Real Medical Datasets Research Articles

Generation of synthetic tympanic membrane images: Development, human validation, and clinical implications of synthetic data.

Synthetic clinical images could augment real medical image datasets, a novel approach in otolaryngology-head and neck surgery (OHNS). Our objective was to develop a generative adversarial network (GAN) for tympanic membrane images and to validate the quality of synthetic images with human reviewers. Our model was developed using a state-of-the-art GAN architecture, StyleGAN2-ADA. The network was trained on intraoperative high-definition (HD) endoscopic images of tympanic membranes collected from pediatric patients undergoing myringotomy with possible tympanostomy tube placement. A human validation survey was administered to a cohort of OHNS and pediatrics trainees at our institution. The primary measure of model quality was the Frechet Inception Distance (FID), a metric comparing the distribution of generated images with the distribution of real images. The measures used for human reviewer validation were the sensitivity, specificity, and area under the curve (AUC) for humans' ability to discern synthetic from real images. Our dataset comprised 202 images. The best GAN was trained at 512x512 image resolution with a FID of 47.0. The progression of images through training showed stepwise "learning" of the anatomic features of a tympanic membrane. The validation survey was taken by 65 persons who reviewed 925 images. Human reviewers demonstrated a sensitivity of 66%, specificity of 73%, and AUC of 0.69 for the detection of synthetic images. In summary, we successfully developed a GAN to produce synthetic tympanic membrane images and validated this with human reviewers. These images could be used to bolster real datasets with various pathologies and develop more robust deep learning models such as those used for diagnostic predictions from otoscopic images. However, caution should be exercised with the use of synthetic data given issues regarding data diversity and performance validation. Any model trained using synthetic data will require robust external validation to ensure validity and generalizability.

An HBase-Based Optimization Model for Distributed Medical Data Storage and Retrieval

In medical services, the amount of data generated by medical devices is increasing explosively, and access to medical data is also put forward with higher requirements. Although HBase-based medical data storage solutions exist, they cannot meet the needs of fast locating and diversified access to medical data. In order to improve the retrieval speed, the recognition model S-TCR and the dynamic management algorithm SL-TCR, based on the behavior characteristics of access, were proposed to identify the frequently accessed hot data and dynamically manage the data storage medium as to maximize the system access performance. In order to improve the search performance of keys, an optimized secondary index strategy was proposed to reduce I/O overhead and optimize the search performance of non-primary key indexes. Comparative experiments were conducted on real medical data sets. The experimental results show that the optimized retrieval model can meet the needs of hot data access and diversified medical data retrieval.

Performance Evaluation of Canonical Correlation Analysis and Generalized Canonical Correlation Analysis with Some Continuous Distributed Data

This study was embarked to examine the performance evaluation of canonical correlation and generalized canonical correlation analysis with some continuous distributed data (Gamma, Gaussian, Exponential and Beta). The objectives of the study were to: ascertain if the anthropometric indicators of patients were correlated; ascertain if there is any relationship between vital signs and anthropometric dimensions of patients; obtain the relative efficiency of CCA and GCCA techniques for four continuous distributed simulated data; and determine the model performance adequacy of CCA and GCCA techniques. Real life medical data set was used, consisting of three response variables (Respiration rate, heart rate, temperature) named the vital signs and three predictor variables (Hip circumference, weight, height) named anthropometric dimensions. The study employed the real life data set to simulate data of sample sizes 15, 30, 45, 60, 100, 120, 140, 160, 400, 600, 800 and 1000 for the four continuous distributions. A computer programming language codes were written via R-Studio package to solve the numerous numerical problems in this study. The result of the study revealed that anthropometric dimensions, being the independent variables were not correlated, which implied that there was no symptom of multicollinearity using the Eigen values/condition index technique. In addition, there was significant relationship between vital signs and anthropometric dimensions of patients using Wilks’ Lambda, Hotelling-Lawley Trace, Pillai’s Bartlett Trace and Roy’s Largest Root multivariate statistics. The adequacy of the CCA and GCCA was evaluated using Wilcoxon rank sum test; and the result revealed that GCCA was more efficient than that of CCA for the Gamma and Beta distributed data, while for Gaussian and Exponential distributed data, the relative efficiency of the CCA and GCCA was the same.

