Risk of data leakage in estimating the diagnostic performance of a deep-learning-based computer-aided system for psychiatric disorders.

Deep-learning approaches with data augmentation have been widely used when developing neuroimaging-based computer-aided diagnosis (CAD) systems. To prevent the inflated diagnostic performance caused by data leakage, a correct cross-validation (CV) method should be employed, but this has been still overlooked in recent deep-learning-based CAD studies. The goal of this study was to investigate the impact of correct and incorrect CV methods on the diagnostic performance of deep-learning-based CAD systems after data augmentation. To this end, resting-state electroencephalogram (EEG) data recorded from post-traumatic stress disorder patients and healthy controls were augmented using a cropping method with different window sizes, respectively. Four different CV approaches were used to estimate the diagnostic performance of the CAD system, i.e., subject-wise CV (sCV), overlapped sCV (oSCV), trial-wise CV (tCV), and overlapped tCV (otCV). Diagnostic performances were evaluated using two deep-learning models based on convolutional neural network. Data augmentation can increase the performance with all CVs, but inflated diagnostic performances were observed when using incorrect CVs (tCV and otCV) due to data leakage. Therefore, the correct CV (sCV and osCV) should be used to develop a deep-learning-based CAD system. We expect that our investigation can provide deep-insight for researchers who plan to develop neuroimaging-based CAD systems for psychiatric disorders using deep-learning algorithms with data augmentation.

A regression-based machine learning approach for pH and glucose detection with redox-sensitive colorimetric paper sensors.

Colorimetric paper sensors are used in various fields due to their convenience and intuitive manner. However, these sensors present low accuracy in practical use because it is difficult to distinguish color changes for a minute amount of analyte with the naked eye. Herein, we demonstrate that a machine learning (ML)-based paper sensor platform accurately determines the color changes. We fabricated a colorimetric paper sensor by adsorbing polyaniline nanoparticles (PAni-NPs), whose color changes from blue to green when the ambient pH decreases. Adding glucose oxidase (GOx) to the paper sensor enables colorimetric glucose detection. Target analytes (10 µL) were aliquoted onto the paper sensors, and their images were taken with a smartphone under the same conditions in a darkroom. The red-green-blue (RGB) data from the images were extracted and used to train and test three regression models: support vector regression (SVR), decision tree regression (DTR), and random forest regression (RFR). Of the three regression models, RFR performed the best at estimating pH levels (R2 = 0.957) ranging from pH 2 to 10 and glucose concentrations (R2 = 0.922) ranging from 0 to 10 mg mL-1.

An artificial neural-network approach to identify motor hotspot for upper-limb based on electroencephalography: a proof-of-concept study.

BACKGROUND: To apply transcranial electrical stimulation (tES) to the motor cortex, motor hotspots are generally identified using motor evoked potentials by transcranial magnetic stimulation (TMS). The objective of this study is to validate the feasibility of a novel electroencephalography (EEG)-based motor-hotspot-identification approach using a machine learning technique as a potential alternative to TMS. METHODS: EEG data were measured using 63 channels from thirty subjects as they performed a simple finger tapping task. Power spectral densities of the EEG data were extracted from six frequency bands (delta, theta, alpha, beta, gamma, and full) and were independently used to train and test an artificial neural network for motor hotspot identification. The 3D coordinate information of individual motor hotspots identified by TMS were quantitatively compared with those estimated by our EEG-based motor-hotspot-identification approach to assess its feasibility. RESULTS: The minimum mean error distance between the motor hotspot locations identified by TMS and our proposed motor-hotspot-identification approach was 0.22 ± 0.03 cm, demonstrating the proof-of-concept of our proposed EEG-based approach. A mean error distance of 1.32 ± 0.15 cm was measured when using only nine channels attached to the middle of the motor cortex, showing the possibility of practically using the proposed motor-hotspot-identification approach based on a relatively small number of EEG channels. CONCLUSION: We demonstrated the feasibility of our novel EEG-based motor-hotspot-identification method. It is expected that our approach can be used as an alternative to TMS for motor hotspot identification. In particular, its usability would significantly increase when using a recently developed portable tES device integrated with an EEG device.

The relationship between visit-to-visit variability in blood pressure and incidence of metabolic syndrome: a general population-based cohort study in Korea.

BACKGROUND: Previous studies demonstrated that visit-to-visit variability of blood pressure (BP) has significant relationship with cardiovascular disease. Visit-to-visit variability in BP might have prognostic value for cardiovascular disease. The aim of this study is to evaluate the effect of visit-to-visit variability in BP on development of metabolic syndrome in general population without cardiovascular disease, diabetes mellitus, metabolic syndrome, and BP medication. METHOD: We used data from the Korean Genome Epidemiology Study conducted by the Korean Centers for Disease Control and Prevention. All cohorts who were followed first 3 periods formed the basis of the study sample, which consisted of 7195 people. Of these samples, 3431 subjects who had cardiovascular disease, diabetes mellitus, or metabolic syndrome were excluded, and 312 subjects who were using antihypertensive medication in first 3 periods were excluded. Our final study sample consisted of 3452 cohorts. RESULTS: The mean age was 53.5 (8.25) years. The proportion of male was 50.2%. Average follow-up duration was 5.91 (0.17) years. In generalized estimating equation, the development of metabolic syndrome was associated with mean systolic BP (SBP) (Odd ratio (OR) 1.042, 95% confidence interval (CI) 1.035-1.048, p < 0.001), mean diastolic BP (DBP) (OR 1.058, 95% CI 1.049-1.069, p < 0.001), standard deviation (SD) of SBP (OR 1.036, 95% CI 1.017-1.055, p < 0.001), SD of DBP (OR 1.053, 95% CI 1.027-1.080, p < 0.001), and coefficient of variation (CV) of DBP (OR 1.025, 95% CI 1.005-1.046, p = 0.016) after adjusted for age, sex, and metabolic syndrome component. When mean SBP, mean DBP, SBP variability, and DBP variability were entered all together in the analysis model, SD of DBP (OR 1.033, 95% CI 1.003-1.063, p = 0.030) and CV of DBP (OR 1.027, 95% CI 1.004-1.051, p = 0.020) were significantly associated with the development of metabolic syndrome. CONCLUSION: In general population without cardiovascular disease, diabetes mellitus, metabolic syndrome, and BP medication, SD of DBP and CV of DBP was associated with the development of metabolic syndrome. Visit-to-visit variability in DBP might be helpful for the prediction of future metabolic syndrome development.

Mediating effects of nocturnal blood pressure and morning surge on the contributions of arterial stiffness and sodium intake to morning blood pressure: A path analysis.

BACKGROUND: The aim of this study was to evaluate the mediating effects of nocturnal blood pressure (BP) and morning surge on the contribution of arterial stiffness and sodium intake to morning BP in a middle-aged general population. METHODS: The study included 124 subjects aged 30-59 years, from rural Yeoju County, Gyeonggi Province, South Korea. Ambulatory BP monitoring, 24 h urinary sodium excretion (24 h UNa) and brachial-ankle pulse wave velocity (baPWV) measurements were performed in all subjects. RESULTS: The mean ± SD age was 48.1 ± 8.2 years and the proportion of male subjects was 41.1%. After adjusting for covariates, morning systolic blood pressure (SBP) was significantly correlated with morning surge [coefficient = 0.761, 95% confidence interval (CI) 0.651-0.872, p < 0.001], nocturnal SBP (coefficient = 0.576, 95% CI 0.454-0.698, p < 0.001) and baPWV (coefficient = 3.550, 95% CI 1.447-5.653, p = 0.001). Path analysis modeling revealed that baPWV had significant direct (0.035, p = 0.013) and indirect effects (0.042, p = 0.012) on morning SBP. The indirect effect mediated by nocturnal SBP was statistically significant (0.054, p = 0.005), but the indirect effect mediated by the morning surge was not significant. The 24 h UNa had no significant direct or indirect effects on morning SBP. CONCLUSION: baPWV had significant direct and indirect effects on morning SBP. The indirect effect was mediated by nocturnal SBP, but not by morning surge. The 24 h UNa had neither significant direct nor indirect effects on morning SBP.

Defining Vitamin D Deficiency and Its Relationship to Hypertension in Postmenopausal Korean Women.

BACKGROUND: Inconsistent results regarding the association between low vitamin D level and hypertension (HTN) have led to uncertainty in clinical practice as to the use of vitamin D cutoff value. The aims of this study were to investigate the association between vitamin D deficiency and HTN in postmenopausal women and to determine the clinically significant cutoff value for vitamin D deficiency. METHODS: We analyzed data from the Korean National Health and Nutrition Examination Survey 5 database (4107 postmenopausal women, aged 50-79 years). The cutoff value for vitamin D deficiency was based on the changes in parathyroid hormone (PTH) level according to serum 25(OH)D value. RESULTS: PTH levels were significantly higher in groups with serum 25(OH)D <15 ng/mL (p < 0.001). When a serum 25(OH)D value of 15 ng/mL was used as a cutoff value, 35.2% of the subjects were found to be vitamin D deficient. Although the association between 25(OH)D level <15 ng/mL and systolic blood pressure did not remain significant after adjusting for season (p = 0.30), multiple logistic regression analysis revealed that 25(OH)D level <15 ng/mL was an independent risk factor for HTN (adjusted odds ratio [OR] 1.285, 95% confidence interval [CI] 1.024-1.614, p = 0.031). CONCLUSIONS: Using serum PTH level, we defined vitamin D deficiency as 25(OH)D <15 ng/mL in postmenopausal women. Vitamin D deficiency [25(OH)D <15 ng/mL] was a significant risk factor for HTN in postmenopausal women.

The relationship between bone mineral density and blood pressure in the Korean elderly population: the Korea National Health and Nutrition Examination Survey, 2008-2011.

It is not clear whether the inverse relationship between bone mineral density (BMD) and blood pressure (BP) could be generalizable to the general elderly population. We used data from the fourth and fifth Korea National Health and Nutrition Examination Survey. The study sample consisted of 8439 men and postmenopausal women aged 50 years and older. We evaluated the relationship between BMD and BP. When adjusted for covariates, femur neck T-score [coefficient = -0.391, 95% confidence interval (CI) -0.766 to -0.016, p = 0.041] had an inverse relationship with diastolic BP (DBP), whereas lumbar spine BMD (coefficient = 0.395, 95% CI 0.058-7.752, p = 0.047) and T-score (coefficient = 0.458, 95% CI 0.005-0.911, p = 0.047) had a positive relationship with systolic BP (SBP). When adjusted for confounding factors, SBP (128.67 ± 0.979 mmHg versus 126.36 ± 0.545 mmHg, p = 0.026) and DBP (78.8 ± 0.622 mmHg versus 77.27 ± 0.283 mmHg, p = 0.016) were significantly higher in femur neck osteoporosis subjects than non-osteoporosis subjects. However, there were no differences in BP in relation to lumbar spine osteoporosis. Femur neck osteoporosis (odds ratio = 1.422, 95% CI 1.107-1.827, p = 0.006) had a significant and positive relationship with hypertension, whereas the other parameters of BMD were not significantly related to hypertension. In conclusion, higher BP and hypertension were significantly and positively correlated with femur neck osteoporosis in men and postmenopausal women aged 50 years and older.

Relationship between bone mineral density and a 10-year risk for coronary artery disease in a healthy Korean population: the Korea National Health and Nutrition Examination Survey 2008-2010.

INTRODUCTION: Bone mineral density (BMD) is associated with atherosclerosis and vascular calcification. If BMD is related independently to the risk of coronary heart disease (CHD), BMD could play an important role in CHD risk prediction. We assessed the hypothesis that BMD is related independently to the risk of CHD. MATERIALS AND METHODS: We used data from the fourth and fifth Korea National Health and Nutrition Examination Surveys. The study sample included men and women aged 20-79 years, who did not have myocardial infarction, angina pectoris, stroke, or diabetes mellitus. We evaluated the relationship between BMD and the Framingham risk score for each sex. RESULTS: In the male population, femur neck BMD [coefficient=-2.167, 95% confidence interval (CI) -3.385 to -0.950, P=0.001] and lumbar spine BMD (coefficient=-1.539, 95% CI -2.546 to -0.532, P=0.003) showed an inverse correlation with the Framingham risk score after adjusting for covariates. In the female population, the relationship between BMD parameters and the Framingham risk score was not significant after adjusting for covariates. In the male population, those with femur neck BMD values in the first quartile had greater odds of a 10-year risk greater than or equal to 10% for CHD compared with those in the fourth quartile (odds ratio=1.942, 95% CI 1.315-2.869, P<0.001). CONCLUSION: BMD was correlated inversely with the 10-year risk for CHD in the healthy male population. This result suggests that in the male population, measurement of BMD could be useful for prediction of the risk of CHD.

Health-related quality of life in coronary heart disease in Korea: the Korea National Health and Nutrition Examination Survey 2007 to 2011.

Using data from 2007 to 2011 of the Korea National Health and Nutrition Examination Survey, we evaluated the influence of coronary heart disease (CHD) on health-related quality of life (HRQoL) as measured by the EQ-5D in comparison with the general population and the predictors of HRQoL in CHD. Compared with the general population, HRQoL was impaired in the EQ-5D dimensions of mobility, usual activities, pain/discomfort, and anxiety/depression. The impairment of HRQoL was much greater in the older age group and in females. In subjects with CHD, the predictors for a low EQ-5D index were old age, female sex, low education, stroke, and noncardiovascular comorbidities, and the predictors for a low EQ Visual Analogue Scale were low income and noncardiovascular comorbidities. For the improvement in HRQoL, preventing stroke and noncardiovascular comorbidities is important, especially among female and older Asian patients with CHD.

Biatrial myxoma and multiple organ infarctions combined with Leriche syndrome in a female patient.

BACKGROUND: Multiple organ infarctions combined with Leriche syndrome due to embolic particles of myxoma are very rare. There is no definite guideline for immediate medical treatment. CASE PRESENTATION: A 36-year-old married female was referred to the emergency department (ED) with severe pain of both lower extremities and gradual decreased mental status. Brain magnetic resonance imaging (MRI) and computed tomography angiography (CTA) revealed acute multiple organ infarctions including the brain, spleen, and bilateral kidneys combined with Leriche syndrome. To evaluate the embolic source, echocardiography was performed and it revealed biatrial myxoma. Because of the risk of progression in systemic embolic events, surgical excision and embolectomy were performed urgently. After the operation, renal function was recovered, and the pain of both limbs was relieved. However, the visual field defect due to the brain infarction remained. She was discharged uneventfully on the fourteenth postoperative day. CONCLUSION: This was an extremely rare case of multiple organ infarctions combined with Leriche syndrome as the initial presentation of biatrial myxoma. The treatment of choice for myxoma is surgical excision, but the optimal timing of operations is still controversial in patients who have had recent neurological insults. Echocardiography was useful to clarify the diagnosis and decide on the proper treatment modality: surgical treatment or thrombolysis.

The relationship between ambulatory arterial stiffness index and blood pressure variability in hypertensive patients.

BACKGROUND AND OBJECTIVES: Ambulatory arterial stiffness index (AASI) is well known as a predictor of cardiovascular mortality in hypertensive patients. Mathematically, AASI reflect the standard deviation (SD) of blood pressure (BP) variation. AASI is measured higher levels in non-dipper than dipper. Thus, AASI has a possibility of not only reflecting arterial stiffness but also BP variability and/or autonomic nervous dysfunction. SUBJECTS AND METHODS: Consecutive data from 418 untreated hypertensive patients were analyzed retrospectively. We examined the association between the 24-hour ambulatory BP monitoring (ABPM) parameters and AASI. RESULTS: AASI had a simple correlation with age (R=0.189, p<0.001), relative wall thickness (RWT) (R=0.115, p=0.019), left ventricular mass index (LVMI) (R=0.192, p<0.001), average systolic BP (SBP) (R=0.232, p<0.001), average pulse pressure (PP) (R=0.363, p<0.001), SD of diastolic BP (DBP) (R=-0.352, p<0.001), SD of PP (R=0.330, p<0.001), SD of heart rate (HR) (R=-0.268, p<0.001), and nocturnal dipping (R=-0.137, p=0.005). In multiple linear regression analysis model including clinical parameters and 24 hour-ABPM parameters, independent predictors of AASI were SD of PP (ß=1.246, p<0.001), SD of DBP (ß=-1.067, p<0.001), SD of SBP (ß=-0.197, p<0.001), and non-dipper (ß=0.054, p=0.033). CONCLUSION: AASI is closely correlated with BP variability. The result of this study shows that AASI is not only a parameter for arterial stiffness, but also a parameter for BP variability.

Non-Dipper Pattern is a Determinant of the Inappropriateness of Left Ventricular Mass in Essential Hypertensive Patients.

BACKGROUND AND OBJECTIVES: Inappropriately high left ventricular mass (iLVM) is known to be related to cardiovascular prognosis. A non-dipper pattern has a greater mean left ventricular (LV) mass than the dipper pattern in hypertensive patients. However, the appropriateness of LV mass in dipper or non-dipper patterns has not been adequately investigated. The aim of this study was to define the relationship between nocturnal dipping and the appropriateness of LV mass. SUBJECTS AND METHODS: Using the ambulatory blood pressure monitoring (ABPM) database, the data of 361 patients who underwent ABPM and echocardiography was analyzed retrospectively. Appropriateness of LV mass was calculated as observed/predicted ratio of LV mass (OPR) using a Korean-specified equation. Nocturnal dipping was expressed as percent fall in systolic blood pressure (BP) during the night compared to the day. RESULTS: Daytime, nighttime and 24 hours BP in hypertensive patients was 140.4±14.8 mmHg, 143.7±15.2 mmHg and 129.4±20.0 mmHg, respectively. OPR was 106.3±19.9% and nocturnal dipping was 10.2±10.9 mmHg. In a multiple linear regression model, 24 hours systolic BP (ß=0.097, p=0.043) and nocturnal dipping (ß=-0.098, p=0.046) were independent determinants of OPR as well as age (ß=0.130, p=0.025) and body mass index (BMI) (ß=0.363, p<0.001). Odds ratio of the non-dipper pattern was 2.134 for iLVM (p=0.021) and 3.694 for obesity (p<0.001; BMI >25 kg/m(2)). CONCLUSION: The non-dipper pattern is independently associated with iLVM in hypertensive patients as well as obesity.

The relationship between coronary artery calcification and bone mineral density in patients according to their metabolic syndrome status.

BACKGROUND AND OBJECTIVES: The extent of coronary artery calcification (CAC) is closely related to total atherosclerotic plaque burden. However, the pathogenesis of CAC is still unclear. Conditions such as diabetes mellitus, renal failure, smoking, and chronic inflammation have been suggested to link vascular calcification and bone loss. In the present study, we hypothesized that bone loss can contribute to the pathogenesis of CAC in patients with the chronic inflammatory condition that accompanies metabolic syndrome (MetS). The objective of this study was to investigate the relationship between CAC and bone mineral density (BMD) in patients with MetS and in patients without MetS, by using coronary multidetector-row computed tomography (MDCT). SUBJECTS AND METHODS: Data from 395 consecutive patients was analyzed retrospectively. From the MDCT database, only those patients who underwent both coronary MDCT and dual-energy X-ray absorptiometry within an interval of one month, were selected. The presence of MetS was determined by the updated criteria as defined by the Third Adult Treatment Panel Report of the National Cholesterol Education Program. RESULTS: In patients with MetS, a significant correlation was found between CAC and age {odds ratio (OR)=1.139, 95% confidence interval (CI) 1.080 to 1.201, p<0.001}, CAC and male sex (OR=3.762, 95% CI 1.339 to 10.569, p=0.012), and CAC and T-score of L-spine (OR=0.740, 95% CI 0.550 to 0.996, p=0.047) using a forward multiple logistic regression analysis model including clinical variables of gender, age, lipid profile, body mass index, diabetes mellitus, hypertension, smoking, and BMD. But in patients without MetS, BMD by itself was not found to contribute to CAC. CONCLUSION: BMD was inversely correlated with CAC only in patients with MetS. This finding suggests that low BMD accompanied by MetS, may have significant clinical implications.

Poly(D,L-lactic acid-co-glycolic acid)-b-poly(ethylene glycol)-b-poly (D,L-lactic acid-co-glycolic acid) triblock copolymer and thermoreversible phase transition in water.

Novel thermoreversible gelation behavior of aqueous solutions of ABA-type triblock copolymers composed of the central polyethylene oxide (PEG) block and two poly(D,L-lactic acid-co-glycolic acid) side blocks was found. Phase transition characteristics, such as critical gel concentration (CGC) and lower and upper critical gel temperature (CGT), are closely related to the molecular structure of the triblock copolymers. The CGC and the lower CGT both increases with increasing PEG/PLGA molecular weight ratio. Increasing the GA content in PLGA block induces a somewhat higher CGC. The copolymer forms micelles with a PLGA loop core and a PEG shell in water. Also grouped micelles are identified seemingly due to the bridging of two micelles sharing two PLGA blocks of a block copolymer chain. As the temperature increases the association of micelles increases, which results in gelation. The ABA-type copolymers exhibit a relatively low CGC (<10%) and low sol-gel transition temperatures compared to BAB-type copolymers. As the temperature increases further gel-sol transition is observed, which would result from the shrinkage of micelles with temperature increase. The hydrodynamic size of the micelles is monitored by dynamic laser scattering, and a possible gelation mechanism was suggested.