CN116313092B - Postoperative delirium prediction system and method for high-throughput target metabonomics analysis - Google Patents

Postoperative delirium prediction system and method for high-throughput target metabonomics analysis Download PDF

Info

Publication number
CN116313092B
CN116313092B CN202310270601.3A CN202310270601A CN116313092B CN 116313092 B CN116313092 B CN 116313092B CN 202310270601 A CN202310270601 A CN 202310270601A CN 116313092 B CN116313092 B CN 116313092B
Authority
CN
China
Prior art keywords
delirium
data
training
serum
metabolites
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202310270601.3A
Other languages
Chinese (zh)
Other versions
CN116313092A (en
Inventor
郭勇
徐卿
周全红
李想
江伟
贾伟
张宜男
赵爱华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Sixth Peoples Hospital
Original Assignee
Shanghai Sixth Peoples Hospital
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai Sixth Peoples Hospital filed Critical Shanghai Sixth Peoples Hospital
Priority to CN202310270601.3A priority Critical patent/CN116313092B/en
Publication of CN116313092A publication Critical patent/CN116313092A/en
Application granted granted Critical
Publication of CN116313092B publication Critical patent/CN116313092B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/30—Prediction of properties of chemical compounds, compositions or mixtures
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/80—Data visualisation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/70—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to mental therapies, e.g. psychological therapy or autogenous training
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • General Health & Medical Sciences (AREA)
  • Epidemiology (AREA)
  • Data Mining & Analysis (AREA)
  • Primary Health Care (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Pathology (AREA)
  • Psychiatry (AREA)
  • Social Psychology (AREA)
  • Psychology (AREA)
  • Hospice & Palliative Care (AREA)
  • Developmental Disabilities (AREA)
  • Child & Adolescent Psychology (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Other Investigation Or Analysis Of Materials By Electrical Means (AREA)

Abstract

The invention provides a postoperative delirium prediction system and method based on high-flux target metabonomics analysis, and belongs to the field of accurate medical treatment. According to the invention, through high-flux target metabonomics analysis and combining with artificial intelligence and machine deep learning modeling, a high-efficiency, high-precision and high-sensitivity prediction model for delirium after old patients is established, the early warning purpose can be achieved before the operation for high-risk patients, delirium after the operation can be identified early, a new method for preventing and treating delirium after the operation can be provided, and accurate medical practice of delirium after the old operations based on high-flux target metabonomics and artificial intelligence analysis technology is realized.

