Infectious fever antibacterial scheme selection method and system based on multi-dimensional data

Abstract

The invention provides an infectious fever antibacterial scheme selection method and system based on multi-dimensional data, and belongs to the technical field of infectious disease diagnosis and treatment. The method comprises the following steps: step (1), identifying a fever type based on symptom and sign data and inspection data, and positioning a pathogen type, an infection site and severity; (2) generating candidate antibiotic schemes based on the antibiotic knowledge base and a guide rule engine; (3) adjusting the dosage according to the individual parameters of the patient, and generating a first scheme and an alternative scheme; through multi-dimensional data integration, infection types are rapidly identified, antibiotics recommended by a guide are matched, and an accurate and safe antibacterial scheme (including a first scheme and an alternative scheme) is generated according to individual conditions of patients.

Description

Technical Field

[0001] This invention relates to the field of infectious disease diagnosis and treatment technology, and in particular to a method and system for selecting antibacterial regimens for infectious fever based on multi-dimensional data integration. Background Technology

[0002] Infectious fever is a medical term that generally refers to a fever caused by pathogen infection. This type of fever is a defensive response of the body to infection and is common in various infectious diseases.

[0003] The selection of existing antibacterial regimens for infectious fever relies on physician experience, which has problems such as non-standard implementation of guidelines, insufficient consideration of individual differences (such as liver and kidney function, age), and high risk of adverse reactions.

[0004] Traditional guidelines are cumbersome to read and make it difficult to integrate patient test data (such as complete blood count, pathogen culture, and drug sensitivity results) with antibiotic characteristics (indications, resistance rates, and dosage adjustment rules) in real time.

[0005] Current technologies lack a systematic closed-loop process of "infection identification - pathogen matching - guideline alignment - individualized adjustment". A dynamically updated antibiotic knowledge base has not been established, making it impossible to automatically generate multiple treatment options for clinical decision-making. Summary of the Invention

[0006] In view of this, in order to overcome the above-mentioned shortcomings in the existing technology, on the one hand, the present invention provides a method for selecting antibacterial regimens for infectious fever based on multi-dimensional data integration. Through multi-dimensional data integration, it can quickly identify the infection type, match the antibiotics recommended by the guidelines, and generate a precise and safe antibacterial regimen (including a first regimen and alternative regimens) according to the individual patient's condition.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for selecting antibacterial regimens for infectious febrile diseases based on multi-dimensional data integration includes the following steps:

[0009] Step (1) Identify the type of fever based on symptom and sign data and laboratory test data, and locate the pathogen type, infection site and severity;

[0010] Step (2) Generate candidate antibiotic regimens based on the antibiotic knowledge base and guideline rule engine;

[0011] Step (3) Adjust the dosage according to the patient's individual parameters to generate the first plan and alternative plans.

[0012] Preferably, in step (1), the symptom and sign data include fever duration, fever pattern, accompanying symptoms, and signs of infection focus.

[0013] Preferably, in step (1), the test data includes basic test data, etiological test data, and imaging data.

[0014] Preferably, in step (2), the antibiotic knowledge base contains structured data on indications, pharmacokinetics, adverse reactions, and drug resistance.

[0015] Preferably, in step (2), the guide rule engine transforms authoritative guidelines into structured rules and matches candidate antibiotics according to the priority of "infection type → pathogen → recommendation level".

[0016] Preferably, in step (3), the individual parameters are age, liver and kidney function, and special populations.

[0017] On the other hand, the present invention also provides a system for performing the above-described method for selecting antibacterial regimens for infectious fever based on multi-dimensional data integration, comprising:

[0018] The infection identification module is used to identify the type of fever based on symptom and sign data and laboratory test data.

[0019] The pathogen analysis module identifies the type of fever, infection site, and severity based on the fever type identified by the infection recognition module.

[0020] The knowledge base module is used to store multi-dimensional data on antibiotics;

[0021] The rules engine module, based on the antibiotic knowledge base and guide rules engine, generates candidate antibiotic regimens;

[0022] The individualized treatment plan generation module is used to adjust the dosage based on the patient's individual parameters and generate a first treatment plan and alternative treatment plans.

