Application of Artificial Intelligence in the Establishment of an Association Model between Metabolic Syndrome, TCM Constitution, and the Guidance of Medicated Diet Care

Metabolic syndrome prediction model using Bayesian optimization and XGBoost based on traditional Chinese medicine features

Metabolic syndrome (MetS) has a high prevalence and is prone to many complications. However, current MetS diagnostic methods require blood tests that are not conducive to self-testing, so a user-friendly and accurate method for predicting MetS is needed to facilitate early detection and treatment. In this study, a MetS prediction model based on a simple, small number of Traditional Chinese Medicine (TCM) clinical indicators and biological indicators combined with machine learning algorithms is investigated. Electronic medical record data from 2040 patients who visited outpatient clinics at Guangdong Chinese medicine hospitals from 2020 to 2021 were used to investigate the fusion of Bayesian optimization (BO) and eXtreme gradient boosting (XGBoost) in order to create a BO-XGBoost model for screening nineteen key features in three categories: individual bio-information, TCM indicators, and TCM habits that influence MetS prediction. Subsequently, the predictive diagnostic model for MetS was developed. The experimental results revealed that the model proposed in this paper achieved values of 93.35 %, 90.67 %, 80.40 %, and 0.920 for the F1, sensitivity, FRS, and AUC metrics, respectively. These values outperformed those of the seven other tested machine learning models. Finally, this study developed an intelligent prediction application for MetS based on the proposed model, which can be utilized by ordinary users to perform self-diagnosis through a web-based questionnaire, thereby accomplishing the objective of early detection and intervention for MetS.

Comparison of the predictive value of four insulin resistance surrogates for the prevalence of hypertension: a population-based study

BACKGROUND

Several studies have investigated the association of insulin resistance (IR) surrogates and the risk of hypertension. However, it is unclear whether there exist differences between different IR surrogates and hypertension risk. Therefore, this study aimed to explore the association of four IR surrogates (triglyceride-glucose index (TyG index), triglyceride-glucose index with body mass index (TyG-BMI), triglycerides/high-density lipoprotein cholesterol ratio (TG/HDL-c), and metabolic score for IR (METS-IR)) with the prevalence of hypertension.

METHODS

This is a cross-sectional study with a total of 117,056 participants. Data were extracted from a computerized database established by Rich Healthcare Group in China, which included all medical records of participants who received a health check-up from 2010 to 2016. IR surrogates were grouped into quartiles as continuous variables, and multivariate logistic regression was performed to estimate the association between different IR surrogate levels and the prevalence of hypertension. Results were expressed as odds ratios (ORs) and 95% confidence intervals (CIs). Missing data were accounted by multiple imputation. These analyses were considered as the sensitivity analysis. Meanwhile, the Bayesian network (BN) model was constructed to further evaluate the relationship between baseline characteristics and the four IR surrogates and the prevalence of hypertension, as well as the importance of every single variable for the prevalence of hypertension.

RESULTS

Multivariate logistic regression analysis revealed that TyG-BMI and METS-IR were independent risk factors for the prevalence of hypertension that increased significantly with increasing TyG-BMI and METS-IR (p for trend < 0.001). The area under the TyG-BMI curve (AUC) was 0.681 [95% CI: 0.677-0.685], and the cut-off value was 199.5, with a sensitivity and specificity of 65.57% and 61.18%, respectively. While the area under the METS-IR curve (AUC) was 0.679 [95% CI: 0.674-0.683], and the cut-off value was 33.61, with a sensitivity and specificity of 69.67% and 56.67%, respectively. The BN model presented that among these four IR surrogates and related variables, TyG-BMI was the most important predictor of hypertension prevalence, with a significance of 34%. The results before and after multiple imputation were similar.

CONCLUSION

TyG-BMI and METS-IR were independent risk factors for the prevalence of hypertension. TyG-BMI and METS-IR had good predictive value for the prevalence of hypertension, and TyG-BMI was superior to METS-IR.

Relationship Between Lipid Profiles and Hypertension: A Cross-Sectional Study of 62,957 Chinese Adult Males

Background

Patterns of dyslipidemia and incidence of hypertension have been rarely reported in Asian populations with inconsistent findings. To accumulate further evidence in Asian populations, the study aimed to investigate the relationship between lipid profiles and hypertension in Chinese adult males.

Methods

We conducted a cross-sectional study based on the data from the DATADRYAD database. The overall population was divided into hypertensive and non-hypertensive groups based on baseline blood pressure levels. For continuous variables, Mann-Whitney test was performed between two groups, while Kruskal-Wallis and Dunn tests were used among multiple groups. The chi-square test was carried out for dichotomous variables. Spearman's correlation coefficient was employed to assess the association between systolic blood pressure (SBP), diastolic blood pressure (DBP) and lipid profiles, whereas the relationship between lipid profiles and the incidence of hypertension was evaluated using multivariate logistic regression. The Bayesian network (BN) model was adopted to investigate the relationship between clinical characteristics and hypertension, and the importance of related predictor to the incidence of hypertension was obtained to make conditional probability analysis.

