RESUMO
Objective: Based on the respiratory disease big data platform in southern Xinjiang, we established a model that predicted and diagnosed chronic obstructive pulmonary disease, bronchiectasis, pulmonary embolism and pulmonary tuberculosis, and provided assistance for primary physicians. Methods: The method combined convolutional neural network (CNN) and long-short-term memory network (LSTM) for prediction and diagnosis of respiratory diseases. We collected the medical records of inpatients in the respiratory department, including: chief complaint, history of present illness, and chest computed tomography. Pre-processing of clinical records with "jieba" word segmentation module, and the Bidirectional Encoder Representation from Transformers (BERT) model was used to perform word vectorization on the text. The partial and total information of the fused feature set was encoded by convolutional layers, while LSTM layers decoded the encoded information. Results: The precisions of traditional machine-learning, deep-learning methods and our proposed method were 0.6, 0.81, 0.89, and F1 scores were 0.6, 0.81, 0.88, respectively. Conclusion: Compared with traditional machine learning and deep-learning methods that our proposed method had a significantly higher performance, and provided precise identification of respiratory disease.
Assuntos
Memória de Curto Prazo , Redes Neurais de Computação , Aprendizado de MáquinaRESUMO
OBJECTIVE: To explore the relationship between the peroxisome proliferator-activated receptor γ(PPARγ) 2 gene Pro12 Ala polymorphism and the risk of obesity in the Chinese population. METHODS: Take "peroxisome proliferator activated receptor gamma2", "PPAR gamma2", "pparg2", "peroxisome proliferator-activated receptor gamma2", "obesity", "Pro12 Ala", "gene polymorphism", "Chinese" as searchable words, Chinese people were selected as study subjects from Chinese databases such as CNKI, Wanfang Database, VIP and foreign language databases such as PubMed, Embase, Cochrane Library and Web of Science, which were published from the establishment of the database until December 31, 2019. RevMan 5. 3 software was used for Meta-analysis of the literature with literature score ≥ 5 points. RESULTS: A total of 12 articles were included, including 2233 cases and 1795 healthy controls. Meta-analysis showed that:(1) Dominant gene model: the risk of obesity for those carrying Ala/Ala+Pro/Ala was 1. 86 times that of those carrying Pro/Pro(OR=1. 86, 95%CI 1. 07-3. 21). (2) Additive gene model: the risk of obesity for those carrying Pro/Ala was 1. 79 times that of those carrying Pro/Pro(OR=1. 79, 95% CI 1. 04-3. 08). (3) Allelic model: the risk of obesity in those carrying the allele Ala was 1. 75 times that of those carrying the allele Pro(OR=1. 75, 95% CI 1. 11-2. 78). The subgroup analysis of different regions showed that the combined OR(95% CI) values of the dominant gene model, additive gene model and allelic model in the southern region were 1. 96(1. 12-3. 42), 2. 02(1. 14-3. 56) and 1. 79(1. 06-3. 03), the combined OR(95% CI) values of the dominant gene model, additive gene model and allele model in the northern region were 1. 89(0. 90-3. 97), 1. 77(0. 85-3. 67) and 1. 80(0. 97-3. 36). CONCLUSION: The polymorphic allele Ala or genotype Ala/Ala or Pro/Ala of PPARγ2 gene may increase the risk of obesity in Chinese population. The correlation between the PPARγ2 gene Pro12 Ala polymorphism and the risk of obesity in the southern population is more obvious.
Assuntos
Predisposição Genética para Doença , PPAR gama , China/epidemiologia , Genótipo , Humanos , Idioma , Obesidade/genética , PPAR gama/genéticaRESUMO
OBJECTIVE: To explore the dose-response relationship between maternal polybrominated diphenyl ethers(PBDEs) exposure and neonatal low birth weight(LBW) by restrictive cubic spline method. METHODS: Blood samples from 96 full-term primiparas in Urumqi were collected and the concentrations of seven PBDEs(BDE-28, BDE-47, BDE-99, BDE-100, BDE-153, BDE-154, BDE-183) in serum were measured to calculate the total PBDEs concentration(ΣPBDEs). The sociodemographic data of pregnant women and their spouses, prenatal examination, nutritional status during pregnancy, internal and external environmental factors and neonatal birth outcomes were collected to analyze the influencing factors affecting the low birth weight of the newborn and explore the dose-response relationship between the exposure level of maternal PBDEs and the birth weight of the newborn. RESULTS: (1) ΣPBDEs ranged from 144.67 to 32884.70 pg/g lipid weight(LW), and the median concentration was 3930.93 pg/g LW. BDE-153 was the main source of contribution. The median concentration was 41.04%.(2) There was a positive correlation between ΣPBDEs and birth outcomes such as birth length and body weight, both P<0.05.(3) There were significant differences in the distribution of PBDEs, weight gain during pregnancy, birth weight, Apgar score, house decoration before and after pregnancy, number of home appliances and number of carpets in normal birth weight infants and low birth weight infants(P<0.05).(4) Restricted cubic splines showed a non-linear "â©" dose-response relationship between the exposure level of PBDEs during pregnancy and the intensity of birth weight. The relationship between the exposure level of PBDEs during pregnancy and the intensity of LBW showed a weak nonlinear "^" dose-response relationship. The correlation intensity between the level and the normal birth weight showed a non-linear "^" dose response relationship(non-linear test, both P<0.05). CONCLUSION: The blood level of PBDEs in pregnant women in Urumqi is low, but it can affect the birth outcome of newborns. The correlation between the exposure level of PBDEs and birth weight during pregnancy is nonlinear dose-response relationship.