Huge Sample Sizes Research Articles

Role of nutrition in orthopedic trauma and surgery-A prospective study

: Nutrition is an age-old issue of human civilization with approximately 40% undernourished hospitalized inpatients. This has effect on surgery as healthy individuals have better tolerance than malnourished patients. : This study was designed to evaluate the effect of nutrition on patients who had undergone different types of Orthopedics surgeries. This prospective research involved 200 participants who received various orthopedic surgeries at a tertiary care facility. The evaluation of nutritional status involved measuring anthropometric (BMI) and biochemical markers (pre-albumin and transferrin) before, after, and three months after surgery for various age groups, genders, comorbidities like diabetes and hypertension, as well as for complications like urinary tract infections and surgical wound dehiscence. Version 16 of the SPSS statistical analysis program was used for the analysis. Chi Square and the repeated variable ANNOVA test were the tests employed.: 113 patients were followed up prospectively with 4.42% under nourishment rate. Nutritional markers were statistically significant values at pre,post and at 3 months of post op. Diabetes and age more than 40 years were two important parameters affecting nutrition of patients undergoing different orthopedic surgeries. Surgical wound infection was not related to nutritional status of the patient in this study. Pre albumin was found to be a better predictor of nutrition than transferrin.: This study clearly showed that nutrition has tremendous impact on the outcome different orthopedics surgeries. But needs more extensive research with huge sample size to prove the same in a more accurate way.

P-087 Obstetrical and perinatal outcomes from intracytoplasmic sperm injection couples with oocyte donation are not affected by advanced paternal age measured as a continuous variable

Abstract Study question Does paternal age (PA), considered a continuous variable, have any effect on obstetrical and perinatal outcomes in intracytoplasmic sperm injection (ICSI) with donated oocytes? Summary answer PA does not affect clinically relevant obstetrical and perinatal outcomes, but it significantly increases the pregnancy’s duration in ICSI with autologous sperm and donated oocytes. What is known already Currently there has been an evident delay in childbearing, therefore studies relating advanced PA with reproductive risks are increasing and results are controversial. These studies frequently study PA by dividing it in groups arbitrarily constructed, but probably a more precise evaluation is obtained using PA as a continuous variable. Furthermore, oocyte donation standardizes female factor, to some extent, allowing a more realistic male factor evaluation. In this regard, we aimed to evaluate the effect of PA at conception (continuous variable) on obstetrical and perinatal outcomes in ICSI couples with autologous sperm and donated oocytes in a large population of patients. Study design, size, duration This retrospective observational multicentric cohort study evaluated data from couples that performed an ICSI between January 2008 and March 2020 using autologous sperm and donated oocytes. We included only singleton deliveries and the first delivery of each couple. PA was analysed as a continuous variable, considered more informative in comparison to its analysis as categorical. Data available and included consisted of 15,696 patients who had a delivery with or without a live birth. Participants/materials, setting, methods PA ranged from 21-54 years. Clinical database was constructed with reproductive outcomes from electronic medical records. Erroneous or incomplete data were removed. Statistical analysis included a descriptive analysis, followed by univariate and multivariate models using binary logistic regression or linear regression. We adjusted for maternal age and body mass index, PA, fresh sperm concentration and progressive motility, cycle type, gestational age, transfer on day 5 and delivery type (when appropriate). P < 0.05 was considered statistically significant. Main results and the role of chance Mean PA was 41.45 (41.38-41.51). We included a total of 15,696 deliveries and 15,673 live births. Results from multivariate models are presented as adjusted odd ratio or regression coefficient with 95% confidence intervals and adjusted p. In terms of obstetrical outcomes, PA results were 1.086 (0.959-1.229) p = 0.194 for gestational diabetes, 0.963 (0.849-1.093) p = 0.561 for anemia, 0.947 (0.837-1.072) p = 0.388 for hypertension, 0.959 (0.878-1.048) p = 0.356 for preterm birth and 0.986 (0.929-1.046) p = 0.641 for caesarean delivery. Regarding perinatal outcomes, PA results were 0.442 (0.005-0.879) p = 0.047 for gestational age, -0.216 (-15.559-15.126) p = 0.978 for infant weight, 1.101 (0.941-1.288) p = 0.231 for low birth weight (<2,500 g), 0.710 (0.463-1.088) p = 0.116 for very low birth weight (<1,500g), 0.988 (0.933-1.047) p = 0.684 for female sex, 0.002 (-0.076-0.08) p = 0.96 for length, -0.012 (-0.088-0.064) p = 0.763 for cranial perimeter, 0.991 (0.876-1.12) p = 0.88 for NICU admission, and 0.027 (-0.028-0.083) p = 0.337, 0.025 (-0.012-0.061) p = 0.18 and -0.045 (-0.096-0.007) p = 0.089 for Apgar 1, 5 and 10 minutes, respectively. According to our results, PA does not affect clinically relevant outcomes for the health during pregnancy and of the infant. However, PA significantly increases the duration of the pregnancy in ICSI with autologous sperm and donated oocytes. Limitations, reasons for caution The study’s major strengths are the consideration of PA as a continuous variable, the use of oocyte donation (controlling female factors) and the inclusion of a huge sample size. However, due to its retrospective nature there are some biases derived from the clinical practice and the presence of missing data. Wider implications of the findings Current evident delay in childbearing merits further research on the effect of PA on reproductive outcomes, addressed in this study. Our study suggests that PA is not associated with obstetrical and perinatal outcomes, except for gestational age, which sends a hopeful message to fathers of advanced paternal age. Trial registration number not applicable