Results

Finally, totally 62,957 participants were included in this study. In the lipid profiles, total cholesterol (TC), low-density cholesterol (LDL-c), and non- high-density lipoprotein cholesterol (non-HDL-c) were higher in the hypertensive population (p <0.001). In the fully multivariate model, for every 1 mg/dl increase in TC, LDL-c and non-HDL, the risk of hypertension increased by 0.2% [1.002 (1.001–1.003)], 0.1% [1.001 (1.000–1.002)], and 0.1% [1.001 (1.000–1.002)]. Meanwhile, HDL-c became positively associated with the incidence of hypertension (p for trend < 0.001) after adjusting for the body mass index (BMI), and 1 mg/dl increment in HDL-c increased the risk of hypertension by 0.2% [1.002 (1.000–1.002)] after fully adjusting for multiple variables. Furthermore, the BN showed that the importance of age, BMI, fasting plasma glucose (FPG), and TC to the effect of hypertension is 43.3, 27.2, 11.8, and 5.1%, respectively.

Conclusion

Elevated TC, LDL-c, and non-HDL-c were related to incidence of hypertension in Chinese adult males, whereas triglycerides (TG) was not significantly associated. The relationship between HDL-c and hypertension incidence shifted from no association to a positive correlation after adjusting for the BMI. Moreover, the BN model displayed that age, the BMI, FPG, and TC were strongly associated with hypertension incidence.

Differences in Lipid Profiles and Atherogenic Indices Between Hypertensive and Normotensive Populations: A Cross-Sectional Study of 11 Chinese Cities

Background

Several previous studies have reported that dyslipidemia is associated with the risk of hypertension, but these studies are mainly conducted in European and US populations, with a very few studies in the Asian population. Moreover, the effects of atherosclerotic indices, including atherogenic coefficient (AC) and atherogenic risk of plasma (AIP), on hypertension in Asians have not been well described so far.

Methods

From 2010 to 2016, altogether 211,833 Chinese adults were ultimately recruited at the health centers in 11 Chinese cities (including Shanghai, Beijing, Nanjing, Suzhou, Shenzhen, Changzhou, Chengdu, Guangzhou, Hefei, Wuhan, and Nantong). Differences in continuous variables between the two groups were analyzed by the Mann-Whitney test, while those in categorical variables were examined by the Chi-squared test. Logistic regression was applied to evaluate the association between lipid profiles and the risk of hypertension. The predictive values of AC and AIP for the incidence of hypertension were analyzed using the area under the receiver operating characteristic (ROC) curve. Meanwhile, Bayesian network (BN) models were performed to further analyze the associations between the different covariates and the incidence of hypertension.

Results

A total of 117,056 participants were included in the final analysis. There were significant differences in baseline characteristics between normotension and hypertension groups (p < 0.001). In multivariate logistic regression, the risk of hypertension increased by 0.2% (1.002 [1.001-1.003]), 0.2% (1.002 [1.001-1.003]), and 0.2% (1.002 [1.001-1.003]) per 1 mg/dl increase in total cholesterol (TC), low-density lipoprotein (LDL), and non-high-density lipoprotein cholesterol (non-HDL-c), respectively. However, after adjusting for body mass index (BMI), an increase in HDL level was associated with a higher risk of hypertension (p for a trend < 0.001), and the risk of hypertension increased by 0.6% per 1 mg/dl increase in HDL-c (1.006 [1.003-1.008]). In women, AC had the highest predictive value for the incidence of hypertension with an area under the curve (AUC) of 0.667 [95% confidence interval (CI): 0.659-0.674]. BN models suggested that TC and LDL were more closely related to the incidence of hypertension.

Conclusions

Overall, lipid profiles were significantly abnormal in the hypertensive population than in the normotensive population. TC and LDL were strongly associated with the incidence of hypertension. TC, LDL, and non-HDL-c levels show a positive association, HDL-c shows a negative association, while TG is not significantly associated with the risk of hypertension. After adjusting for BMI, HDL-c turns out to be positively associated with the risk of hypertension. In addition, AC has a good predictive value for the incidence of hypertension in women.

Application of Machine Learning in Developing Decision-Making Support Models for Decompressed Vertebroplasty

BACKGROUND

The common treatment methods for vertebral compression fractures with osteoporosis are vertebroplasty and kyphoplasty, and the result of the operation may be related to the value of various measurement data during the operation.