Selection of Optimal Regression-like Equations for Circular Regression Model via Mallows’ $$C_p$$ and AIC Criteria

The problem of choosing the best regressors to fit the circular regression data has not been addressed. We focus on the problem of finding the optimal regression-like equations (ORLE) in the Sarma and Jammalamadaka (SJ) circular regression model (Sarma and Jammalamadaka 1993). First, the issues of under-fitting and over-fitting of regression equations in the SJ model are addressed. Then, we extend Mallows’ $$C_p$$ and AIC and their robust versions to the SJ circular regression model. A simulation study is used to investigate the performance of the proposed criteria. Results showed that robust circular Mallows’ $$C_p$$ and AIC are effective in selecting an accurate ORLE for circular regression models in both the clean and contaminated data sets. An application of the proposed procedures is discussed using a real medical data set.

Secure Medical Data Collection in the Internet of Medical Things Based on Local Differential Privacy

As big data and data mining technology advance, research on the collection and analysis of medical data on the internet of medical things (IoMT) has gained increasing attention. Medical institutions often collect users’ signs and symptoms from their devices for analysis. However, the process of data collection may pose a risk of privacy leakage without a trusted third party. To address this issue, we propose a medical data collection based on local differential privacy and Count Sketch (MDLDP). The algorithm first uses a random sampling technique to select only one symptom for perturbation by a single user. The perturbed data is then uploaded using Count Sketch. The third-party aggregates the user-submitted data to estimate the frequencies of the symptoms and the mean extent of their occurrence. This paper theoretically demonstrates that the designed algorithm satisfies local differential privacy and unbiased estimation. We also evaluated the algorithm experimentally with existing algorithms on a real medical dataset. The results show that the MDLDP algorithm has good utility for key-value type medical data collection statistics in the IoMT.

Variational Approach for Joint Kidney Segmentation and Registration from DCE-MRI Using Fuzzy Clustering with Shape Priors.

The dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) technique has great potential in the diagnosis, therapy, and follow-up of patients with chronic kidney disease (CKD). Towards that end, precise kidney segmentation from DCE-MRI data becomes a prerequisite processing step. Exploiting the useful information about the kidney's shape in this step mandates a registration operation beforehand to relate the shape model coordinates to those of the image to be segmented. Imprecise alignment of the shape model induces errors in the segmentation results. In this paper, we propose a new variational formulation to jointly segment and register DCE-MRI kidney images based on fuzzy c-means clustering embedded within a level-set (LSet) method. The image pixels' fuzzy memberships and the spatial registration parameters are simultaneously updated in each evolution step to direct the LSet contour toward the target kidney. Results on real medical datasets of 45 subjects demonstrate the superior performance of the proposed approach, reporting a Dice similarity coefficient of 0.94 ± 0.03, Intersection-over-Union of 0.89 ± 0.05, and 2.2 ± 2.3 in 95-percentile of Hausdorff distance. Extensive experiments show that our approach outperforms several state-of-the-art LSet-based methods as well as two UNet-based deep neural models trained for the same task in terms of accuracy and consistency.

A distribution-free smoothed combination method to improve discrimination accuracy in multi-category classification.

Results from multiple diagnostic tests are combined in many ways to improve the overall diagnostic accuracy. For binary classification, maximization of the empirical estimate of the area under the receiver operating characteristic curve has widely been used to produce an optimal linear combination of multiple biomarkers. However, in the presence of a large number of biomarkers, this method proves to be computationally expensive and difficult to implement since it involves maximization of a discontinuous, non-smooth function for which gradient-based methods cannot be used directly. The complexity of this problem further increases when the classification problem becomes multi-category. In this article, we develop a linear combination method that maximizes a smooth approximation of the empirical Hyper-volume Under Manifolds for the multi-category outcome. We approximate HUM by replacing the indicator function with the sigmoid function and normal cumulative distribution function. With such smooth approximations, efficient gradient-based algorithms are employed to obtain better solutions with less computing time. We show that under some regularity conditions, the proposed method yields consistent estimates of the coefficient parameters. We derive the asymptotic normality of the coefficient estimates. A simulation study is performed to study the effectiveness of our proposed method as compared to other existing methods. The method is illustrated using two real medical data sets.

Voltage slope guided learning in spiking neural networks.