Soleus Muscle H-reflex: Reference Data of Adult Population From a Tertiary Care Center in India.

Introduction The Hoffmann reflex (H reflex) is one of the most studied reflexes in human neurophysiological literature. Detection of the H reflex is useful in the diagnosis of early polyneuropathy, S1 radiculopathy, early GBS, tibial neuropathy and sciatica, and sacral plexopathy. The H reflex is also used as a tool to measure the excitability of the nervous components of the arc, regardless of the sensitivity of the sensory organs. The monosynaptic nature of reflex circuits makes H-reflex an attractive tool for clinical neurophysiology and research. Objective The objective is to create reference data of soleus H-reflex latency in an adult population from a tertiary care center in India. Materials and methods Seven hundred eighty-four healthy volunteers underwent a physical examination and brief electrophysiological examination before elicitation of the soleus H reflex of both lower extremities using standard techniques. Reference values ​​are expressed as mean ± standard deviation as well as the third and 97th percentiles for latency as the dependent variable. Results The study population included 346 (44.1%) women and 438 (55.9%) men. The men were aged 40.46 ± 14.76 years, and the women were aged 41.63 ± 13.49 years. The average weight of the men was 73.32 ± 10.28 kilograms, and the women were 62.91 ± 7.46 kilograms. The average height of the men was 172.06 ± 4.22 cm, and the women were 159.12 ± 2.42 cm. The third and 97th percentiles for H-reflex latency on the right side were 22.86 ms to 34.22 ms and on the left side were 22.86 ms to 35.39 ms. The average right tibial H latency and left tibial H latency were 28.18 ± 2.59 ms and 28.14 ± 2.70 ms, respectively. Conclusion A sizable subject population was used to provide reference data for this study. Because of the huge sample size and nearly appropriate coverage of different age groups, reference ranges have been established for various age, height, and BMI groups.

Hypothesis testing of one sample mean vector in distributed frameworks

Distributed frameworks are commonly used in the setting where data are stored in k different local machines and cannot be merged due to privacy protections or the huge sample size. For a random vector X ∈ R p with expectation μ , testing the mean vector H 0 : μ = μ 0 vs H 1 : μ ≠ μ 0 for a given vector μ 0 is a basic problem in statistics. In distributed frameworks, the computation of the centralized test statistics is not privacy-preserving and often requires heavy communication costs, which can be a burden when p or k is large. To deal with this problem, we extend two commonly used centralized test statistics to the distributed ones based on the divide and conquer technique. It is observed that the proposed test statistics are effective and can reduce communication costs and computation complexity. Numerical results confirm the theoretical findings.

Transkingdom Network Analysis (TkNA): a systems framework for inferring causal factors underlying host-microbiota and other multi-omic interactions.

