echemi logo
Product
  • Product
  • Supplier
  • Inquiry
    Home > Active Ingredient News > Endocrine System > Nat Med: A neural network model based on blood metabolomics to predict 24 diseases

    Nat Med: A neural network model based on blood metabolomics to predict 24 diseases

    • Last Update: 2022-10-12
    • Source: Internet
    • Author: User
    Search more information of high quality chemicals, good prices and reliable suppliers, visit www.echemi.com

    Risk stratification is central


    On September 22, 2022, the Roland Eils team at the Charite Medical School in Berlin and Duo published an article


    Study population and neural network models for predicting diseases based on metabolic profiles

    The authors first extracted valid information on metabolic profiles and disease development risk based on the UK Biobank cohort of 22 central 117981 participants, trained a neural network model based on metabolic profiles to predict 24 disease risks, and validated


     

    Metabolic profiles stratify disease risk at the beginning of the disease

    The key to prevention is to identify high-risk people


     

    Predictive effect of metabolic spectroscopy in combination with clinical predictors

    Information such as gender and age are clinical predictors that are readily available in primary care and are often used for risk stratification


    The authors used the Cox proportional risk model to predict the risk of disease development by predicting the risk of disease alone or in combination with metabolic profile, such as age and sex, ASCVD (set of cardiovascular disease predictors) and PANEL (set of clinical predictors such as blood tests), and quantitatively compared the stratification ability of different indicators with Harrell's C index, and found that the disease prediction stratification of metabolic spectrogram indicators was significantly better than that of age and sex, ASCVD indicators, and the treatment of kidney disease, The prediction effect of liver disease and type II diabetes was significantly better than that of clinical indicators, and the prediction effect of cataracts, glaucoma, skin cancer, colon cancer, rectal cancer and prostate cancer was weaker than that of traditional clinical indicators


    In the comparison of metabolic and clinical indicators, the authors found that metabolic profiles predicted kidney disease, liver disease, type II diabetes, COPD, and heart failure better than metabolic profiles combined with age and sex


    Manifestations of model stratification capabilities in clinical practice

    While the stratification capability of the model is critical, the clinical utility of any risk model depends on calibration and selection of appropriate intervention thresholds


     

    The disadvantage of neural networks in practical applications is that they are very poorly


    The authors found that glutamine, glycine, tyrosine, metabolites of sugar metabolism, albumin, creatinine, acetylated glycoproteins, ketone bodies, fatty acids, and lipoproteins all contributed to


    Next, the authors looked at type II diabetes mellitus and all-cause dementia, where metabolic states make a significant contribution


     

    This article is an English version of an article which is originally in the Chinese language on echemi.com and is provided for information purposes only. This website makes no representation or warranty of any kind, either expressed or implied, as to the accuracy, completeness ownership or reliability of the article or any translations thereof. If you have any concerns or complaints relating to the article, please send an email, providing a detailed description of the concern or complaint, to service@echemi.com. A staff member will contact you within 5 working days. Once verified, infringing content will be removed immediately.

    Contact Us

    The source of this page with content of products and services is from Internet, which doesn't represent ECHEMI's opinion. If you have any queries, please write to service@echemi.com. It will be replied within 5 days.

    Moreover, if you find any instances of plagiarism from the page, please send email to service@echemi.com with relevant evidence.