Description

Postoperative delirium prediction system and method for high-throughput target metabonomics analysis
Technical Field
The invention belongs to the field of accurate medical treatment, and particularly relates to a postoperative delirium prediction system and method based on high-flux target metabonomics analysis.
Background
Delirium is an acute or subacute, fluctuating neuropsychiatric syndrome characterized by abnormal levels of consciousness, attention deficit associated with cognitive impairment, and symptoms including thinking, behavior, emotion, and perception. Post-operative delirium (Post-Operative Delirium, POD) is a common postoperative complication for elderly patients, and studies have shown that the incidence of Post-operative delirium in elderly patients is 10% -50% or even higher, usually occurs within 5 days after surgery, and the core features become acute attacks, condition fluctuations, attention damage, confusion, disturbance of consciousness, and the like. The harm of the POD is not only that the mechanical ventilation time is prolonged, the ICU residence time and the hospitalization time are increased, the higher tracheotomy probability and the medical cost are consumed, but also that the nursing burden is increased, the hospitalization time is prolonged, the death rate of the patient in the hospitalization period is increased, and long-term cognitive function damage is remained in some senile patients, and even the senile patients are converted into permanent dementia.
Assessment and prediction of delirium after elderly surgery is one of the focus of medical community, and existing assessment tools include CAM and CAM-ICU, ICDSC, DRS-98, CTD, 4AT, etc., which have advantages in screening and diagnosis of delirium, respectively, with respective applicable subjects. However, most of the current tools are based on observation and simple detection, and the accuracy of evaluation and prediction is not high enough to meet clinical needs. Currently, there is an urgent need for a high-precision, high-reliability method for predicting delirium after senile surgery.
Disclosure of Invention
The invention aims to solve the problems in the prior art and provide a high-precision and high-sensitivity postoperative delirium prediction system and method.
The high-flux target metabolomics integrates the technical advantages of non-target metabolomics and target metabolomics, realizes high-flux and high-sensitivity target metabolite detection, and provides an efficient method for qualitatively and quantitatively detecting mass low-abundance metabolites. According to the invention, through high-flux target metabonomics analysis and collection of relevant clinical data, artificial intelligence and machine deep learning modeling are combined, a prediction model for delirium after operation of an elderly patient is provided, early warning is carried out on a high-risk patient before operation, early recognition is carried out, intervention measures are taken, and postoperative recovery is promoted.
Postoperative delirium prediction systems based on high-throughput target metabonomics analysis are achieved by:
a data acquisition unit: collecting blood of a patient on the day of operation and in the morning after operation, freezing serum, performing liquid chromatography analysis and high-resolution mass spectrometry on the serum, and obtaining original data of metabolites by adopting a liquid chromatography tandem mass spectrometry method;
a data analysis unit: the original data is imported into MS-dial software for preprocessing, including peak extraction, peak positioning, noise removal, deconvolution, digital peak calibration, filtering and normalization, and then a mass spectrum database is used for identifying and screening metabolites;
model training unit: the machine learning method is used for deep learning of the preprocessed data, the serum sample on the current day of operation and the serum sample after operation are used as training sets, random circulation is carried out for multiple times, diversified fitting is achieved, and the following model is adopted to optimize a deep learning algorithm:
L t (W) is an objective function, x i And y i Respectively representing the expression level of the metabolite in different dimensions,representing the minimum training data amount, gamma being the forgetting factor, l being the loss function, b being the training sample number, W being the weight, r being the convex normalization function;
further, the weight W in equation (1) is updated using the following algorithm:
Step1：M 0 ＝0，R 0 ＝0
Step2：
Step3：
Step4：
Step5：
Step6：
Step7：Return W
where M is a first momentum predicted value, R is a second momentum predicted value,is a first momentum correction value,/>Is a second momentum correction value, W is a weight, alpha is a self-learning rate, beta 1 And beta 2 Is a superparameter->Is the estimated gradient of iteration t;
visual output unit: and visually displaying the prediction result, predicting the probability of postoperative delirium of the patient, and giving early warning to the high-risk patient before the operation.