We present Transkingdom Network Analysis (TkNA), a unique causal-inference analytical framework that offers a holistic view of biological systems by integrating data from multiple cohorts and diverse omics types. TkNA helps to decipher key players and mechanisms governing host-microbiota (or any multi-omic data) interactions in specific conditions or diseases. TkNA reconstructs a network that represents a statistical model capturing the complex relationships between different omics in the biological system. It identifies robust and reproducible patterns of fold change direction and correlation sign across several cohorts to select differential features and their per-group correlations. The framework then uses causality-sensitive metrics, statistical thresholds and topological criteria to determine the final edges forming the transkingdom network. With the subsequent network's topological features, TkNA identifies nodes controlling a given subnetwork or governing communication between kingdoms and/or subnetworks. The computational time for the millions of correlations necessary for network reconstruction in TkNA typically takes only a few minutes, varying with the study design. Unlike most other multi-omics approaches that find only associations, TkNA focuses on establishing causality while accounting for the complex structure of multi-omic data. It achieves this without requiring huge sample sizes. Moreover, the TkNA protocol is user friendly, requiring minimal installation and basic familiarity with Unix. Researchers can access the TkNA software at https://github.com/CAnBioNet/TkNA/ .

Feature Screening with Conditional Rank Utility for Big-Data Classification

Feature screening is a commonly used strategy to eliminate irrelevant features in high-dimensional classification. When one encounters big datasets with both high dimensionality and huge sample size, the conventional screening methods become computationally costly or even infeasible. In this article, we introduce a novel screening utility, Conditional Rank Utility (CRU), and propose a distributed feature screening procedure for the big-data classification. The proposed CRU effectively quantifies the significance of a numerical feature on the categorical response. Since CRU is constructed based on the ratio of the mean conditional rank to the mean unconditional rank of a feature, it is robust against model misspecification and the presence of outliers. Structurally, CRU can be expressed as a simple function of a few component parameters, each of which can be distributively estimated using a natural unbiased estimator from the data segments. Under mild conditions, we show that the distributed estimator of CRU is fully efficient in terms of the probability convergence bound and the mean squared error rate; the corresponding distributed screening procedure enjoys the sure screening and ranking properties. The promising performances of the CRU-based screening are supported by extensive numerical examples. Supplementary materials for this article are available online.

Quantile regression for massive data set

Traditional statistical analysis is challenged by modern massive data sets, which have huge sample size and dimension. Quantile regression has become a popular alternative to least squares method for providing comprehensive description of the response distribution and robustness against heavy-tailed error distributions. On the other hand, non-smooth quantile loss poses a new challenge to massive data sets. To address the problem, we transform the non-differentiable quantile loss function into a convex quadratic loss function based on Expectation-maximization (EM) algorithm using an asymmetric Laplace distribution. Both simulations and real data application are conducted to illustrate the performance of the proposed methods.

Hyoid Bone Fractures in Mechanical Asphyxiation Deaths: A Study at Osmania General Hospital, Hyderabad, India

The fractures of the hyoid bone in the neck are one of the important indicators of any pressure over the neck. Thepresent study includes a huge sample size of mechanical asphyxiation deaths (791 cases) to study the incidence ofhyoid bone fractures in hanging, ligature strangulation, and throttling. Fractures of the hyoid bone were noticed ina total of 12 cases, 9 male and 3 female with 33.33% incidence in the 41-50 years age group. Throttling dominatedthe strangulation group in our study with a share of 42.85%. A total of 6 (5 male, 1 female) cases showed hyoidbone fractures in the strangulation group wherein the predominant incidence is seen in 41-50 years(50%). In ourstudy, right cornu fractures are frequent in hanging and throttling. Except for one outward compression fracturein a hanging instance, all of the fractures identified in our investigation are inward compression fractures. Thepresent study adds to the existing literature that hyoid bone fractures are less common in hanging than throttlingand any fracture of hyoid bone is common after the age of 40 years because of calcification.

Block-Fading Non-Stationary Channel Estimation for MIMO-OFDM Systems via Meta-Learning

Deep learning (DL)-based channel estimations for multiple-input-multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) systems have shown remarkable performance at the cost of huge sample size and complexity. While such complexity can be offloaded onto offline training phase for stationary channels, this becomes problematic when non-stationary channels are arisen. In this letter, we resolve this issue by proposing meta-learning-aided online training that only requires small sample size with reduced complexity. Numerical results under 3GPP channel models verify that proposed meta-learning approach outperforms not only conventional DL-based estimators but also conventional model-based estimators, e.g., least squares and linear minimum mean square error estimators, especially in the small sample size/low complexity regime.

