Categories
Nevin Manimala Statistics

Jointly Defending DeepFake Manipulation and Adversarial Attack using Decoy Mechanism

IEEE Trans Pattern Anal Mach Intell. 2023 Mar 6;PP. doi: 10.1109/TPAMI.2023.3253390. Online ahead of print.

ABSTRACT

Highly realistic imaging and video synthesis have become possible and relatively simple tasks with the rapid growth of generative adversarial networks (GANs). GAN-related applications, such as DeepFake image and video manipulation and adversarial attacks, have been used to disrupt and confound the truth in images and videos over social media. DeepFake technology aims to synthesize high visual quality image content that can mislead the human vision system, while the adversarial perturbation attempts to mislead the deep neural networks to a wrong prediction. Defense strategy becomes difficult when adversarial perturbation and DeepFake are combined. This study examined a novel deceptive mechanism based on statistical hypothesis testing against DeepFake manipulation and adversarial attacks. Firstly, a deceptive model based on two isolated sub-networks was designed to generate two-dimensional random variables with a specific distribution for detecting the DeepFake image and video. This research proposes a maximum likelihood loss for training the deceptive model with two isolated sub-networks. Afterward, a novel hypothesis was proposed for a testing scheme to detect the DeepFake video and images with a well-trained deceptive model. The comprehensive experiments demonstrated that the proposed decoy mechanism could be generalized to compressed and unseen manipulation methods for both DeepFake and attack detection.

PMID:37028044 | DOI:10.1109/TPAMI.2023.3253390

Categories
Nevin Manimala Statistics

Efficient Observer Design for Ambulatory Estimation of Body Centre of Mass Position

IEEE Trans Neural Syst Rehabil Eng. 2023 Mar 6;PP. doi: 10.1109/TNSRE.2023.3253051. Online ahead of print.

ABSTRACT

Complementary Linear Filter (CLF) is a common techinque employed for estimating the ground projection of body Centre of Mass starting from ground reaction forces. This method fuses centre of pressure position and double integration of horizontal forces, selecting best cut-off frequencies for low-pass and high-pass filters. Classical Kalman filter is a substantially equivalent approach, as both methods rely on an overall quantification of error/noise and don’t analyze its origin and time-dependence. In order to overcome such limitations, a Time-Varying Kalman Filter (TVKF) is proposed in this paper: the effect of unknown variables is directly taken into account by employing a statistical description which is obtained from experimental data. To this end, in this paper we have employed a dataset of 8 walking healthy subjects: beside supplying gait cycles at different speeds, it deals with subjects in age of development and provides a wide range of body sizes, allowing therefore to assess the observers’ behaviour under different conditions. The comparison carried out between CLF and TVKF appears to highlight several advantages of the latter method in terms of better average performance and smaller variability. Results presented in this paper suggest that a strategy which incorporates a statistical description of unknown variables and a time-varying structure can yield a more reliable observer. The demonstrated methodology sets a tool that can undergo a broader investigation to be carried out including more subjects and different walking styles.

PMID:37028027 | DOI:10.1109/TNSRE.2023.3253051

Categories
Nevin Manimala Statistics

Improved Electrical Impedance Tomography Reconstruction via a Bayesian Approach with an Anatomical Statistical Shape Model

IEEE Trans Biomed Eng. 2023 Mar 6;PP. doi: 10.1109/TBME.2023.3250650. Online ahead of print.

ABSTRACT

OBJECTIVE: electrical impedance tomography (EIT) is a promising technique for rapid and continuous bedside monitoring of lung function. Accurate and reliable EIT reconstruction of ventilation requires patient-specific shape information. However, this shape information is often not available and current EIT reconstruction methods typically have limited spatial fidelity. This study sought to develop a statistical shape model (SSM) of the torso and lungs and evaluate whether patient-specific predictions of torso and lung shape could enhance EIT reconstructions in a Bayesian framework.

METHODS: torso and lung finite element surface meshes were fitted to computed tomography data from 81 participants, and a SSM was generated using principal component analysis and regression analyses. Predicted shapes were implemented in a Bayesian EIT framework and were quantitatively compared to generic reconstruction methods.

RESULTS: Five principal shape modes explained 38% of the cohort variance in lung and torso geometry, and regression analysis yielded nine total anthropometrics and pulmonary function metrics that significantly predicted these shape modes. Incorporation of SSM-derived structural information enhanced the accuracy and reliability of the EIT reconstruction as compared to generic reconstructions, demonstrated by reduced relative error, total variation, and Mahalanobis distance.

CONCLUSION: As compared to deterministic approaches, Bayesian EIT afforded more reliable quantitative and visual interpretation of the reconstructed ventilation distribution. However, no conclusive improvement of reconstruction performance using patient specific structural information was observed as compared to the mean shape of the SSM.

SIGNIFICANCE: The presented Bayesian framework builds towards a more accurate and reliable method for ventilation monitoring via EIT.

PMID:37028024 | DOI:10.1109/TBME.2023.3250650

Categories
Nevin Manimala Statistics

Diagnosis of Coexisting Valvular Heart Diseases Using Image-to-Sequence Translation of Contact Microphone Recordings

IEEE Trans Biomed Eng. 2023 Mar 6;PP. doi: 10.1109/TBME.2023.3253381. Online ahead of print.

ABSTRACT

OBJECTIVE: Development of a contact microphone-driven screening framework for the diagnosis of coexisting valvular heart diseases (VHDs).

METHODS: A sensitive accelerometer contact microphone (ACM) is employed to capture heart-induced acoustic components on the chest wall. Inspired by the human auditory system, ACM recordings are initially transformed into Mel-frequency cepstral coefficients (MFCCs) and their first and second derivatives, resulting in 3-channel images. An image-to-sequence translation network based on the convolution-meets-transformer (CMT) architecture is then applied to each image to find local and global dependencies in images, and predict a 5-digit binary sequence, where each digit corresponds to the presence of a specific type of VHD. The performance of the proposed framework is evaluated on 58 VHD patients and 52 healthy individuals using a 10-fold leave-subject-out cross-validation (10-LSOCV) approach.

RESULTS: Statistical analyses suggest an average sensitivity, specificity, accuracy, positive predictive value, and F1 score of 93.28%, 98.07%, 96.87%, 92.97%, and 92.4% respectively, for the detection of coexisting VHDs. Furthermore, areas under the curve (AUC) of 0.99 and 0.98 are respectively reported for the validation and test sets.

CONCLUSION: The high performances achieved prove that local and global features of ACM recordings effectively characterize heart murmurs associated with valvular abnormalities.

SIGNIFICANCE: Limited access of primary care physicians to echocardiography machines has resulted in a low sensitivity of 44% when using a stethoscope for the identification of heart murmurs. The proposed framework provides accurate decision-making on the presence of VHDs, thus reducing the number of undetected VHD patients in primary care settings.

PMID:37028021 | DOI:10.1109/TBME.2023.3253381

Categories
Nevin Manimala Statistics

Clinically Relevant Myocardium Segmentation in Cardiac Magnetic Resonance Images

IEEE J Biomed Health Inform. 2023 Mar 2;PP. doi: 10.1109/JBHI.2023.3250429. Online ahead of print.

ABSTRACT

Deep learning approaches have shown great success in myocardium region segmentation in Cardiac MR (CMR) images. However, most of these often ignore irregularities such as protrusions, breaks in contour, etc. As a result, the common practice by clinicians is to manually correct the obtained outputs for the evaluation of myocardium condition. This paper aims to make the deep learning systems capable of handling the aforementioned irregularities and satisfy desired clinical constraints, necessary for various downstream clinical analysis. We propose a refinement model which imposes structural constraints on the outputs of the existing deep learning-based myocardium segmentation methods. The complete system is a pipeline of deep neural networks where an initial network performs myocardium segmentation as accurate as possible and the refinement network removes defects from the initial output to make it suitable for clinical decision support systems. We experiment with datasets collected from four different sources and observe consistent final segmentation outputs with improvement up to 8% in Dice Coefficient and up to 18 pixels in Hausdorff Distance due to the proposed refinement model. The proposed refinement strategy leads to qualitative and quantitative improvements in the performances of all the considered segmentation networks. Our work is an important step towards the development of a fully automatic myocardium segmentation system. It can also be generalized for other tasks where the object of interest has regular structure and the defects can be modelled statistically.

PMID:37028020 | DOI:10.1109/JBHI.2023.3250429

Categories
Nevin Manimala Statistics

Internationalizing the business school: A comparative analysis of English-medium and Spanish-medium instruction impact on student performance

Eval Program Plann. 2023 Mar 23;98:102279. doi: 10.1016/j.evalprogplan.2023.102279. Online ahead of print.

ABSTRACT

Business degrees have been pioneers in adopting the internationalization of Higher Education Institutions with the option of English as Medium of Instruction (EMI). Research has grown about the EMI versus non-EMI lecturers and students’ performance measured through perception, motivation, discursive analysis or satisfaction measures. However, results have not been conclusive in the scarce number of papers comparing quantitative course grades of EMI versus non-EMI students. The aim of this research paper is to prove that there is no difference in attaining learning objectives among students within a Business Administration Degree in Spain regardless the language of instruction. The present observational study considers all enrolled freshman throughout a horizon of six consecutive years allowing more reliable results not affected by the specificities of courses or years. All 212 students in the EMI track were matched to non-EMI track counterparts taking into account all available covariates. Results not only show that there is no difference in the attained learning objectives between the two tracks, but also that EMI students’ grades are in fact better than their non-EMI counterparts, which might help to remove the believe many still have on the lower academic attainment of those following an EMI track.

PMID:37027996 | DOI:10.1016/j.evalprogplan.2023.102279

Categories
Nevin Manimala Statistics

Implementation of indication-based antibiotic order sentences improves antibiotic use in emergency departments

Am J Emerg Med. 2023 Mar 29;69:5-10. doi: 10.1016/j.ajem.2023.03.048. Online ahead of print.

ABSTRACT

INTRODUCTION: Prior data have suggested that suboptimal antibiotic prescribing in the emergency department (ED) is common for uncomplicated lower respiratory tract infections (LRTI), urinary tract infections (UTI), and acute bacterial skin and skin structure infections (ABSSSI). The objective of this study was to measure the effect of indication-based antibiotic order sentences (AOS) on optimal antibiotic prescribing in the ED.

METHODS: This was an IRB-approved quasi-experiment of adults prescribed antibiotics in EDs for uncomplicated LRTI, UTI, or ABSSSI from January to June 2019 (pre-implementation) and September to December 2021 (post-implementation). AOS implementation occurred in July 2021. AOS are lean process, electronic discharge prescriptions retrievable by name or indication within the discharge order field. The primary outcome was optimal prescribing, defined as correct antibiotic selection, dose, and duration per local and national guidelines. Descriptive and bivariate statistics were performed; multivariable logistic regression was used to determine variables associated with optimal prescribing.

RESULTS: A total of 294 patients were included: 147 pre-group and 147 post-group. Overall optimal prescribing improved from 12 (8%) to 34 (23%) (P < 0.001). Individual components of optimal prescribing were optimal selection at 90 (61%) vs 117 (80%) (P < 0.001), optimal dose at 99 (67%) vs 115 (78%) (P = 0.036), and optimal duration at 38 (26%) vs 50 (34%) (P = 0.13) for pre- and post-group, respectively. AOS was independently associated with optimal prescribing after multivariable logistic regression analysis (adjOR, 3.6; 95%CI,1.7-7.2). A post-hoc analysis showed low uptake of AOS by ED prescribers.

CONCLUSIONS: AOS are an efficient and promising strategy to enhance antimicrobial stewardship in the ED.

PMID:37027958 | DOI:10.1016/j.ajem.2023.03.048

Categories
Nevin Manimala Statistics

Climate plays a dominant role over land management in governing soil carbon dynamics in North Western Himalayas

J Environ Manage. 2023 Apr 5;338:117740. doi: 10.1016/j.jenvman.2023.117740. Online ahead of print.

ABSTRACT

The soil carbon (C) dynamics is strongly influenced by climate and land-use patterns in the Himalayas. Therefore, soils under five prominent land use [e.g., maize (Zea mays), horticulture, natural forest, grassland, and wasteland] were sampled down up to 30 cm depth under two climatic conditions viz., temperate and subtropical to assess the impacts of climate and landuse on soil C dynamics. Results demonstrated that irrespective of land use, temperate soil contains 30.66% higher C than subtropical soils. Temperate soils under natural forests had the higher total organic carbon (TOC, 21.90 g kg-1), Walkley-Black carbon (WBC, 16.42 g kg-1), contents, and stocks (TOC, 66.92 Mg ha-1 and WBC, 50.24 Mg ha-1), and total soil organic matter (TSOM, 3.78%) concentration as compared to other land uses like maize, horticulture, grassland, and wasteland. Under both climatic conditions, maize land use had the lowest TOC 9.63, 6.55 g kg-1 and WBC 7.22, 4.91 g kg-1 at 0-15 and 15-30 cm soil depth, respectively. Horticulture land use had 62.58 and 62.61% higher TOC and WBC over maize-based land use under subtropical and temperate climatic conditions at 0-30 cm soil depth, respectively. However, soils of maize land use under temperate conditions had ∼2 times more TOC than in subtropical conditions. The study inferred that the C-losses is more in the subtropical soil than in temperate soils. Hence, the subtropical region needs more rigorous adoption of C conservation farming practices than the temperate climatic setting. Although, the adoption of C storing and conserving practices is crucial under both climatic settings to arrest land degradation. Horticultural land uses along with conservation effective soil management practices may be encouraged to restore more soil C and to improve the livelihood security of the hill populace in the North Western Himalayas.

PMID:37027954 | DOI:10.1016/j.jenvman.2023.117740

Categories
Nevin Manimala Statistics

Sofa and bed-related pediatric trauma injuries treated in United States emergency departments

Am J Emerg Med. 2023 Mar 31;68:155-160. doi: 10.1016/j.ajem.2023.03.055. Online ahead of print.

ABSTRACT

INTRODUCTION: Children under the age of 5 years suffer from the highest rates of fall-related injuries. Caretakers often leave young children on sofas and beds, however, falling and rolling off these fixtures can lead to serious injury. We investigated the epidemiologic characteristics and trends of bed and sofa-related injuries among children aged <5 years treated in US emergency departments (EDs).

METHODS: We conducted a retrospective analysis of data from the National Electronic Injury Surveillance System from 2007 through 2021 using sample weights to estimate national numbers and rates of bed and sofa-related injuries. Descriptive statistics and regression analyses were employed.

RESULTS: An estimated 3,414,007 children aged <5 years were treated for bed and sofa-related injuries in emergency departments (EDs) in the United States from 2007 through 2021, averaging 115.2 injuries per 10,000 persons annually. Closed head injuries (30%) and lacerations (24%) comprised the majority of injuries. The primary location of injury was the head (71%) and upper extremity (17%). Children <1 year of age accounted for most injuries, with a 67% increase in incidence within the age group between 2007 and 2021 (p < 0.001). Falling, jumping, and rolling off beds and sofas were the primary mechanisms of injury. The proportion of jumping injuries increased with age. Approximately 4% of all injuries required hospitalization. Children <1 year of age were 1.58 times more likely to be hospitalized after injury than all other age groups (p < 0.001).

CONCLUSION: Beds and sofas can be associated with injury among young children, especially infants. The annual rate of bed and sofa-related injuries among infants <1 year old is increasing, which underscores the need for increased prevention efforts, including parental education and improved safety design, to decrease these injuries.

PMID:37027936 | DOI:10.1016/j.ajem.2023.03.055

Categories
Nevin Manimala Statistics

Fasting plasma glucose and alanine aminotransferase on the risk of hepatocellular carcinoma: A nested case-control study

Cancer Epidemiol. 2023 Apr 5;84:102362. doi: 10.1016/j.canep.2023.102362. Online ahead of print.

ABSTRACT

BACKGROUND: The risk of hepatocellular carcinoma (HCC) is associated with a variety of factors. However, the possible association between the abnormal metabolism of fasting plasma glucose (FPG) and alanine aminotransferase (ALT) and the risk of HCC has not been widely studied. We examined this relationship based on a prospective cohort study.

METHODS: 162 first-attack HCC cases during three follow-up periods (2014-2020) were selected as the case group. A control group of 648 participants was obtained by 1:4 matching of age (± 2 years) and sex with noncancer participants in the same period. Conditional logistic regression models, restricted cubic spline models, additive interaction models, and generalized additive models were used to explore the effects of FPG and ALT on the risk of HCC.

RESULTS: After correction for confounding factors, we found that abnormal FPG and elevated ALT increased the risk of HCC, respectively. Compared with the normal FPG group, the risk of HCC was significantly increased in the impaired fasting glucose (IFG) (OR = 1.91, 95 %CI: 1.04, 3.50) and diabetes groups (OR = 2.12, 95 %CI: 1.24, 3.63). Compared with the lowest quartile of ALT, subjects in the fourth quartile had an 84 % increased risk of HCC (OR = 1.84, 95 %CI: 1.05-3.21). Moreover, there was an interaction between FPG and ALT on the risk of HCC, and 74 % of the HCC risk could be attributed to their synergistic effect (AP = 0.74, 95 %CI: 0.56-0.92).

CONCLUSION: Abnormal FPG and elevated ALT are independent risk factors for HCC, and they have a synergistic effect on the risk of HCC. Therefore, serum FPG and ALT levels should be monitored to prevent the development of HCC.

PMID:37027905 | DOI:10.1016/j.canep.2023.102362