Further, the model training further comprises sample training, and the sample training method comprises the following steps:
1) Selecting a sample, selecting 600 patients with age more than or equal to 60 years old subjected to semi-hip joint replacement surgery to form an experimental group, collecting blood of the patients on the day of surgery and in the morning of the first day after surgery, freezing serum, performing liquid chromatography analysis and high-resolution mass spectrometry on the serum, and obtaining original data of metabolites by adopting a liquid chromatography tandem mass spectrometry;
2) Data analysis, namely importing original data into MS-real software to perform preprocessing, including peak extraction, peak positioning, denoising, deconvolution, digital peak calibration, filtering and normalization, and identifying metabolites by using a mass spectrum database;
3) Diagnosis of delirium after operation, namely performing diagnosis of delirium after operation on an experiment group by adopting a consciousness confusion evaluation method, and dividing the experiment group into a delirium group and a comparison group;
4) Model training, namely performing contrast iteration on delirium groups and contrast groups by using data subjected to deep learning pretreatment by a machine learning method, and optimizing the model.
Further, the metabolites include, but are not limited to, omega 3 fatty acids, omega 6 fatty acids, AAA, and BCAA.
Further, the visual output unit comprises a display screen, and the probability is displayed on the display screen by using colors, lines, pictures or characters.
The postoperative delirium prediction method based on high-throughput target metabonomics analysis by using the system comprises the following steps:
1) Collecting data, namely collecting blood of a patient on the day of operation and in the morning after operation, freezing serum, performing liquid chromatography analysis and high-resolution mass spectrometry on the serum, and obtaining original data of metabolites by adopting a liquid chromatography tandem mass spectrometry method;
2) Data processing, namely, importing the original data into MS-real software for preprocessing, including peak extraction, peak positioning, denoising, deconvolution, number peak calibration, filtering and normalization, and then using a mass spectrum database to identify and screen metabolites;
3) Model fitting, namely deep learning the preprocessed data by using a machine learning method, taking a serum sample on the current day of operation and a serum sample after operation as training sets, and randomly circulating for multiple iterations to realize diversified fitting, wherein the following model optimization deep learning algorithm is adopted:
L t (W) is an objective function, x i And y i Respectively representing the expression level of the metabolite in different dimensions,representing the minimum training data amount, gamma being the forgetting factor, l being the loss function, b being the training sample number, W being the weight, r being the convex normalization function;
further, the weight W in equation (1) is updated using the following algorithm:
Step1：M 0 ＝0，R 0 ＝0
Step2：
Step3：
Step4：
Step5：
Step6：
Step7：Return W
where M is a first momentum predicted value, R is a second momentum predicted value,is a first momentum correction value,/>Is a second momentum correction value, W is a weight, alpha is a self-learning rate, beta 1 And beta 2 Is a superparameter->Is the estimated gradient of iteration t;
4) And outputting a fitting result, and outputting the prediction probability obtained by model fitting to the visual equipment.
Further, the metabolites include, but are not limited to, omega 3 fatty acids, omega 6 fatty acids, AAA, and BCAA.
Further, the probability is displayed on the visual device using colors, lines, pictures or text.
Compared with the prior art, the invention has the beneficial effects that: through high flux target metabonomics analysis, combining artificial intelligence and machine deep learning modeling, a high-precision and high-sensitivity prediction model of delirium after old patients are established, the purpose of early warning can be achieved before an operation for high-risk patients, delirium after the operation can be identified early, and a new method for preventing and treating delirium after the operation can be provided. In addition, the method specifically optimizes the machine learning algorithm based on postoperative delirium prediction, improves the iteration efficiency and accuracy of the prediction model, and can more efficiently and accurately predict postoperative delirium of elderly patients.
Drawings
Fig. 1 is a schematic diagram of a post-operative delirium prediction system based on high throughput target metabonomics analysis.
Fig. 2 is a flow chart of a method of postoperative delirium prediction based on high throughput target metabonomics analysis.
Detailed Description
The invention is described in further detail below with reference to the attached drawing figures:
fig. 1 is a schematic diagram of a post-operative delirium prediction system based on high throughput target metabonomics analysis, the post-operative delirium prediction system comprising:
and the data acquisition unit (01) is used for collecting blood of a patient on the day of operation and in the morning after operation, freezing and storing serum, carrying out liquid chromatography analysis and high-resolution mass spectrometry on the serum, and obtaining the original data of the metabolite by adopting liquid chromatography tandem mass spectrometry.
Liquid chromatography tandem mass spectrometry (LC-MS/MS) is a separation detection technology, and is a sensitive high-flux identification and quantitative analysis means. And separating out each component of the 40 mu ml serum sample by adopting a liquid chromatography tandem mass spectrometry, and then arranging and separating in a liquid chromatography emission mechanism to obtain separation peaks, wherein the liquid chromatography and the mass spectrometry are combined, so that the quantitative and qualitative analysis of each metabolite in the serum sample is more accurate, and higher sensitivity and accuracy are obtained.
And the data analysis unit (02) is used for guiding the original data of the metabolite obtained by the liquid chromatography tandem mass spectrometry into MS-real software for preprocessing, including peak extraction, peak positioning, noise removal, deconvolution, number peak calibration, filtering and normalization, and then using a mass spectrum database to identify and screen the metabolite.
And (3) preparing a reference peak map, fitting each sample peak with the reference peak map, filtering the aligned peaks, and carrying out missing value assignment so as to remove false peaks. All peak signal intensities were converted to relative intensities and the data for each serum sample were processed one by one to complete normalization.
The model training unit (03) uses a machine learning method to deeply learn the preprocessed data, takes a serum sample on the current day of operation and a serum sample after operation as a training set, randomly circulates for multiple iterations to realize diversified fitting, wherein the following model optimization deep learning algorithm is adopted:
L t (W) is an objective function, x i And y i Respectively representing the expression level of the metabolite in different dimensions,representing the minimum training data amount, gamma being the forgetting factor, l being the loss function, b being the training sample number, W being the weight, r being the convex normalization function;
further, the weight W in equation (1) is updated using the following algorithm:
Step1：M 0 ＝0，R 0 ＝0
Step2：
Step3：
Step4：
Step5：
Step6：
Step7：Return W
where M is a first momentum predicted value, R is a second momentum predicted value,is a first momentum correction value,/>Is a second momentum correction value, W is a weight, alpha is a self-learning rate, beta 1 And beta 2 Is a superparameter->Is the estimated gradient of iteration t.
And the visual output unit (04) is used for visually displaying the prediction result, predicting the probability of postoperative delirium of the patient and giving early warning to the high-risk patient before the operation.
In order to further optimize the model, the invention also carries out sample training, which is specifically as follows:
selecting samples, selecting 600 patients with age more than or equal to 60 years old subjected to semi-hip replacement surgery to form an experimental group, collecting blood of the patients on the day of surgery and in the morning of the first day after surgery, freezing serum, numbering serum samples of each person according to 1-600, respectively performing liquid chromatography analysis and high-resolution mass spectrometry on all the serum samples, and obtaining original data of metabolites by adopting a liquid chromatography tandem mass spectrometry method;
step two, data analysis, namely importing original data into MS-real software for preprocessing, including peak extraction, peak positioning, noise removal, deconvolution, number peak calibration, filtering and normalization, and identifying metabolites by using a mass spectrum database;
step three, diagnosis of delirium after operation, wherein delirium assessment diagnosis is carried out twice daily on 600 patients subjected to semi-hip replacement operation in study on the first to third days after operation, and the assessment diagnosis adopts a consciousness confusion assessment method: (1) acute onset, fluctuating conditions; (2) inattention; (3) disordered thinking; (4) Altered consciousness levels, (1) simultaneous presence of (2), plus either (3) or (4), are diagnosed as the occurrence of postoperative delirium. According to the judgment, dividing the 600 patients into delirium groups and contrast groups, and storing the corresponding serum sample pretreatment data groups;