Evaluation of Liver Function Tests in Normal Pregnancy-An Observational Epidemiological Study-Kumbakonam Urban Rural Epidemiological Study-KURES 10

Early interpretation of liver function tests (LFTs) can result in timely management and it may decrease the incidence of complications in both the mother and the foetus. Normal LFTs don't always imply a normal liver. A number of drawbacks can occur when interpreting basic blood LFTs. Abnormal Liver Function Tests (LFTs) in pregnancy must be properly interpreted to avoid diagnostic pitfalls. we did a routine testing in 498 asymptomatic antenatal mothers in the third trimester. Ninety-seven percentage of mothers had normal bilirubin while 93 % had normal enzyme levels. The other abnormal values were also within 1 to 1.5 % above the normal. No mother had any symptoms. Only two patients had gallstones whose LFTs were normal. Two mothers had tattoo marks while the other two were HbSAg positive. All the 498 mothers had an uneventful progress towards delivery. Cross consultation done on a few cases with abnormal values were non-contributory. Pregnancy-related disorders are the most common reason for unusual liver function tests during pregnancy, especially in the third trimester. We omitted such cases to support our hypotheses of avoiding undue testing. The most prevalent is pre-eclampsia-related disorder. This is the first such study on a huge sample size. We suggest a routine testing of LFTs is not needed in otherwise healthy antenatal mothers.

O-296 Paternal age does not affect obstetric and perinatal outcomes in IVF or ICSI cycles with autologous oocytes

Abstract Study question Does paternal age have any effect on obstetric and perinatal outcomes in in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI) cycles with autologous oocytes? Summary answer Gestational diabetes, delivery type, cranial circumference and preterm birth were significantly different between our age groups, although after adjusted analysis no significant differences were seen. What is known already In recent years, there has been an increase of the average male age at which the first child is conceived. Therefore, studies addressing paternal age and reproductive risks are increasing and results are controversial, so there is a need of more research about it, which this study tries to address. Although several studies suggest the threshold to consider paternal age ‘advanced’ at 40 years, this limit is not well established. The assessment of risks concerning the mother and offspring is important and, thus, so is the study of its association with paternal age. Study design, size, duration This retrospective observational multicentric cohort study has considered pregnancies and children conceived from IVF-ICSI performed to couples in Spain IVIRMA clinics between January 2008 and March 2020 using patients’ own sperm and autologous oocytes. Paternal age ranged from 18 to 51 years. The study population was categorized in three groups: <30(A, reference), 30-40(B) and >40(C) years. The data available included 14402 patients who had a delivery with a live birth and 16118 singleton newborns. Participants/materials, setting, methods We evaluated pregnancies and children from autologous IVF-ICSI with own semen, known age and resulting in delivery. Data was exported to obtain the clinical database followed by the statistical analysis of the study main outcomes. P < 0.05 was considered statistically significant. Obstetric and perinatal outcomes were measured including delivery type, incidence of preterm birth, gestational diabetes, anemia, hypertension and pre-eclampsia; and newborn sex, weight, length, cranial circumference, Apgar score and neonatal intensive care unit (NICU) admission. Main results and the role of chance Results are expressed as rate or mean with 95%CI. There were no statistically significant differences for the incidence of anemia, hypertension or pre-eclampsia; newborn sex, weight, length, Apgar score (1, 5 and 10 min) and NICU admission between the reference younger group (A) vs. the older paternal age groups (B and C). However, there were statistically significant differences for the incidence of gestational diabetes when comparing A 5.1%(3.5-7.2) with B 7.7%(7.1-8.3) with OR = 1.5(1.1-2.2)(p = 0.021) and with C 9.1%(7.1-11.4) with OR = 1.9(1.2-2.9)(p = 0.005); for delivery type when comparing A[cesarean 33.8%(31.3-36.4)] with B[cesarean 37.8%(37.0-38.7)] with OR = 1.2(1.1-1.3) (p = 0.004) and with C[cesarean 49.3%(46.7-52.0)] with OR = 1.9(1.6-2.2) (p < 0.001); for cranial circumference when comparing A 34.5(34.3-34.7) with C 34.8(34.6-35.0) with p = 0.03; and for the incidence of preterm birth when comparing A 7.9%(6.6-9.5) with C 5.9%(4.7-7.2) with OR = 0.7(0.5-1.0)(p = 0.034). These results suggest an increased risk of gestational diabetes and cesarean delivery as paternal age increases. Moreover, group C showed a larger cranial perimeter and a lower risk of preterm birth when compared to group A. However, after adjusted analysis (according to selected patients’, cycles’ and semen samples’ characteristics, gestational age, newborn sex and type of delivery) these differences were no longer statistically significant, so probably other clinical variables are involved in the differences seen. Limitations, reasons for caution Despite the retrospective nature of this study that leads to biases derived from the clinical practice and to the presence of missing data limiting sample size, the study includes a huge sample size. Wider implications of the findings Currently, fatherhood delay has led to a growing interest in studies addressing paternal age and reproductive risks. Our paternal age comparison showed significant differences for gestational diabetes, delivery type, cranial circumference and preterm, no longer significant after adjusted analysis, so probably other clinical variables were involved. Further studies are needed. Trial registration number Not applicable

Stacked deep analytic model for human activity recognition on a UCI HAR database.

Background Owing to low cost and ubiquity, human activity recognition using smartphones is emerging as a trendy mobile application in diverse appliances such as assisted living, healthcare monitoring, etc. Analysing this one-dimensional time-series signal is rather challenging due to its spatial and temporal variances. Numerous deep neural networks (DNNs) are conducted to unveil deep features of complex real-world data. However, the drawback of DNNs is the un-interpretation of the network's internal logic to achieve the output. Furthermore, a huge training sample size (i.e. millions of samples) is required to ensure great performance. Methods In this work, a simpler yet effective stacked deep network, known as Stacked Discriminant Feature Learning (SDFL), is proposed to analyse inertial motion data for activity recognition. Contrary to DNNs, this deep model extracts rich features without the prerequisite of a gigantic training sample set and tenuous hyper-parameter tuning. SDFL is a stacking deep network with multiple learning modules, appearing in a serialized layout for multi-level feature learning from shallow to deeper features. In each learning module, Rayleigh coefficient optimized learning is accomplished to extort discriminant features. A subject-independent protocol is implemented where the system model (trained by data from a group of users) is used to recognize data from another group of users. Results Empirical results demonstrate that SDFL surpasses state-of-the-art methods, including DNNs like Convolutional Neural Network, Deep Belief Network, etc., with ~97% accuracy from the UCI HAR database with thousands of training samples. Additionally, the model training time of SDFL is merely a few minutes, compared with DNNs, which require hours for model training. Conclusions The supremacy of SDFL is corroborated in analysing motion data for human activity recognition requiring no GPU but only a CPU with a fast- learning rate.

Stacked deep analytic model for human activity recognition on a UCI HAR database

Background Owing to low cost and ubiquity, human activity recognition using smartphones is emerging as a trendy mobile application in diverse appliances such as assisted living, healthcare monitoring, etc. Analysing this one-dimensional time-series signal is rather challenging due to its spatial and temporal variances. Numerous deep neural networks (DNNs) are conducted to unveil deep features of complex real-world data. However, the drawback of DNNs is the un-interpretation of the network's internal logic to achieve the output. Furthermore, a huge training sample size (i.e. millions of samples) is required to ensure great performance. Methods In this work, a simpler yet effective stacked deep network, known as Stacked Discriminant Feature Learning (SDFL), is proposed to analyse inertial motion data for activity recognition. Contrary to DNNs, this deep model extracts rich features without the prerequisite of a gigantic training sample set and tenuous hyper-parameter tuning. SDFL is a stacking deep network with multiple learning modules, appearing in a serialized layout for multi-level feature learning from shallow to deeper features. In each learning module, Rayleigh coefficient optimized learning is accomplished to extort discriminant features. A subject-independent protocol is implemented where the system model (trained by data from a group of users) is used to recognize data from another group of users. Results Empirical results demonstrate that SDFL surpasses state-of-the-art methods, including DNNs like Convolutional Neural Network, Deep Belief Network, etc., with ~97% accuracy from the UCI HAR database with thousands of training samples. Additionally, the model training time of SDFL is merely a few minutes, compared with DNNs, which require hours for model training. Conclusions The supremacy of SDFL is corroborated in analysing motion data for human activity recognition requiring no GPU but only a CPU with a fast- learning rate.

Stacked deep analytic model for human activity recognition on a UCI HAR database

Background Owing to low cost and ubiquity, human activity recognition using smartphones is emerging as a trendy mobile application in diverse appliances such as assisted living, healthcare monitoring, etc. Analysing this one-dimensional time-series signal is rather challenging due to its spatial and temporal variances. Numerous deep neural networks (DNNs) are conducted to unveil deep features of complex real-world data. However, the drawback of DNNs is the un-interpretation of the network's internal logic to achieve the output. Furthermore, a huge training sample size (i.e. millions of samples) is required to ensure great performance. Methods In this work, a simpler yet effective stacked deep network, known as Stacked Discriminant Feature Learning (SDFL), is proposed to analyse inertial motion data for activity recognition. Contrary to DNNs, this deep model extracts rich features without the prerequisite of a gigantic training sample set and tenuous hyper-parameter tuning. SDFL is a stacking deep network with multiple learning modules, appearing in a serialized layout for multi-level feature learning from shallow to deeper features. In each learning module, Rayleigh coefficient optimized learning is accomplished to extort discriminant features. A subject-independent protocol is implemented where the system model (trained by data from a group of users) is used to recognize data from another group of users. Results Empirical results demonstrate that SDFL surpasses state-of-the-art methods, including DNNs like Convolutional Neural Network, Deep Belief Network, etc., with ~97% accuracy from the UCI HAR database with thousands of training samples. Additionally, the model training time of SDFL is merely a few minutes, compared with DNNs, which require hours for model training. Conclusions The supremacy of SDFL is corroborated in analysing motion data for human activity recognition requiring no GPU but only a CPU with a fast- learning rate.

Methodology for Interactive Labeling of Patched Asphalt Pavement Images Based on U-Net Convolutional Neural Network

Image recognition based on deep learning generally demands a huge sample size for training, for which the image labeling becomes inevitably laborious and time-consuming. In the case of evaluating the pavement quality condition, many pavement distress patching images would need manual screening and labeling, meanwhile the subjectivity of the labeling personnel would greatly affect the accuracy of image labeling. In this study, in order for an accurate and efficient recognition of the pavement patching images, an interactive labeling method is proposed based on the U-Net convolutional neural network, using active learning combined with reverse and correction labeling. According to the calculation results in this paper, the sample size required by the interactive labeling is about half of the traditional labeling method for the same recognition precision. Meanwhile, the accuracy of interactive labeling method based on the mean intersection over union (mean_IOU) index is 6% higher than that of the traditional method using the same sample size and training epochs. In addition, the accuracy analysis of the noise and boundary of the prediction results shows that this method eliminates 92% of the noise in the predictions (the proportion of noise is reduced from 13.85% to 1.06%), and the image definition is improved by 14.1% in terms of the boundary gray area ratio. The interactive labeling is considered as a significantly valuable approach, as it reduces the sample size in each epoch of active learning, greatly alleviates the demand for manpower, and improves learning efficiency and accuracy.

Possible different genotypes for human papillomavirus vaccination in lower middle-income countries towards cervical cancer elimination in 2030: a cross-sectional study.

