2002 nutritional risk screening table, and train the auxiliary judgment model until the trained auxiliary judgment model is obtained;

According to the NRS-2002 nutritional risk screening table, the embodiment of the present invention selects auxiliary judgment characteristic values from the data set constructed in step S21: gender, age, BMI, dietary intake, albumin, and types of concomitant diseases; The selection of the above-mentioned auxiliary judgment eigenvalues is not limited, and can be selected according to specific needs;

Perform one-hot encoding on the auxiliary judgment eigenvalues:

Dietary intake: Whether the weekly dietary intake is reduced by > 25%: 0 means no, 1 means yes;

Albumin: Whether it is <30g/L: 0 means no, 1 means yes;

BMI score: divided into three grades according to BMI value: BMI<18.5, 24.0>BMI>18.50, BMI>24.0, one-hot encoding;

Types of concomitant diseases: bone fractures, acute complications of chronic diseases, liver cirrhosis, COPD, hemodialysis, diabetes, major abdominal surgery, stroke, severe pneumonia, hematological malignancies, craniocerebral injury, bone marrow transplantation, and concomitant diseases One-hot encoding is performed for various types, 0 means yes, 1 means no;

Age: divided into 10 grades, one grade per 10 years old, one-hot encoding;

Gender: 1 for male, 0 for female;

After the above one-hot encoding process, there are a total of 28 auxiliary judgment features, which are used to assist judgment model training. In the model training process, the nutritional risk score is used as a label, and the above 28 auxiliary judgment features are used to construct a decision tree for training, which is used to predict the nutritional risk score of the patient, and the predicted nutritional risk score and the actual nutritional risk score corresponding to the patient are combined. Compare and continuously optimize the parameters of the auxiliary judgment model until a trained auxiliary judgment model is obtained;

Step S23: Input the auxiliary judgment characteristic value of the patient to be predicted into the trained auxiliary judgment model; output the predicted nutritional risk score, if the nutritional risk score is greater than the threshold value and the hemodynamic stability condition is met, then go to step S3, otherwise enter the observation period.

This embodiment of the present invention does not specifically limit which decision tree model is adopted. Obtain the information record of the patient to be predicted, extract the corresponding auxiliary judgment feature value and input it into the auxiliary judgment model, and the output is the predicted nutritional risk score, and the score value is greater than or equal to 3 points or less than 3 points. Patients with nutritional risk score < 3 entered the observation period; when nutritional risk score ≥ 3, the auxiliary judgment model further judged the patient's hemodynamic stability conditions: mean arterial pressure > 65 mmHg, blood lactate < 4 mmol/L, norepinephrine <0.2 μg·min-1·kg-1, if the patient meets the above-mentioned hemodynamic stability conditions, go to step S3.

In one embodiment, the above step S3: according to the data in the information record, the degree of gastrointestinal injury of the patient is judged, and the nutrition supply method is determined: if the injury is severe, parenteral nutrition is recommended, and if the injury is mild or moderate, enteral nutrition is recommended Internal nutrition, including:

The evaluation indicators were selected according to the definition of acute gastrointestinal injury by the European Critical Care Society in 2012, and the evaluation indicators in the information records were evaluated. The evaluation results were graded according to the degree of gastrointestinal injury of the patients. Mild or moderate injury, enteral nutrition is recommended, grade IV is severe injury, parenteral nutrition is recommended.

The embodiment of the present invention uses the definition of acute gastrointestinal injury (AGI) by the European Society of Critical Care in 2012, and is graded according to the degree of gastrointestinal injury of the patient to be tested. , whether diarrhea. The patients are graded according to the evaluation index, grade I: there is a risk of gastrointestinal dysfunction and failure, grade II: gastrointestinal dysfunction, grade III: gastrointestinal failure, grade IV: gastrointestinal dysfunction with distant organ function obstacle. Grades I to III were defined as mild or moderate injury, and enteral nutrition was recommended, and grade IV was defined as severe injury, and parenteral nutrition was recommended, and the patient's assessment results were stored in the CETAT database.

In one embodiment, the above step S4: input the data in the information record into the auxiliary recommendation model to obtain auxiliary data of the energy metabolism value, and combine the basic information and indicators during hospitalization to determine the final energy metabolism value; Combined with the nutrition supply method, the final recommended supplementary nutrition plan is obtained, including:

Step S41: Perform principal component analysis on the data set, and select the auxiliary recommendation feature value for training the auxiliary recommendation model based on the K-means decision tree, until the trained auxiliary recommendation model is obtained;

Using PCA principal component analysis technology, extract characteristic attributes that can reflect the characteristics of sepsis patients from the data set obtained in step S21, and retain auxiliary recommended characteristic values with a factor variance value greater than 1: heart rate, body temperature, pulse oximetry, contraction blood pressure, diastolic blood pressure, mean arterial pressure, respiratory rate, bicarbonate HCO ₃ , blood glucose.