A thorny problem in machine learning is how to extract useful clues related to delayed feedback signals from the clutter of input activity, known as the temporal credit-assignment problem. The aggregate-label learning algorithms make an explicit representation of this problem by training spiking neurons to assign the aggregate feedback signal to potentially effective clues. However, earlier aggregate-label learning algorithms suffered from inefficiencies due to the large amount of computation, while recent algorithms that have solved this problem may fail to learn due to the inability to find adjustment points. Therefore, we propose a membrane voltage slope guided algorithm (VSG) to further cope with this limitation. Direct dependence on the membrane voltage when finding the key point of weight adjustment makes VSG avoid intensive calculation, but more importantly, the membrane voltage that always exists makes it impossible to lose the adjustment point. Experimental results show that the proposed algorithm can correlate delayed feedback signals with the effective clues embedded in background spiking activity, and also achieves excellent performance on real medical classification datasets and speech classification datasets. The superior performance makes it a meaningful reference for aggregate-label learning on spiking neural networks.

Level-Set-Based Kidney Segmentation from DCE-MRI Using Fuzzy Clustering with Population-Based and Subject-Specific Shape Statistics.

The segmentation of dynamic contrast-enhanced magnetic resonance images (DCE-MRI) of the kidney is a fundamental step in the early and noninvasive detection of acute renal allograft rejection. In this paper, a new and accurate DCE‑MRI kidney segmentation method is proposed. In this method, fuzzy c-means (FCM) clustering is embedded into a level set method, with the fuzzy memberships being iteratively updated during the level set contour evolution. Moreover, population‑based shape (PB-shape) and subject-specific shape (SS-shape) statistics are both exploited. The PB-shape model is trained offline from ground-truth kidney segmentations of various subjects, whereas the SS-shape model is trained on the fly using the segmentation results that are obtained for a specific subject. The proposed method was evaluated on the real medical datasets of 45 subjects and reports a Dice similarity coefficient (DSC) of 0.953 ± 0.018, an intersection-over-union (IoU) of 0.91 ± 0.033, and 1.10 ± 1.4 in the 95-percentile of Hausdorff distance (HD95). Extensive experiments confirm the superiority of the proposed method over several state-of-the-art level set methods, with an average improvement of 0.7 in terms of HD95. It also offers an HD95 improvement of 9.5 and 3.8 over two deep neural networks based on the U-Net architecture. The accuracy improvements have been experimentally found to be more prominent on low-contrast and noisy images.

Time-Aware Multi-Type Data Fusion Representation Learning Framework for Risk Prediction of Cardiovascular Diseases.

Predicting the future risk of cardiovascular diseases from the historical Electronic Health Records (EHRs) is a significant research task in personalized healthcare fields. In recent years, many deep neural network-based methods have emerged, which model patient disease progression by capturing the temporal patterns in sequential visit data. However, existing methods usually cannot effectively integrate the features of heterogeneous clinical data, and do not fully consider the impact of patients age and irregular time interval between consecutive medical records on the patients disease development. To address these challenges, we propose a Time-Aware Multi-type Data fUsion Representation learning framework (TAMDUR) for CVDs risk prediction. In this framework, we design a time-aware decay function, which is based on the patients age and the elapsed time between visits, to model the disease progression pattern. A parallel combination of Bi LSTM and CNN is constructed to respectively learn the temporal and non-temporal features from various types of clinical data. Finally, a multi-type data fusion representation layer based on self-attention is utilized to integrate various features and their correlations to obtain the final patient representation. We evaluate our model on a real medical dataset, and the experimental results demonstrate that TAMDUR outperforms the state-of-the-art approaches.

On Suitability of Mixture of Generalized Exponential Models in Modeling Right-Censored Medical Datasets Using Conditional Expectations.

The exploration of suitable models for modeling censored medical datasets is of great importance. There are numerous studies dealing with modeling the censored medical datasets. However, majority of the earlier contributions have utilized the conventional models for modeling the said datasets. Unfortunately, the conventional models are not capable of capturing the behavior of the heterogeneous datasets involving the mixture of two or more subpopulations. In addition, the earlier contributions have considered conventional censoring schemes by replacing all the censored items with the largest failed item. This paper is aimed at proposing the analysis of right-censored mixture medical datasets. The mixture of the generalized exponential distribution has been proposed to model the right-censored heterogeneous medical datasets. In converse to conventional censoring schemes, we have proposed censoring schemes which replace the censored items with conditional expectation (CE) of the random variable. In addition, the Bayesian methods have been proposed to estimate the model parameters. The performance and sensitivity of the proposed estimators have been evaluated using a detailed simulation study. The detailed simulation study suggests that censoring schemes based on CE provide improved estimation as compared to conventional censoring schemes. The suitability of the model in modeling heterogeneous datasets has been verified by modeling two real right-censored medical datasets. The comparison of the proposed model with existing mixture model under Bayesian methods advocated the improved performance of the proposed model.