PurposeHuman papillomavirus (HPV) genotype and age distribution of HPV infection were crucial for the national vaccination and screening program planning. However, there was a limited study providing these data in the normal cervix population. This study aimed to explore the HPV genotypes profile of women with clinically normal cervix based on Visual Inspection of Acetic Acid (VIA) test.Materials and MethodsA 7-year cross-sectional study was conducted from 2012 to 2018 in private and public health care centers in Jakarta. Subjects were recruited consecutively. Data were collected by anamnesis, VIA, and HPV DNA test using the polymerase chain reaction (PCR; SPF10-DEIA-LiPA25) method. HPV genotyping procedures include DNA extraction, PCR (SPF10-DEIA-LiPA25) using the HPV XpressMatrix kit (PT KalGen DNA, East Jakarta, Indonesia), and hybridization. The IBM SPSS ver. 20.0 (IBM Corp., Armonk, NY, USA) were used to analyze the data.ResultsA total of 1,397 subjects were collected. Positive HPV-DNA tests were found in 52 subjects (3.7%); 67% were single and 33% were multiple HPV infections. HPV 52 was the most frequently detected HPV genotype, followed by HPV 39, 16, 18 74, 44, 31, 54, and 66, respectively. The highest HPV infections in this population were in the 31–40 and 41–50 years old group.ConclusionThis study suggested beneficial screening for women aged 31–50 years old. Instead of “original” nonavalent (HPV 16, 18, 6, 11, 31, 33, 45, 52, 58), the different “nonavalent” formula for HPV vaccines protecting against HPV 16, 18, 6, 11, 31, 39, 44, 52, 74 might be useful for Indonesian population. However, further multicenter studies with a huge sample size are still needed.

Two-stage feature selection for classification of gene expression data based on an improved Salp Swarm Algorithm.

Microarray technology has developed rapidly in recent years, producing a large number of ultra-high dimensional gene expression data. However, due to the huge sample size and dimension proportion of gene expression data, it is very challenging work to screen important genes from gene expression data. For small samples of high-dimensional biomedical data, this paper proposes a two-stage feature selection framework combining Wrapper, embedding and filtering to avoid the curse of dimensionality. The proposed framework uses weighted gene co-expression network (WGCNA), random forest and minimal redundancy maximal relevance (mRMR) for first stage feature selection. In the second stage, a new gene selection method based on the improved binary Salp Swarm Algorithm is proposed, which combines machine learning methods to adaptively select feature subsets suitable for classification algorithms. Finally, the classification accuracy is evaluated using six methods: lightGBM, RF, SVM, XGBoost, MLP and KNN. To verify the performance of the framework and the effectiveness of the proposed algorithm, the number of genes selected and the classification accuracy was compared with the other five intelligent optimization algorithms. The results show that the proposed framework achieves an accuracy equal to or higher than other advanced intelligent algorithms on 10 datasets, and achieves an accuracy of over 97.6% on all 10 datasets. This shows that the method proposed in this paper can solve the feature selection problem related to high-dimensional data, and the proposed framework has no data set limitation, and it can be applied to other fields involving feature selection.

Stacked deep analytic model for human activity recognition on a UCI HAR database

Background Owing to low cost and ubiquity, human activity recognition using smartphones is emerging as a trendy mobile application in diverse appliances such as assisted living, healthcare monitoring, etc. Analysing this one-dimensional time-series signal is rather challenging due to its spatial and temporal variances. Numerous deep neural networks (DNNs) are conducted to unveil deep features of complex real-world data. However, the drawback of DNNs is the un-interpretation of the network's internal logic to achieve the output. Furthermore, a huge training sample size (i.e. millions of samples) is required to ensure great performance. Methods In this work, a simpler yet effective stacked deep network, known as Stacked Discriminant Feature Learning (SDFL), is proposed to analyse inertial motion data for activity recognition. Contrary to DNNs, this deep model extracts rich features without the prerequisite of a gigantic training sample set and tenuous hyper-parameter tuning. SDFL is a stacking deep network with multiple learning modules, appearing in a serialized layout for multi-level feature learning from shallow to deeper features. In each learning module, Rayleigh coefficient optimized learning is accomplished to extort discriminant features. A subject-independent protocol is implemented where the system model (trained by data from a group of users) is used to recognize data from another group of users. Results Empirical results demonstrate that SDFL surpasses state-of-the-art methods, including DNNs like Convolutional Neural Network, Deep Belief Network, etc., with ~97% accuracy from the UCI HAR database with thousands of training samples. Additionally, the model training time of SDFL is merely a few minutes, compared with DNNs, which require hours for model training. Conclusions The supremacy of SDFL is corroborated in analysing motion data for human activity recognition requiring no GPU but only a CPU with a fast- learning rate.