[0023] The present invention has the following advantages over the prior art:

[0024] Multi-dimensional data integration closed loop: For the first time, symptom and sign data, laboratory test data, etiological tests, guidelines and rules, and individual parameters are integrated to achieve a closed loop of "precise identification - intelligent matching - safe adjustment". This not only covers the initial observation of patient symptoms and the formulation of the final treatment plan, but also includes the comprehensive analysis of various medical test results and in-depth research on pathogens. By integrating this key information, the medical system can more accurately identify the patient's health status, intelligently match the most suitable treatment plan, and make safe adjustments when necessary to ensure the effectiveness and safety of treatment.

[0025] Dynamic knowledge base and rule engine: Supports guideline updates and drug resistance data synchronization to ensure that protocols meet the latest clinical standards.

[0026] Multiple treatment options mechanism: While adhering to guideline recommendations, this mechanism simultaneously provides alternative options based on factors such as drug resistance, drug accessibility, and adverse reactions, enhancing the flexibility and adaptability of clinical decision-making. This mechanism aims to provide physicians and patients with more comprehensive and personalized treatment choices, ensuring that the most suitable decisions can be made quickly and accurately when faced with different conditions and treatment needs.

[0027] Personalized dosing algorithm: Built-in age and liver and kidney function correction formulas automatically generate dosing regimens that conform to pharmacokinetic principles, reducing the risk of dose-related adverse reactions. Attached Figure Description

[0028] Figure 1 This is a flowchart of the process of the present invention. Detailed Implementation

[0029] like Figure 1 As shown, this invention provides a method for selecting antibacterial regimens for infectious fever based on multi-dimensional data integration, comprising the following steps:

[0030] Step (1) Identify the fever type and locate the pathogen category, infection site, and severity based on symptom and sign data and laboratory test data. Symptom and sign data include fever duration, fever pattern (e.g., continuous fever / remittent fever), accompanying symptoms (e.g., cough, dysuria, rash), and signs of infection focus (e.g., lung rales, local redness and swelling). Laboratory test data includes basic examination data, etiological examination data, and imaging data. Basic examination data includes complete blood count (e.g., WBC, neutrophil percentage, CRP, PCT) and blood biochemistry (liver and kidney function, electrolytes). Etiological examination data includes blood / sputum / urine culture, PCR nucleic acid detection, and serum antibodies (e.g., G test, GM test). Imaging data includes chest X-ray, CT scan, and ultrasound (to locate the infection site).

[0031] Among these methods, fever type identification can preferably be based on a decision tree algorithm to distinguish between infectious fever and non-infectious fever (such as tumors or autoimmune diseases), achieving an accuracy rate of over 95%. The decision tree algorithm used in this invention is a conventional decision tree algorithm in the field and can be selected according to actual needs. For example:

[0032] In this invention, the decision tree identification method for fever type is specifically implemented as follows:

[0033] Extract the following key features from electronic medical records:

[0034] Basic characteristics: age, sex, duration of fever, fever pattern (continuous fever / remittent fever / intermittent fever, etc.);

[0035] Laboratory indicators: white blood cell count, neutrophil percentage, CRP, PCT, ESR, autoantibody profile.

[0036] Imaging features: presence of exudative shadows on lung CT scan, and metabolic status on PET-CT scan;

[0037] Clinical symptoms: Whether accompanied by skin rash, joint pain, swollen lymph nodes, etc.;

[0038] Decision tree construction

[0039] The decision tree is generated using the C4.5 algorithm. Example of core classification rules:

[0040] IFPCT > 2 ng / mL AND fever duration < 7 days THEEN infectious fever

[0041] ELSEIF autoantibody profile positive AND ESR > 50 mm / h THEN autoimmune disease

[0042] ELSEIF PET-CT showed hypermetabolic nodules AND elevated tumor markers THEN tumor-related fever.

[0043] ELSEIF: Fever pattern is continuous fever AND WBC < 4 × 10^9 / L THEN: Viral infection

[0044] Model Validation

[0045] Validation based on a dataset of 3000 febrile patients:

[0046] Training set: 2000 cases (1200 infectious diseases, 500 autoimmune diseases, and 300 neoplastic diseases);

[0047] Test set: 1000 cases;

[0048] Using 10-fold cross-validation, the final accuracy reached 95.2%, specificity 96.7%, and sensitivity 94.3%.

[0049] This invention, based on etiological evidence, identifies the type of pathogen, the site of infection, and the severity. The pathogen types include bacteria (Gram-positive / Gram-negative), viruses, fungi, and atypical pathogens (such as mycoplasma and chlamydia). Sites of infection include the respiratory tract, urinary tract, abdominal cavity, and bloodstream infections. Severity is categorized as mild or severe (such as sepsis and septic shock).

[0050] This step, based on multi-dimensional data, identifies and locates the type of heat, offering the following advantages:

[0051] I. Accurately identify fever type and pathogen

[0052] By comprehensively analyzing symptom and physical examination data, as well as laboratory test results, the system can accurately identify fever types and differentiate between infectious and non-infectious fevers with an accuracy rate exceeding 95%. Furthermore, based on etiological evidence, it can precisely pinpoint the pathogen category, including bacteria, viruses, fungi, and atypical pathogens, providing a scientific basis for treatment.

[0053] II. Comprehensively locate the site and severity of infection.

[0054] In-depth analysis of symptoms, signs, and imaging data allows for accurate identification of the infection site, such as the respiratory tract, urinary tract, abdominal cavity, or bloodstream infection. Furthermore, based on baseline and etiological examination data, the severity of the infection can be assessed, ranging from mild to severe (e.g., sepsis, septic shock), providing strong support for the development of treatment plans.

[0055] III. Improving Diagnostic Efficiency and Accuracy

[0056] This solution combines artificial intelligence technology with medical knowledge to achieve rapid and accurate diagnosis of febrile diseases. It avoids the missed diagnoses and misdiagnoses caused by insufficient doctor experience or misdiagnosis in traditional diagnostic methods, improving diagnostic efficiency and accuracy and providing strong support for timely treatment of patients.

[0057] Step (2) generates candidate antibiotic regimens based on the antibiotic knowledge base and guideline rule engine. In this step (2), the antibiotic knowledge base contains structured data on indications, pharmacokinetics, adverse reactions, and drug resistance.

[0058] The antibiotic knowledge base includes:

[0059] Indications cover authoritative guidelines such as the "Guidelines for Clinical Application of Antimicrobial Drugs" and "Fever Diseases," and are labeled according to pathogen-antibiotic matching relationships (e.g., Streptococcus pneumoniae → penicillins) and infection sites (e.g., urinary tract infections → quinolones).

[0060] Pharmacokinetics, effects of liver and kidney function (e.g., β-lactams excreted by the kidneys require dose adjustment based on creatinine clearance), and age correction (dosage formulas for newborns / elderly).

[0061] The neonatal dose correction formula is as follows:

[0062] Fried's formula (for babies aged 1-24 months)

[0063]

[0064] Yound Formula (for children aged 12-18)

[0065]

[0066] Clark's Formula (calculated as a percentage of body weight)

[0067]

[0068] The dosage correction formula for older adults is as follows:

[0069] Cockcroft-Gault Formula (Kidney Function Correction)

[0070]

[0071] CrCl = Male value × 0.85 (Female)

[0072] Jelliffe Formula (Simplified Version for Kidney Function Correction)

[0073]

[0074] CrCl = Male value × 0.9 (Female)

[0075] Body surface area correction method (applicable to all age groups)

[0076]

[0077] Body surface area is calculated using the Mosteller formula:

[0078]

[0079] Adverse reactions and contraindications, history of allergies (e.g., β-lactams are contraindicated in patients with penicillin allergy), contraindications in special populations (tetracyclines are contraindicated in pregnant women), drug interactions (e.g., rifampin induces liver enzymes that affect the metabolism of other drugs).

[0080] Drug resistance data, regional drug resistance monitoring data (e.g., automatic alert when the resistance rate of Escherichia coli to levofloxacin in a certain region is >50%).

[0081] Examples of structured data are as follows:

[0082] {"Antibiotic Name":"Amoxicillin Clavulanate Potassium","Indications":["Streptococcus pneumoniae infection","β-lactamase-producing Haemophilus influenzae infection"],"Recommendation Level":"First-line (IDSA Guidelines 2023)","Renal Function Adjustment":"Dose halved when CrCl < 30 ml / min","Contraindications":["History of penicillin allergy","Severe liver impairment"],"Common Adverse Reactions":["Gastrointestinal reactions","Rash"]}

[0083] In step (2), based on the structured data described above, the guideline rule engine transforms authoritative guidelines into structured rules and matches candidate antibiotics according to the priority of "infection type → pathogen → recommendation level". Specific operations can be performed as follows:

[0084] Digitalization of the Guidelines:

[0085] Transform authoritative guidelines (such as IDSA and China's "Guidelines for Clinical Application of Antimicrobial Drugs") into structured rules, for example:

[0086] Rule 1: For community-acquired pneumonia (CAP) in adults without underlying diseases, outpatient treatment → macrolides (such as azithromycin) or doxycycline (first-line regimen) are recommended; if the local Streptococcus pneumoniae resistance rate to macrolides is >25%, then β-lactams (alternative regimen) should be used.

[0087] Rule 2: Patients with septic shock → Follow “early goal-oriented therapy” and administer broad-spectrum antibiotics (such as piperacillin-tazobactam + vancomycin) within 1 hour.

[0088] The preferred matching logic is as follows:

[0089] Sorting by "infection type → pathogen → guideline recommendation level", preliminary candidate regimens (including 1 to 3 antibiotics) are generated;

[0090] Automatically exclude drugs that conflict with the patient's contraindications (e.g., levofloxacin is excluded if the patient has a history of quinolone allergy).

[0091] The advantages of this step are mainly reflected in the following aspects:

[0092] I. Highly Efficient and Precise Matching: By transforming authoritative guidelines into structured rules, the guideline rule engine can quickly and accurately match candidate antibiotic regimens based on the patient's infection type, pathogen, and guideline recommendation level. This significantly reduces the time doctors spend consulting guidelines, analyzing cases, and developing treatment plans, thus improving work efficiency.

[0093] II. Personalized Treatment Plans: The structured rule engine can automatically exclude medications that conflict with a patient's contraindications based on their specific circumstances, such as age, underlying diseases, and allergy history, thus tailoring a personalized treatment plan for each patient. This helps improve the targeting and effectiveness of treatment and reduces the occurrence of adverse reactions.

[0094] Third, improving the quality of medical care: A structured rule engine based on authoritative guidelines ensures the scientific rigor and authority of treatment plans. Doctors can rely more heavily on these validated rules when developing treatment plans, thereby improving the quality of medical care and reducing medical risks.

[0095] Fourth, promoting the rational use of medical resources: Through automated matching and personalized treatment plans, the guideline rule engine helps reduce unnecessary antibiotic use and avoid drug abuse and resistance. This helps promote the rational use of medical resources and reduce medical costs.

[0096] Step (3) adjusts the dosage according to the patient's individual parameters to generate a first treatment plan and an alternative treatment plan. In step (3), the individual parameters are age, liver and kidney function, and special populations.

[0097] Among them, age: the dosage for newborns is calculated according to their age in days (e.g., ceftriaxone: <7 days, 50mg / kg q12h; >7 days, 50mg / kg qd); for the elderly, the decline in liver and kidney function should be considered (e.g., adjusting the dosage of chloramphenicol).

[0098] Liver and kidney function: Liver dysfunction: The dosage of drugs metabolized by the liver (such as rifampin) needs to be reduced, and hepatotoxic drugs (such as isoniazid) should be used with caution;

[0099] Renal insufficiency: Adjust the dosage of drugs excreted by the kidneys according to creatinine clearance (CrCl) (e.g., if CrCl < 60 ml / min, adjust meropenem from 1 g q8h to 1 g q12h).

[0100] Special populations: pregnant women (choose FDA pregnancy category B drugs, such as penicillins), and immunosuppressed patients (opportunistic infections must be covered, such as caspofungin, which is contraindicated in patients with granulocytopenia and fever).

[0101] The solution output is as follows:

[0102] First option: The option based on the highest level of guideline recommendation, the lowest risk of drug resistance, and the highest degree of matching with the patient's individual parameters;

[0103] Alternative options (multiple options are possible), as follows:

[0104] Consider the possibility of drug resistance (e.g., if the local pathogen has a resistance rate >30% to the first treatment, provide a second-line drug);

[0105] Consider drug accessibility (e.g., if a certain drug is unavailable in a primary care hospital, replace it with a recommended drug of the same level);

[0106] Considering the risk of adverse reactions (e.g., if the patient has underlying gastrointestinal disease, when the primary regimen is macrolides, alternative regimens may include gastric mucosal protectants or be replaced with β-lactams).

[0107] On the other hand, the present invention also provides a system for performing the above-described method for selecting antibacterial regimens for infectious fever based on multi-dimensional data integration, comprising:

[0108] The infection identification module is used to identify the type of fever based on symptom and sign data and laboratory test data.

[0109] The pathogen analysis module identifies the type of fever, infection site, and severity based on the fever type identified by the infection recognition module.

[0110] The knowledge base module is used to store multi-dimensional data on antibiotics;

[0111] The rules engine module, based on the antibiotic knowledge base and guide rules engine, generates candidate antibiotic regimens;

[0112] The individualized treatment plan generation module is used to adjust the dosage based on the patient's individual parameters and generate a first treatment plan and alternative treatment plans.

[0113] The system provided by this invention is suitable for performing the above-described method, and its modules are used to perform the steps of the above-described method.

[0114] The technical solution of the present invention will be clearly and thoroughly described below with reference to specific embodiments.

[0115] Example 1

[0116] Taking adult patients with community-acquired pneumonia (CAP) as an example, the specific procedures are as follows:

[0117] 1. Input data:

[0118] Symptoms: Fever for 3 days, cough with yellow sputum, and moist rales in the right lower lung;

[0119] Test: WBC 12×10 9 / L, CRP 80mg / L, sputum culture indicated Streptococcus pneumoniae (sensitive to penicillin);

[0120] Individual parameters: Male, 55 years old, CrCl 80ml / min, no history of drug allergy.

[0121] 2. System Processing:

[0122] The infection identification module determined it to be "infectious fever (bacterial pneumonia)";

[0123] The pathogen analysis module identified the pathogen as "Gram-positive bacteria (Streptococcus pneumoniae)" and the site of infection as "lower respiratory tract".

[0124] Knowledge base module + rule engine module matching guide: IDSA 2023 CAP guidelines recommend "penicillins (first-line) or macrolides (alternative)";

[0125] The individualized treatment plan generation module was adjusted as follows: The patient has normal renal function and no history of allergies. The first treatment plan is "Amoxicillin 1g q8hpo", and the alternative treatment plan is "Azithromycin 500mg qdpo (for patients with penicillin intolerance)".

[0126] 3. Output results:

[0127] First regimen: Amoxicillin 1g orally every 8 hours, for a course of 7 days;

[0128] Alternative treatment: Azithromycin 500mg orally once daily for 5 days (to be used if the patient experiences a penicillin allergic reaction).

[0129] The above description is merely a preferred embodiment of the present invention. However, the scope of protection of the present invention is not limited thereto; any equivalent substitutions or modifications made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and its improved concept, should be covered within the scope of protection of the present invention.

Claims

1. An infectious fever antibacterial regimen selection method based on multi-dimensional data integration, characterized in that, The method comprises the following steps: Step (1) identifies fever types based on symptom and sign data and test and examination data, and locates pathogen categories, infection sites and severity; Step (2) generates candidate antibiotic schemes based on an antibiotic knowledge base and a guideline rule engine; Step (3) adjusts doses according to patient individual parameters to generate first schemes and alternative schemes.

2. The method according to claim 1, wherein the method is characterized by, In step (1), the symptom and sign data are fever duration, fever type, accompanying symptoms and infection focus signs.

3. The method of claim 1, wherein the method is based on multi-dimensional data integration. In step (1), the test and examination data are basic examination data, pathogenic examination data and imaging data.

4. The method of claim 1, wherein the method is based on multi-dimensional data integration. In step (2), the antibiotic knowledge base comprises structured data of indications, pharmacokinetics, adverse reactions and drug resistance data.

5. The method of claim 1, wherein the method is based on multi-dimensional data integration. In step (2), the guideline rule engine converts authoritative guidelines into structured rules, and matches candidate antibiotics according to the priority of "infection type→pathogen→recommended level".

6. The method of claim 1-5, wherein the method is characterized in that, In step (3), the individual parameters are age, liver and kidney function and special groups.

7. A system for performing the multi-dimensional data integration based infectious fever antibacterial regimen selection method of any one of claims 1-6, characterized in that, It comprises: An infection identification module for identifying fever types based on symptom and sign data and test and examination data; A pathogen analysis module for locating pathogen categories, infection sites and severity based on the fever types identified by the infection identification module; A knowledge base module for storing multi-dimensional antibiotic data; A rule engine module for generating candidate antibiotic schemes based on an antibiotic knowledge base and a guideline rule engine; An individualized scheme generation module for adjusting doses according to patient individual parameters to generate first schemes and alternative schemes.

Images