and fourthly, training the model, namely performing contrast iteration on the delirium group and the contrast group by using data subjected to deep learning pretreatment by a machine learning method, optimizing the model, and improving the expected accuracy of the model.
Further, the visual output unit comprises a display screen on which the probability is displayed using colors, lines, pictures or text.
The inventors have found that the absence of omega 3 and omega 6 fatty acids, disorders of energy metabolism and oxidative stress and imbalance of AAA and BCAA metabolism all lead to an increased probability of postoperative delirium. Accordingly, metabolites of great interest in serum samples of the present invention include, but are not limited to, omega 3 fatty acids, omega 6 fatty acids, aromatic Amino Acids (AAA), and Branched Chain Amino Acids (BCAA).
The invention also provides a method for predicting postoperative delirium based on high-throughput target metabonomics analysis, and fig. 2 is a flowchart of the method for predicting postoperative delirium, specifically comprising:
1) Collecting data, namely collecting blood of a patient on the day of operation and in the morning after operation, freezing serum, performing liquid chromatography analysis and high-resolution mass spectrometry on the serum, and obtaining original data of metabolites by adopting a liquid chromatography tandem mass spectrometry method;
2) Data processing, namely, importing the original data into MS-real software for preprocessing, including peak extraction, peak positioning, denoising, deconvolution, number peak calibration, filtering and normalization, and then using a mass spectrum database to identify and screen metabolites;
3) Model fitting, namely deep learning the preprocessed data by using a machine learning method, taking a serum sample on the current day of operation and a serum sample after operation as training sets, and randomly circulating for multiple iterations to realize diversified fitting, wherein the following model optimization deep learning algorithm is adopted:
L t (W) is an objective function, x i And y i Respectively representing the expression level of the metabolite in different dimensions,representing the minimum training data amount, gamma being the forgetting factor, l being the loss function, b being the training sample number, W being the weight, r being the convex normalization function;
further, the weight W in equation (1) is updated using the following algorithm:
Step1：M 0 ＝0，R 0 ＝0
Step2：
Step3：
Step4：
Step5：
Step6：
Step7：Return W
where M is a first momentum predicted value, R is a second momentum predicted value,is a first momentum correction value,/>Is a second momentum correction value, W is a weight, alpha is a self-learning rate, beta 1 And beta 2 Is a superparameter->Is the estimated gradient of iteration t;
4) And outputting a fitting result, and outputting the prediction probability obtained by model fitting to the visual equipment.
Further, the metabolites include, but are not limited to, omega 3 fatty acids, omega 6 fatty acids, aromatic Amino Acids (AAA), and Branched Chain Amino Acids (BCAA).
Further, the probability is displayed on the visual device using colors, lines, pictures or text.
According to the invention, through high-flux target metabonomics analysis and combining artificial intelligence and machine deep learning modeling, a high-precision and high-sensitivity prediction model of delirium after operation of the elderly patient is established, the purpose of early warning can be achieved before operation for the high-risk patient, delirium after operation can be identified early, and a new method for preventing and treating delirium after operation can be provided. In addition, the method specifically optimizes the machine learning algorithm based on postoperative delirium prediction, improves the iteration efficiency and accuracy of the prediction model, and can more efficiently and accurately predict postoperative delirium of elderly patients.
In the description of the present invention, it should be noted that, unless explicitly specified and limited otherwise, the terms "connected," "connected," and "connected" are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
In the description of the present invention, unless otherwise indicated, the terms "upper," "lower," "left," "right," "inner," "outer," and the like are used for convenience in describing the present invention and simplifying the description based on the orientation or positional relationship shown in the drawings, and do not denote or imply that the devices or elements in question must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention.
Finally, it should be noted that the above-mentioned technical solution is only one embodiment of the present invention, and various modifications and variations can be easily made by those skilled in the art based on the application methods and principles disclosed in the present invention, and are not limited to the methods described in the above-mentioned specific embodiments of the present invention, therefore, the foregoing description is only preferred, and not meant to be limiting.