EDITORIAL COMMENT

In this interesting study, an automatically calculated electronic comorbidity measure, the Care Assessment Needs (CAN) score, was studied as predictor of life expectancy in a huge Veterans Health Administration (VHA) patient sample. 1 N. N. et al. Using an automated electronic health record score to estimate life expectancy in men diagnosed with prostate cancer in the Veterans Health Administration. Urology. 2021; ([in press]) Google Scholar The score was originally designed to estimate short-term risks of hospitalization and mortality. The authors demonstrated a strong association of the CAN score with medium-term mortality after prostate cancer diagnosis. 1 N. N. et al. Using an automated electronic health record score to estimate life expectancy in men diagnosed with prostate cancer in the Veterans Health Administration. Urology. 2021; ([in press]) Google Scholar Although this instrument may be automatically available in VHA patients, its high complexity may hinder its routine use in different institutions. Instruments predicting life expectancy to support treatment selection in men with early prostate cancer should not only be accurate but also simple to use. 2 Daskivich TJ Kwan L Dash A et al. Weighted versus unweighted Charlson score to predict long-term other-cause mortality in men with early-stage prostate cancer. Eur Urol. 2014; 66: 1002-1009 Abstract Full Text Full Text PDF PubMed Scopus (20) Google Scholar In contrast to some contrary claims, 3 Walz J Gallina A Perrotte P et al. Clinicians are poor raters of life-expectancy before radical prostatectomy or definitive radiotherapy for localized prostate cancer. BJU Int. 2007; 100: 1254-1258 Crossref PubMed Scopus (113) Google Scholar at least in candidates for radical prostatectomy, clinicians are probably able raters of life expectancy. In our radical prostatectomy series, even patients with the highest risk of competing mortality (age-adjusted Charlson score of 6 or higher, comprising only 4% of the sample) had a 10-year non-prostate cancer mortality rate of only 40% 4 Froehner M. Patient reported vs claims based measures of health for modeling life expectancy in men with prostate cancer. Letter. J Urol. 2021; 205: 1235-1236 Crossref Scopus (1) Google Scholar and the 50% level was not reached before narrowly 14 years. Complex tools requiring comprehensive data collection will probably not change routine decision making in candidates for radical prostatectomy regardless of their accuracy. If the risk classifications were, however, automatically calculated and readily available, there may be a different situation. Because of the huge sample size and the detailed information collected, the impressive data source used in this study 1 N. N. et al. Using an automated electronic health record score to estimate life expectancy in men diagnosed with prostate cancer in the Veterans Health Administration. Urology. 2021; ([in press]) Google Scholar should encourage further studies, for instance predicting 90-day mortality in vulnerable candidates for radical cystectomy.

Joint super-resolution and synthesis of 1 mm isotropic MP-RAGE volumes from clinical MRI exams with scans of different orientation, resolution and contrast

Most existing algorithms for automatic 3D morphometry of human brain MRI scans are designed for data with near-isotropic voxels at approximately 1 mm resolution, and frequently have contrast constraints as well-typically requiring T1-weighted images (e.g., MP-RAGE scans). This limitation prevents the analysis of millions of MRI scans acquired with large inter-slice spacing in clinical settings every year. In turn, the inability to quantitatively analyze these scans hinders the adoption of quantitative neuro imaging in healthcare, and also precludes research studies that could attain huge sample sizes and hence greatly improve our understanding of the human brain. Recent advances in convolutional neural networks (CNNs) are producing outstanding results in super-resolution and contrast synthesis of MRI. However, these approaches are very sensitive to the specific combination of contrast, resolution and orientation of the input images, and thus do not generalize to diverse clinical acquisition protocols – even within sites. In this article, we present SynthSR, a method to train a CNN that receives one or more scans with spaced slices, acquired with different contrast, resolution and orientation, and produces an isotropic scan of canonical contrast (typically a 1 mm MP-RAGE). The presented method does not require any preprocessing, beyond rigid coregistration of the input scans. Crucially, SynthSR trains on synthetic input images generated from 3D segmentations, and can thus be used to train CNNs for any combination of contrasts, resolutions and orientations without high-resolution real images of the input contrasts. We test the images generated with SynthSR in an array of common downstream analyses, and show that they can be reliably used for subcortical segmentation and volumetry, image registration (e.g., for tensor-based morphometry), and, if some image quality requirements are met, even cortical thickness morphometry. The source code is publicly available at https://github.com/BBillot/SynthSR.