Differential privacy medical data publishing method based on attribute correlation

The advent of the era of big data promotes the further development of medicine, and data release is an important step in it. The existing medical data release methods mostly use the k-anonymity model as the basis for data protection. With the advancement of technology, anonymous models are progressively less resistant to consistency attacks and background knowledge attacks. In order to better protect the private information of patients, this paper makes two major contributions: (1) The method of calculating the correlation between attributes is used to ensure the validity of the data after the data is released; (2) On the basis of the previous step, combined with the difference privacy-preserving model and tree model, this paper proposes an attribute association-based differential privacy classification tree data publishing method (ACDP-Tree). In this paper, simulation experiments are carried out on real medical data sets. The experimental results show that the algorithm ensures the validity and availability of the data to a certain extent while ensuring that the patient's privacy is not leaked.

A new cure rate regression framework for bivariate data based on the Chen distribution.

The present study introduces a new multivariate mixture cure rate model based on the Chen probability distribution to model recurrent event data in the presence of cure fraction. In this context, we provide an alternative for the use of some usual modeling approaches as the semiparametric Cox proportional hazards model commonly used in lifetime data analysis, considering a new bivariate parametric model to be used in the data analysis of bivariate lifetime data assuming a mixture structure for the bivariate data in presence of covariates, censored data and cure fraction. Under a Bayesian setting, the proposed methodology was considered to analyze two real medical datasets from a retrospective cohort study related to leukemia and diabetic retinopathy diseases. The model validation process was addressed by using the Cox-Snell residuals, which allowed us to identify the suitability of the new proposed mixture cure rate model.

Discrete Bilal Distribution in the Presence of Right-Censored Data and a Cure Fraction

The statistical literature presents many continuous probability distributions with only one parameter, which are extensively used in the analysis of lifetime data, such as the exponential, the Lindley, and the Rayleigh distributions. Alternatively, the use of discretized versions of these distributions can provide a better fit for the data in many applications. As the novelty of this study, we present inferences for the discrete Bilal distribution (DB) with one parameter introduced by Altun et al. (2020) in the presence of right-censored data and cure fraction. We assume standard maximum likelihood methods based on asymptotic normality of the maximum likelihood estimators and also a Bayesian approach based on MCMC (Markov Chain Monte Carlo) simulation methods to get inferences for the parameters of the discrete BD distribution. The use of the proposed model was illustrated with three examples considering real medical lifetime data sets. From these applications, we concluded that the proposed model based on the discrete DB distribution has good performance even with the inclusion of a cure fraction in comparison to other existing discrete models, such as the DsFx-I, Lindley, Rayleigh, and Burr-Hatke probability distributions. Moreover, the model can be easily implemented in standard existing software, such as the R package. Under a Bayesian approach, we assumed a gamma prior distribution for the parameter of the DB discrete distribution. We also provided a brief sensitivity analysis assuming the half-normal distribution in place of the gamma distribution for the parameter of the DB distribution. From the obtained results of this study, we can conclude that the proposed methodology can be very useful for researchers dealing with medical discrete lifetime data in the presence of right-censored data and cure fraction.

Gamma Generalized Logistic Distribution: Properties and Applications

In this paper we consider a new class of asymmetric logistic distribution that contains both the type I and type II generalized logistic distributions of Balakrishnan and Leung (Commun Stat Simul Comput 17(1):25–50, 1988) as its special cases. We investigate some important properties of the distribution such as expressions for its mean, variance, characteristic function, measure of skewness and kurtosis, entropy etc. along with the distribution of its order statistics. A location-scale extension of the distribution is defined and discussed the maximum likelihood estimation of its parameters. Further, two real life medical data sets are utilized for illustrating the usefulness of the model and a simulation study is conducted for examining the performance of the maximum likelihood estimators of the parameters of the distribution.

QAPP: A quality-aware and privacy-preserving medical image release scheme

The direct release of medical image may face the dilemma: the privacy protection of medical images inevitably affects the visual quality of images. To balance medical image quality and privacy, this paper proposes a quality-aware and privacy-preserving medical image release scheme, QAPP, which effectively integrates the discrete cosine transform (DCT) with differential privacy (DP). Specifically, QAPP is composed of three phases. First, DCT is applied to each medical image to obtain its cosine coefficients matrix. Second, the original cosine coefficients matrix is compressed into k*k cosine coefficients matrix, which can retain the main features of each image. Third, the appropriate Laplace noise is injected into the formed k*k matrix to achieve differential privacy, and these noise-added coefficients are used to reconstruct the noise-added medical images through inverse DCT. Especially, considering there two error sources affecting the image quality in our work: the compression error caused by DCT, and the injected noise error caused by DP, Therefore, a selection function is proposed to determine the optimal compression dimension k, which can minimize the influence of these two errors to improve the visualization quality of the medical image. Subjective and objective image quality evaluation, and extensive experiments of image classification and segmentation using the real medical image dataset demonstrate that the proposed method QAPP can better balance medical image quality and privacy than other similar DP-based methods.

Multi-kernel graph fusion for spectral clustering

Many methods of multi-kernel clustering have a bias to power base kernels by ignoring other kernels. To address this issue, in this paper, we propose a new method of multi-kernel graph fusion based on min–max optimization (namely MKGF-MM) for spectral clustering by making full use of all base kernels. Specifically, the proposed method investigates a novel min–max weight strategy to capture the complementary information among all base kernels. As a result, every base kernel contributes to the construction of the fusion graph from all base kernels so that the quality of the fusion graph is guaranteed. In addition, we design an iterative optimization method to solve the proposed objective function. Furthermore, we theoretically prove that our optimization method achieves convergence. Experimental results on real medical datasets and scientific datasets demonstrate that the proposed method outperforms all comparison methods and the proposed optimization method achieves fast convergence.

Exponentiated Ailamujia distribution with statistical inference and applications of medical data

In this article, we study some properties of a new distribution, namely Exponentiated Ailamujia distribution (EAD). The Exponentiated Ailamujia has two parameters (scale and shape). The different structural properties of the EAD have been obtained as the moments, reliability analysis, and the harmonic mean. The parameters of the proposed model are estimated through the maximum likelihood method. Real life medical data sets are considered to show that the exponentiated Ailamujia distribution has a better fit as compared to the Ailamujia and Lindley distributions.

Detecting Ransomware within Real Healthcare Medical Records Adopting Internet of Medical Things using Machine and Deep Learning Techniques

The Internet of Medical Things was immensely implemented in healthcare systems during the covid 19 pandemic to enhance the patient's circumstances remotely in critical care units while keeping the medical staff safe from being infected. However, Healthcare systems were severely affected by ransomware attacks that may override data or lock systems from caregivers' access. In this work, after obtaining the required approval, we have got a real medical dataset from actual critical care units. For the sake of research, a portion of data was used, transformed, and manifested using laboratory-made payload ransomware and successfully labeled. The detection mechanism adopted supervised machine learning techniques of K Nearest Neighbor, Support Vector Machine, Decision Trees, Random Forest, and Logistic Regression in contrast with deep learning technique of Artificial Neural Network. The methods of KNN, SVM, and DT successfully detected ransomware's signature with an accuracy of 100%. However, ANN detected the signature with an accuracy of 99.9%. The results of this work were validated using precision, recall, and f1 score metrics.

A Hybrid Imputation Method for Multi-Pattern Missing Data: A Case Study on Type II Diabetes Diagnosis

Real medical datasets usually consist of missing data with different patterns which decrease the performance of classifiers used in intelligent healthcare and disease diagnosis systems. Many methods have been proposed to impute missing data, however, they do not fulfill the need for data quality especially in real datasets with different missing data patterns. In this paper, a four-layer model is introduced, and then a hybrid imputation (HIMP) method using this model is proposed to impute multi-pattern missing data including non-random, random, and completely random patterns. In HIMP, first, non-random missing data patterns are imputed, and then the obtained dataset is decomposed into two datasets containing random and completely random missing data patterns. Then, concerning the missing data patterns in each dataset, different single or multiple imputation methods are used. Finally, the best-imputed datasets gained from random and completely random patterns are merged to form the final dataset. The experimental evaluation was conducted by a real dataset named IRDia including all three missing data patterns. The proposed method and comparative methods were compared using different classifiers in terms of accuracy, precision, recall, and F1-score. The classifiers’ performances show that the HIMP can impute multi-pattern missing values more effectively than other comparative methods.