MATERIAL AND METHOD

This study mainly uses machine learning algorithms, including Bayesian networks, neural networks, and discriminant analysis, to predict the effects of different decompression vertebroplasty methods on preoperative symptoms and changes in vital signs and oxygen saturation in intraoperative measurement data.

RESULT

The neural network shows better analysis results, and the area under the curve is >0.7. In general, important determinants of surgery include numbness and immobility of the lower limbs before surgery.

CONCLUSION

In the future, this association model can be used to assist in decision making regarding surgical methods. The results show that different surgical methods are related to abnormal vital signs and may affect the length of hospital stay.

Association of Lipid Levels With the Prevalence of Hypertension in Chinese Women: A Cross-Sectional Study Based on 32 Health Check Centers

BACKGROUND

Dyslipidemia is strongly associated with the development of hypertension. In our previous study, it was shown that elevated TC, LDL-c, and non-HDL-c were associated with the prevalence of hypertension in Chinese men, whereas the relationship between HDL-c and hypertension shifted from no association to a positive association after adjusting for the BMI. To further accumulate epidemiological evidence in Asian women, this study aimed to investigate the relationship between lipid profile and prevalence of hypertension in Chinese adult women.

METHODS

This is a cross-sectional study including 54,099 Chinese women aged>20 years at 32 health screening centers in 11 cities from 2010-2016. The original data were obtained from DATADRYAD database (www.datadryad.org). Besides, the overall women were classified into non-hypertensive and hypertensive groups based on baseline blood pressure levels. Differences between the two groups were examined by Man-Whitney test or Chi-square test. Spearman's correlation coefficient was employed to evaluate the correlation between systolic blood pressure (SBP), diastolic blood pressure (DBP) and lipid profiles. Multivariate logistic regression was performed to estimate the relationship between different lipid levels and the prevalence of hypertension. Odds ratios (ORs) and 95% confidence intervals (CIs) indicated the risk of lipid and hypertension. Bayesian model (BN) model was constructed to further assess the relationship between baseline characteristics and the prevalence of hypertension, as well as the importance of each variable for the prevalence of hypertension.

RESULTS

Compared to the non-hypertensive population, the hypertensive population was older, and had the higher body mass index (BMI), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-c), serum creatinine (Scr), fasting blood glucose (FPG), blood urea nitrogen (BUN), alanine aminotransferase (ALT), aspartate aminotransferase (AST), and non-high-density lipoprotein cholesterol (non-HDL-c), but HDL-c and the presence concerning the family history of diabetes were lower. Multivariate logistic regression analysis revealed that TC, LDL-c, and non-HDL-c showed a positive trend with hypertension risk (p for trend < 0.05) whereas TC and HDL-c were not significantly associated with hypertension prevalence. Moreover, each 1 mg/dl increase in TC, LDL, and non-HDL hypertension prevalence increased by 0.2% [1.002 (1.000-1.003)], 0.2% [1.002 (1.000-1.004)], and 0.2% [1.002(1.001-1.004)], respectively. BN suggested that the importance of age, BMI, FPG, non-HDL-c on the prevalence of hypertension was 52.73%, 24.98%, 11.22%, and 2.34%, respectively.

CONCLUSION

Overall, in Chinese adult women, TC, LDL-c and non-HDL-c levels were higher and HDL-c level was lower in the hypertensive population, whereas TG did not differ significantly from the non-hypertensive population. Meanwhile, TC, LDL-c, and non-HDL-c were positively associated with prevalence of hypertension, and HDL-c was negatively associated with prevalence of hypertension but became nonsignificant after full adjustment for variables. Moreover, BN model suggested that age, BMI, FPG, and non-HDL-c had a greater effect on the development of hypertension.

Comparison of different predicting models to assist the diagnosis of spinal lesions

In neurosurgical or orthopedic clinics, the differential diagnosis of lower back pain is often time-consuming and costly. This is especially true when there are several candidate diagnoses with similar symptoms that might confuse clinic physicians. Therefore, methods for the efficient differential diagnosis can help physicians to implement the most appropriate treatment and achieve the goal of pain reduction for their patients.In this study, we applied data-mining techniques from artificial intelligence technologies, in order to implement a computer-aided auxiliary differential diagnosis for a herniated intervertebral disc, spondylolithesis, and spinal stenosis. We collected questionnaires from 361 patients and analyzed the resulting data by using a linear discriminant analysis, clustering, and artificial neural network techniques to construct a related classification model and to compare the accuracy and implementation efficiency of the different methods.Our results indicate that a linear discriminant analysis has obvious advantages for classification and diagnosis, in terms of accuracy.We concluded that the judgment results from artificial intelligence can be used as a reference for medical personnel in their clinical diagnoses. Our method is expected to facilitate the early detection of symptoms and early treatment, so as to reduce the social resource costs and the huge burden of medical expenses, and to increase the quality of medical care.