The training process of the auxiliary recommendation model based on the K-means decision tree in the embodiment of the present invention is:

First, let the data set consisting of auxiliary recommendation feature values be S, use the energy metabolism value as the label, randomly select k samples as the initial mean, and use the K-means mean algorithm to group the data set; among them, the adjustment process of parameter k is: : Let S={X ₁ , X ₂ ,...,X _n }, n be the number of samples in the data set S, select the initial cluster center from S, denote it as c ₁ , calculate the cluster center and S The _shortest distance of data Xi, which is expressed as formula (2):

D(X _i )=min{dis(X _i ,c _j )} (2)

Among them, i=1...n, j=1...k;

Calculate the probability P that each sample becomes the next cluster center, as shown in formula (3), and take the sample with the highest probability as the next cluster center:

Repeat the above steps until k selected cluster centers that meet the requirements. However, the k value may not be the optimal value, so the "elbow method" is used to select the optimal k value. In the embodiment of the present invention, the error sum of squares is used as an index for selecting the optimal k value, as shown in formula (4):

Among them, C _i is the ith cluster, p is the sample in C _i , c _i is the cluster center of cluster C _i , and SSE is the clustering error of all samples, indicating the quality of the clustering effect. According to the patient data of the sample set in the embodiment of the present invention, the optimal k value is 4.

According to formula (5), the evaluation index DBI of the clustering effect is calculated:

表示的是任意的两个簇的距离的平方和，c_i和c_j分别是两个簇的聚类中心；DBI值越大，则表示两个簇之间的距离远。in,

It represents the sum of the squares of the distances of any two clusters, c _i and c _j are the cluster centers of the two clusters respectively; the larger the DBI value, the farther the distance between the two clusters is.

Randomly select samples from the 4 clusters obtained above to form a subset Y, input the auxiliary recommendation model constructed based on the ID3 decision tree algorithm, and perform decision analysis on the clustered data, and select the test attribute based on the speed of information descending. In each decision node, the attribute with the highest information gain that has not yet been selected is selected as the dividing standard of the decision tree, until the decision tree is finally generated.

The total information entropy of the sample calculated in the ID3 decision tree algorithm is shown in formula (6):

Among them, assuming that the sample set is Y, it contains y samples in total, the category attribute is energy metabolism value, and there are m different values, which is b _i (i=1,2,3,...,m), and B _i is b _i The number of samples in , P _i =B _i /y represents the probability that the sample belongs to _bi .

Calculate the information entropy of each auxiliary recommendation feature value, and express it as, as shown in formula (7):

Among them, D is the auxiliary recommendation feature value, and there are k different values, where B _ij is the number of samples with the category D _i in the subset B _j , and the information gain is shown in formula (8):

Gain(D)=I(B ₁ ,B ₂ ,...,B _m )-E(D) (8)

The decision tree is divided according to the eigenvalue with the largest information gain, and so on, and finally the auxiliary recommendation model is constructed.

Input the auxiliary recommendation feature value into the auxiliary recommendation model for training, obtain the auxiliary data X of the predicted energy metabolism value, compare the predicted X with the patient's real energy metabolism value, and continuously optimize the parameters of the auxiliary recommendation model until the training is completed. The auxiliary recommendation model of ;

Step S42: Input the auxiliary recommendation feature value of the patient to be predicted into the trained auxiliary recommendation model, and obtain auxiliary data X of the energy metabolism value;

Input the auxiliary recommendation feature value corresponding to the patient to be tested in step S23 into the trained auxiliary recommendation model to obtain auxiliary data X of the required energy metabolism value of the patient;

Step S43: Combine the resting energy metabolism value Y, the auxiliary data X of the energy metabolism value, and the ICU length of time T to obtain the final energy metabolism value Z, and calculate the corresponding calories of sugar, fat, and amino acids according to the Z value. amount; combined with the enteral nutrition method obtained in step S3, the final recommended supplementary nutrition scheme is obtained.