Claims (7)

1. A post-operative delirium prediction system based on high-throughput target metabonomics analysis, comprising:
(1) A data acquisition unit: collecting blood of a patient on the day of operation and in the morning after operation, freezing serum, performing liquid chromatography analysis and high-resolution mass spectrometry on the serum, and obtaining original data of metabolites by adopting a liquid chromatography tandem mass spectrometry method;
(2) A data analysis unit: the original data is imported into MS-dial software for preprocessing, including peak extraction, peak positioning, noise removal, deconvolution, digital peak calibration, filtering and normalization, and then a mass spectrum database is used for identifying and screening metabolites;
(3) Model training unit: deeply learning the preprocessed data by using a machine learning method, taking a serum sample on the current day of operation and a serum sample after operation as training sets, and randomly circulating for multiple iterations to realize diversified fitting; wherein, the following model is adopted to optimize the deep learning algorithm:
L t (W) is an objective function, x i And y i Respectively representing the expression level of the metabolite in different dimensions,representing the minimum training data amount, gamma being the forgetting factor, l being the loss function, b being the training sample number, W being the weight, r being the convex normalization function;
further, the weight W in equation (1) is updated using the following algorithm:
Step1：M 0 ＝0，R 0 ＝0
Step2：M t ＝β 1 M t-1 +(1-β 1 )▽l t (W t-1 )
Step3：R t ＝β 2 R t-1 +(1-β 2 )▽l t (W t-1 ) 2
Step4：
Step5：
Step6：
Step7：Return W
where M is a first momentum predicted value, R is a second momentum predicted value,is a first momentum correction value,/>Is a second momentum correction value, W is a weight, alpha is a self-learning rate, beta 1 And beta 2 Is a superparameter->Is the estimated gradient of iteration t;
(4) Visual output unit: and visually displaying the prediction result, predicting the probability of postoperative delirium of the patient, and giving early warning to the high-risk patient before the operation.
2. The postoperative delirium prediction system according to claim 1, characterized in that: the model training further comprises a sample training process, and the method for training the sample comprises the following steps:
1) Selecting a sample, selecting 600 patients with age more than or equal to 60 years old subjected to semi-hip joint replacement surgery to form an experimental group, collecting blood of the patients on the day of surgery and in the morning of the first day after surgery, freezing serum, performing liquid chromatography analysis and high-resolution mass spectrometry on the serum, and obtaining original data of metabolites by adopting a liquid chromatography tandem mass spectrometry;
2) Data analysis, namely importing original data into MS-real software to perform preprocessing, including peak extraction, peak positioning, denoising, deconvolution, digital peak calibration, filtering and normalization, and identifying metabolites by using a mass spectrum database;
3) Diagnosis of delirium after operation, namely performing diagnosis of delirium after operation on an experiment group by adopting a consciousness confusion evaluation method, and dividing the experiment group into a delirium group and a comparison group;
4) Model training, namely performing contrast iteration on delirium groups and contrast groups by using data subjected to deep learning pretreatment by a machine learning method, and optimizing the model.
3. The postoperative delirium prediction system according to claim 1, characterized in that: such metabolites include, but are not limited to, omega 3 fatty acids, omega 6 fatty acids, aromatic amino acids, and branched chain amino acids.
4. The postoperative delirium prediction system according to claim 1, characterized in that: the visual output unit comprises a display screen, and the probability is displayed on the display screen by using colors, lines, pictures or characters.
5. A method of postoperative delirium prediction based on high-throughput target metabonomics analysis, comprising:
1) Collecting data, namely collecting blood of a patient on the day of operation and in the morning after operation, freezing serum, performing liquid chromatography analysis and high-resolution mass spectrometry on the serum, and obtaining original data of metabolites by adopting a liquid chromatography tandem mass spectrometry method;
2) Data processing, namely, importing the original data into MS-real software for preprocessing, including peak extraction, peak positioning, denoising, deconvolution, number peak calibration, filtering and normalization, and then using a mass spectrum database to identify and screen metabolites;
3) Model fitting, namely deep learning the preprocessed data by using a machine learning method, taking a serum sample on the current day of operation and a serum sample after operation as training sets, and randomly circulating for multiple iterations to realize diversified fitting, wherein the following model optimization deep learning algorithm is adopted:
L t (W) is an objective function, x i And y i Respectively representing the expression level of the metabolite in different dimensions,representing the minimum training data amount, gamma being the forgetting factor, l being the loss function, b being the training sample number, W being the weight, r being the convex normalization function;
further, the weight W in equation (1) is updated using the following algorithm:
Step1：M 0 ＝0，R 0 ＝0
Step2：
Step3：
Step4：
Step5：
Step6：
Step7：Return W
where M is a first momentum predicted value, R is a second momentum predicted value,is a first momentum correction value,/>Is a second momentum correction value, W is a weight, alpha is a self-learning rate, beta 1 And beta 2 Is a superparameter->Is the estimated gradient of iteration t;
4) And outputting a fitting result, and outputting the prediction probability obtained by model fitting to the visual equipment.
6. The method according to claim 5, wherein: such metabolites include, but are not limited to, omega 3 fatty acids, omega 6 fatty acids, aromatic amino acids, and branched chain amino acids.
7. The method according to claim 5, wherein: the probability is displayed on the visual device using colors, lines, pictures or text.
CN202310270601.3A 2023-03-20 2023-03-20 Postoperative delirium prediction system and method for high-throughput target metabonomics analysis Active CN116313092B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310270601.3A CN116313092B (en) 2023-03-20 2023-03-20 Postoperative delirium prediction system and method for high-throughput target metabonomics analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310270601.3A CN116313092B (en) 2023-03-20 2023-03-20 Postoperative delirium prediction system and method for high-throughput target metabonomics analysis