The embodiment of the present invention defines the final energy metabolism value Z based on the resting energy metabolism value Y, the auxiliary data X of the energy metabolism value, and the ICU admission time T (unit: day), as shown in Table 1:

Table 1 Final energy metabolism value Z

0<T<7 7≤T<10 10<T Y>110%X Z=110%X Z=X Z=X X<Y<110%X Z=Y Z=Y Z=X 90% X<Y<X Z=X Z=X Z=90%X Y<90%X Z=90%X Z=90%X Z=90%X

For example, a patient to be tested obtains that its nutritional risk score is 5 according to step S2 and satisfies the stability of hemodynamics, indicating that the patient can carry out nutritional feeding, then enter step S3 to determine the degree of gastrointestinal damage of the patient, if it is determined that the patient is The patient is severely injured, and parenteral nutrition is recommended. According to step S4, the auxiliary data X=2000kcal of the energy metabolism value required by the patient is obtained, and the resting energy metabolism value Y=2300kcal can be calculated in step S11, and the number of days in ICU is 2 days. According to Table 1, it can be known that the final energy of the patient is provided. Metabolized Z = 110% X = 2000 x 110% = 2200 kcal.

According to the formula: the daily protein requirement P = 6.25 × 24 hours the patient's urine nitrogen amount (g), 1 g of protein produces 4.3 kcal of energy, that is, the sugar and fat calorie content = Z-4.3 × P, and the known 24 hours The patient's urine nitrogen amount=12g, the daily protein requirement of the patient can be calculated P=6.25×12=75g; sugar and fat calorie content=Z-4.3×P=2200-4.3×75=1877.5kcal.

Determine the patient's final recommended supplementary nutrition scheme: recommended parenteral nutrition, protein requirement P=75g, sugar and fat calorie content=1877.5kcal.

Embodiment 2

As shown in FIG. 2 , an embodiment of the present invention provides a nutritional decision-making method for patients with sepsis, comprising the following steps:

Step S1: obtain the patient's basic information and indicators during hospitalization, and construct the patient's information record;

Step S2: Input the data in the information record into the auxiliary judgment model, and evaluate the feeding feasibility of the patient. If nutritious feeding can be performed, transfer to S3, otherwise the patient enters the observation period;

Step S3: Judging the degree of gastrointestinal injury of the patient according to the data in the information record, and determining the nutrition supply method: if the injury is severe, the parenteral nutrition method is recommended, and if the injury is mild or moderate, the enteral nutrition method is recommended;

Step S4: Input the data in the information record into the auxiliary recommendation model, obtain auxiliary data of energy metabolism value, and determine the final energy metabolism value by combining the basic information and indicators during hospitalization; combine the final energy metabolism value with the nutrition supply method to obtain the final energy metabolism value. recommended supplementary nutrition programs;

S5: Tolerance assessment of patients is carried out according to the enteral nutrition tolerance score table, and the final recommended supplementary nutrition scheme is adjusted according to the assessment results.

The specific implementation details of the above steps S1 to S4 are the same as the steps S1 to S4 in the embodiment;

Step S5: Evaluate the patient's tolerance according to the enteral nutrition tolerance score table, and adjust the final recommended supplementary nutrition scheme according to the evaluation results, including:

The patients were evaluated according to the enteral nutrition tolerance scale, and the evaluation indicators were: abdominal pain NRS grade, abdominal distension grade, intra-abdominal pressure grade, diarrhea grade, bowel sounds, aspiration, vomiting.

If the patient's nutrition supply method is enteral nutrition supply, according to the score results of the enteral nutrition tolerance scale, it is divided into: 0-6 points: enteral nutrition maintains the original speed, 7-12 points: enteral nutrition slows down the speed, ≥13 points: Stop enteral nutrition.

If the patient's nutritional supply method is parenteral nutrition supply, according to the degree of gastrointestinal injury of the patient, the evaluation result is grade I to III, it is recommended that doctors combine enteral nutrition and parenteral nutrition supply, and then adjust the nutritional supply method every day. The adjustment method is as follows: According to the scoring results of the enteral nutrition tolerance scale, it is divided into: 0-6 points: the amount of parenteral nutrition is reduced by 10%, provided by enteral nutrition; 7-12 points: no additional enteral nutrition is provided, ≥ 13 points : Stop enteral nutrition. Continue to adjust the nutrition supply method until the patient is completely on enteral nutrition.

The invention provides a nutrition decision-making method for patients with sepsis, which can intelligently provide patients with a personalized auxiliary nutrition decision-making scheme, realizes the process, standardization and informationization of the nutrition decision-making process, and improves the clinical nutrition treatment in the whole hospital. efficiency.

Embodiment 3

As shown in FIG. 3 , an embodiment of the present invention provides a nutrition decision-making system for patients with sepsis, including the following modules:

The basic information acquisition module 61 is used to acquire the basic information of the patient and the indicators during hospitalization, and construct the patient information record;

The patient feeding feasibility assessment module 62 is used for inputting the data in the information record into the auxiliary judgment model to assess the feeding feasibility of the patient; if nutritional feeding can be performed, it is transferred to the patient gastrointestinal function assessment module, otherwise the patient enters the observation period;

The patient's gastrointestinal function evaluation module 63 is used to judge the degree of gastrointestinal injury of the patient according to the data in the information record, and determine the nutrition supply method: if the injury is severe, parenteral nutrition is recommended; if the injury is mild or moderate, enteral nutrition is recommended nutrition;

The patient auxiliary nutrition recommendation scheme module 64 is used to input the data in the information record into the auxiliary recommendation model, obtain auxiliary data of energy metabolism value, and determine the final energy metabolism value in combination with the basic information and indicators during hospitalization; combine the final energy metabolism value with Enteral nutrition, get the final recommended supplementary nutrition program.

Embodiment 4

As shown in FIG. 4 , an embodiment of the present invention provides a nutrition decision-making system for patients with sepsis, including the following modules:

The basic information acquisition module 61 is used to acquire the basic information of the patient and the indicators during hospitalization, and construct the patient information record;

The patient feeding feasibility assessment module 62 is used for inputting the data in the information record into the auxiliary judgment model to assess the feeding feasibility of the patient; if nutritional feeding can be performed, it is transferred to the patient gastrointestinal function assessment module, otherwise the patient enters the observation period;

The patient's gastrointestinal function evaluation module 63 is used to judge the degree of gastrointestinal injury of the patient according to the data in the information record, and determine the nutrition supply method: if the injury is severe, parenteral nutrition is recommended; if the injury is mild or moderate, enteral nutrition is recommended nutrition;

The patient auxiliary nutrition recommendation scheme module 64 is used to input the data in the information record into the auxiliary recommendation model, obtain auxiliary data of energy metabolism value, and determine the final energy metabolism value in combination with the basic information and indicators during hospitalization; combine the final energy metabolism value with Enteral nutrition, get the final recommended supplementary nutrition plan;

Nutritional supply mode adjustment module 65 : evaluate the patient's tolerance according to the enteral nutrition tolerance score table, and adjust the final recommended supplementary nutrition program according to the evaluation result.

The above embodiments are provided for the purpose of describing the present invention only, and are not intended to limit the scope of the present invention. The scope of the invention is defined by the appended claims. Various equivalent replacements and modifications made without departing from the spirit and principle of the present invention should be included within the scope of the present invention.

Claims (9)

1. A method of nutritional decision making for sepsis patients, comprising the steps of:

step S1: acquiring basic information and inpatient period indexes of a patient, and constructing an information record of the patient;

step S2: inputting the data in the information record into an auxiliary judgment model, carrying out feeding feasibility evaluation on the patient, and switching to S3 if the patient can carry out nutrition feeding, otherwise, switching to an observation period;

step S3: and judging the gastrointestinal injury degree of the patient according to the data in the information record, and determining a nutrition supply mode: if severe injury occurs, recommending an parenteral nutrition mode, and if mild or moderate injury occurs, recommending an enteral nutrition mode;

step S4: inputting data in the information record into an auxiliary recommendation model to obtain auxiliary data of an energy metabolism value, and determining a final energy metabolism value by combining the basic information and the hospitalization period index; and combining the final energy metabolism value with the nutrition supply mode to obtain a final auxiliary nutrition recommendation scheme.

2. The nutritional decision method for sepsis patients according to claim 1, characterized in that the step S1: acquiring basic information and inpatient period indexes of a patient, and constructing an information record of the patient, wherein the information record specifically comprises the following steps:

step S11: acquiring basic information of a patient, comprising: direct information of the patient measured at the time of admission of the patient and energy metabolism information of the patient measured by a professional device; wherein,

the direct information includes: gender, level of consciousness, age, height, weight, body temperature, pulse, respiration, systolic pressure, diastolic pressure, blood oxygen saturation and time of admission;

the energy metabolism information includes: a resting energy metabolism value, a respiratory quotient and a corresponding measurement time;

step S12: obtaining patient hospitalization indices including: laboratory examination data, vital sign data, ICU entry duration, and nutrient supply duration.

3. The nutritional decision method for sepsis patients according to claim 2, wherein the calculation of the resting energy metabolism value in step S11 specifically comprises:

step S111: obtaining oxygen uptake VO of a patient in a specific time ₂ And carbon dioxide exhalation rate VCO ₂ ；

Step S112: calculating the resting energy metabolism value Y using equation (1):

Y(kcal/day)＝1.44×([VO ₂ (ml/min)×3.94]+[VCO ₂ (ml/min)×1.11]) (1)。

4. the nutritional decision method for sepsis patients according to claim 1, characterized in that the step S2: inputting the data in the information record into an auxiliary judgment model, and carrying out feeding feasibility evaluation on the patient to obtain a nutritional risk score of the patient; if the nutrition feeding is available, the method proceeds to step S3, otherwise, the patient enters an observation period, which specifically includes:

step S21: data cleaning is carried out on the data in the information record to obtain a cleaned data set;

step S22: selecting an auxiliary judgment characteristic value from the data set according to an NRS-2002 nutrition risk screening table, and using the auxiliary judgment characteristic value to train an auxiliary judgment model based on a decision tree until a trained auxiliary judgment model is obtained;

step S23: inputting the auxiliary judgment characteristic value of the patient to be predicted into the trained auxiliary judgment model, outputting the predicted nutritional risk score, if the nutritional risk score is larger than a threshold value and meets the hemodynamic stability condition, turning to the step S3, otherwise, entering an observation period.

5. The nutritional decision method for sepsis patients according to claim 1, characterized in that the step S3: and judging the gastrointestinal injury degree of the patient according to the data in the information record, and determining a nutrition supply mode: if the damage is severe, a parenteral nutrition mode is recommended, and if the damage is mild or moderate, an enteral nutrition mode is recommended, which specifically comprises the following steps:

the assessment indexes are selected according to the definition of acute gastrointestinal injury of the European Critical society of 2012, the assessment indexes in the information record are assessed, the assessment results are graded according to the gastrointestinal injury degree of the patient, the grade I-III is mild or moderate injury, an enteral nutrition mode is recommended, the grade IV is severe injury, and an parenteral nutrition mode is recommended.

6. The nutritional decision method for sepsis patients according to claim 1, characterized in that the step S4: inputting data in the information record into an auxiliary recommendation model to obtain auxiliary data of an energy metabolism value, and determining a final energy metabolism value by combining the basic information and the hospitalization period index; and combining the final energy metabolism value with the intestinal nutrition mode to obtain a final auxiliary nutrition recommendation scheme, which specifically comprises the following steps:

step S41: selecting an auxiliary recommendation characteristic value in the data set for training an auxiliary recommendation model based on a K-means decision tree until a trained auxiliary recommendation model is obtained;

step S42: inputting the auxiliary recommendation characteristic value of the patient to be predicted into the trained auxiliary recommendation model to obtain auxiliary data X of the energy metabolism value;

step S43: combining the rest energy metabolism value Y with the auxiliary data X of the energy metabolism value and the ICU entering time T to obtain a final energy metabolism value Z, and calculating the calorie content of corresponding sugar, fat and amino acid according to the Z value; and combining the nutrition supply modes obtained in the step S3 to obtain a final auxiliary nutrition recommendation scheme.

7. The nutritional decision method for sepsis patients according to claim 1, characterized by further comprising the steps of:

s5: the patient is assessed for tolerance according to an enteral nutritional tolerance scoring table and the final recommended supplemental nutrition regimen is adjusted according to the assessment.

8. A nutritional decision system for sepsis patients, comprising the following modules:

the basic information acquisition module is used for acquiring basic information and inpatient period indexes of a patient and constructing a patient information record;

the patient feeding feasibility evaluation module is used for inputting the data in the information record into an auxiliary judgment model and evaluating the feeding feasibility of the patient; if the nutrition feeding can be carried out, the gastrointestinal function evaluation module is switched to, otherwise, the patient enters an observation period;

the gastrointestinal function evaluation module of the patient is used for judging the gastrointestinal damage degree of the patient according to the data in the information record and determining the nutrition supply mode: if severe injury occurs, recommending an parenteral nutrition mode, and if mild or moderate injury occurs, recommending an enteral nutrition mode;

the patient auxiliary nutrition recommendation scheme module is used for inputting data in the information record into an auxiliary recommendation model to obtain auxiliary data of an energy metabolism value, and determining a final energy metabolism value by combining the basic information and the hospitalization period index; and combining the final energy metabolism value with the intestinal nutrition mode to obtain a final auxiliary nutrition recommendation scheme.

9. The nutritional decision system for sepsis patients according to claim 8, further comprising a module:

a nutrition supply mode adjustment module: the patient is subjected to tolerance assessment according to an enteral nutrition tolerance score table, and the final supplemental nutrition recommendation is adjusted according to the assessment results.

Images