Publications (2)

Publication Number Publication Date
CN116313092A CN116313092A (en) 2023-06-23
CN116313092B true CN116313092B (en) 2023-09-01

Family

ID=86825350

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310270601.3A Active CN116313092B (en) 2023-03-20 2023-03-20 Postoperative delirium prediction system and method for high-throughput target metabonomics analysis

Country Status (1)

Country Link
CN (1) CN116313092B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115192716A (en) * 2022-08-09 2022-10-18 中国人民解放军空军特色医学中心 Methods and systems for predicting, preventing or treating post-operative delirium
CN115274115A (en) * 2022-08-26 2022-11-01 南通大学附属医院 Method for constructing Nomogram prediction model for mechanical ventilation time extension of patient after great cardiac vascular surgery
CN115565683A (en) * 2021-07-02 2023-01-03 南通大学附属医院 Method for establishing and verifying heart great vessel postoperative delirium risk prediction model

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115565683A (en) * 2021-07-02 2023-01-03 南通大学附属医院 Method for establishing and verifying heart great vessel postoperative delirium risk prediction model
CN115192716A (en) * 2022-08-09 2022-10-18 中国人民解放军空军特色医学中心 Methods and systems for predicting, preventing or treating post-operative delirium
CN115274115A (en) * 2022-08-26 2022-11-01 南通大学附属医院 Method for constructing Nomogram prediction model for mechanical ventilation time extension of patient after great cardiac vascular surgery

Also Published As

Publication number Publication date
CN116313092A (en) 2023-06-23

Similar Documents

Publication Publication Date Title
CN110141219A (en) Myocardial infarction automatic testing method based on lead fusion deep neural network
CN110033860B (en) Method for improving detection rate of genetic metabolic diseases based on machine learning
CN108549875B (en) Electroencephalogram epileptic seizure detection method based on depth channel attention perception
CN103776891B (en) A kind of method of detection differential expression protein
Mattauch et al. A bibliometric approach for detecting the gender gap in computer science
US20100036791A1 (en) Examination value predicting device using electrophoresis waveform, prediction method, and predicting program
Mantini et al. Independent component analysis for the extraction of reliable protein signal profiles from MALDI-TOF mass spectra
CN111557672A (en) Nicotinic acid skin reaction image analysis method and equipment
CN105611872A (en) An apparatus and method for evaluating multichannel ECG signals
CN114550929A (en) Juvenile bone age data management method based on big data and abnormity early warning method
Alexander et al. Using unsupervised learning to identify clinical subtypes of Alzheimer's disease in electronic health records
CN116313092B (en) Postoperative delirium prediction system and method for high-throughput target metabonomics analysis
Pagali et al. Delirium occurrence and association with outcomes in hospitalized COVID-19 patients
CN111710423A (en) Method for determining mood disorder morbidity risk probability based on regression model
US20210158899A1 (en) Biomarker detection method, disease assessment method, biomarker detection device, and computer readable medium
CN111161868A (en) Medical quick inspection management system
CN115206420B (en) Construction method and application of schizophrenia abnormal gene-metabolism regulation network
Cochrane Infectious Diseases Group et al. Impact of diagnostic strategies for tuberculosis using lateral flow urine lipoarabinomannan assay in people living with HIV
CN111210876B (en) Disturbed metabolic pathway determination method and system
CN111462842B (en) Clinical application evaluation system and method for liquid chromatography-tandem mass spectrometry platform
CN113539397A (en) Real-time quality control system only using patient sample
CN113255789A (en) Video quality evaluation method based on confrontation network and multi-tested electroencephalogram signals
CN112288033B (en) Full-automatic fluorescence immunoassay data processing method, system and device
CN113180701B (en) Electroencephalogram signal deep learning method for image label labeling
CN110889836A (en) Image data analysis method and device, terminal equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant