Explainable Risk Prediction of Post-Stroke Adverse Mental Outcomes Using Machine Learning Techniques in a Population of 1780 Patients

Post-stroke depression and anxiety, collectively known as post-stroke adverse mental outcome (PSAMO) are common sequelae of stroke. About 30% of stroke survivors develop depression and about 20% develop anxiety. Stroke survivors with PSAMO have poorer health outcomes with higher mortality and greater functional disability. In this study, we aimed to develop a machine learning (ML) model to predict the risk of PSAMO. We retrospectively studied 1780 patients with stroke who were divided into PSAMO vs. no PSAMO groups based on results of validated depression and anxiety questionnaires. The features collected included demographic and sociological data, quality of life scores, stroke-related information, medical and medication history, and comorbidities. Recursive feature elimination was used to select features to input in parallel to eight ML algorithms to train and test the model. Bayesian optimization was used for hyperparameter tuning. Shapley additive explanations (SHAP), an explainable AI (XAI) method, was applied to interpret the model. The best performing ML algorithm was gradient-boosted tree, which attained 74.7% binary classification accuracy. Feature importance calculated by SHAP produced a list of ranked important features that contributed to the prediction, which were consistent with findings of prior clinical studies. Some of these factors were modifiable, and potentially amenable to intervention at early stages of stroke to reduce the incidence of PSAMO.

Artificial intelligence assisted tools for the detection of anxiety and depression leading to suicidal ideation in adolescents: a review

Epidemiological studies report high levels of anxiety and depression amongst adolescents. These psychiatric conditions and complex interplays of biological, social and environmental factors are important risk factors for suicidal behaviours and suicide, which show a peak in late adolescence and early adulthood. Although deaths by suicide have fallen globally in recent years, suicide deaths are increasing in some countries, such as the US. Suicide prevention is a challenging global public health problem. Currently, there aren't any validated clinical biomarkers for suicidal diagnosis, and traditional methods exhibit limitations. Artificial intelligence (AI) is budding in many fields, including in the diagnosis of medical conditions. This review paper summarizes recent studies (past 8 years) that employed AI tools for the automated detection of depression and/or anxiety disorder and discusses the limitations and effects of some modalities. The studies assert that AI tools produce promising results and could overcome the limitations of traditional diagnostic methods. Although using AI tools for suicidal ideation exhibits limitations, these are outweighed by the advantages. Thus, this review article also proposes extracting a fusion of features such as facial images, speech signals, and visual and clinical history features from deep models for the automated detection of depression and/or anxiety disorder in individuals, for future work. This may pave the way for the identification of individuals with suicidal thoughts.

Epileptic Seizures Detection Using Deep Learning Techniques: A Review

A variety of screening approaches have been proposed to diagnose epileptic seizures, using electroencephalography (EEG) and magnetic resonance imaging (MRI) modalities. Artificial intelligence encompasses a variety of areas, and one of its branches is deep learning (DL). Before the rise of DL, conventional machine learning algorithms involving feature extraction were performed. This limited their performance to the ability of those handcrafting the features. However, in DL, the extraction of features and classification are entirely automated. The advent of these techniques in many areas of medicine, such as in the diagnosis of epileptic seizures, has made significant advances. In this study, a comprehensive overview of works focused on automated epileptic seizure detection using DL techniques and neuroimaging modalities is presented. Various methods proposed to diagnose epileptic seizures automatically using EEG and MRI modalities are described. In addition, rehabilitation systems developed for epileptic seizures using DL have been analyzed, and a summary is provided. The rehabilitation tools include cloud computing techniques and hardware required for implementation of DL algorithms. The important challenges in accurate detection of automated epileptic seizures using DL with EEG and MRI modalities are discussed. The advantages and limitations in employing DL-based techniques for epileptic seizures diagnosis are presented. Finally, the most promising DL models proposed and possible future works on automated epileptic seizure detection are delineated.

Evolutionary algorithm-based classifier parameter tuning for automatic ovarian cancer tissue characterization and classification

PURPOSE

Ovarian cancer is one of the most common gynecological cancers in women. It is difficult to accurately and objectively diagnose benign and malignant ovarian tumors using ultrasound and other tests. Hence, there is an imperative need to develop a computer-aided diagnostic (CAD) system for ovarian tumor classification in order to reduce patient anxiety and the cost of unnecessary biopsies. In this paper, we present an automatic CAD system for the detection of benign and malignant ovarian tumors using advanced image processing and data mining techniques.

MATERIALS AND METHODS

In the proposed system, Hu's invariant moments, Gabor transform parameters and entropies are first extracted from the acquired ultrasound images. Significant features are then used to train a probabilistic neural network (PNN) classifier for classifying the images into benign and malignant categories. The model parameter (σ) for which the PNN classifier performs the best is identified using a genetic algorithm (GA).

RESULTS

The proposed system was validated using 1300 benign images and 1300 malignant images, obtained from 10 patients with a benign disease and 10 with a malignant disease. We used 23 statistically significant (p < 0.0001) features. By evaluating the classifier using a ten-fold cross-validation technique, we were able to achieve an average classification accuracy of 99.8 %, sensitivity of 99.2 % and specificity of 99.6 % with a σ of 0.264.

CONCLUSION

The proposed system is automated and hence is more objective, can be easily deployed in any computer, is fast and accurate and can act as an adjunct tool in helping physicians make a confident call about the nature of the ovarian tumor under evaluation.

Automated carotid intima-media thickness and its link for prediction of SYNTAX score in Japanese coronary artery disease patients

AIM

The purpose of this study was to evaluate whether the automated carotid intima-media thickness (CIMT) identified by using automated software could predict the SYNTAX score for coronary artery disease (CAD) patients.

METHODS

Three-hundred-seventy consecutive patients (males 218; median age 69±11 years) who underwent carotid-ultrasound and coronary angiography were analyzed. Two experienced interventional cardiologists calculated the SYNTAX score from the carotid angiograms. After ultrasonographic examinations were performed, the plaque score (PS) was calculated and automated carotid IMT analysis was obtained by a fully automated algorithm. Correlation and stepwise logistic regression analysis were calculated and also the receiver operating characteristics (ROC) curve analysis was computed.

RESULTS

The mean SYNTAX score was 8.1±14.4; the PS was 7.1±14.4 and the mean CIMT was 0.86±0.23 mm (Normality rejected with a P-value of 0.001). A statistically significant correlation was found between the CIMT and SYNTAX score (r=0.323; P=0.0001) and between the PS and SYNTAX score (r=0.583; P=0.0001). The area under the ROC curve (Az) between CIMT and coronary artery disease was 0.648 (P=0.0001) and the CIMT of 1 mm or more was associated with the presence coronary artery disease with a specificity of 90.5%. Logistic regression analysis confirmed the association between CIMT and SYNTAX score (P=0.0002).

CONCLUSIONS

Results of our study using an automated algorithm showed a statistical significant association between CIMT and SYNTAX score and indicated that CIMT may be considered a reliable parameter for prediction of SYNTAX score in coronary artery disease patient population from Japan.

Diagnosis of Hashimoto's thyroiditis in ultrasound using tissue characterization and pixel classification

Hashimoto's thyroiditis is the most common type of inflammation of the thyroid gland, and accurate diagnosis of Hashimoto's thyroiditis would be helpful to better manage the disease process and predict thyroid failure. Most of the published computer-based techniques that use ultrasound thyroid images for Hashimoto's thyroiditis diagnosis are limited by lack of procedure standardization because individual investigators use various initial ultrasound settings. This article presents a computer-aided diagnostic technique that uses grayscale features and classifiers to provide a more objective and reproducible classification of normal and Hashimoto's thyroiditis-affected cases. In this paradigm, we extracted grayscale features based on entropy, Gabor wavelet, moments, image texture, and higher order spectra from the 100 normal and 100 Hashimoto's thyroiditis-affected ultrasound thyroid images. Significant features were selected using t-test. The resulting feature vectors were used to build the following three classifiers using tenfold stratified cross validation technique: support vector machine, k-nearest neighbor, and radial basis probabilistic neural network. Our results show that a combination of 12 features coupled with support vector machine classifier with the polynomial kernel of order 1 and linear kernel gives the highest accuracy of 80%, sensitivity of 76%, specificity of 84%, and positive predictive value of 83.3% for the detection of Hashimoto's thyroiditis. The proposed computer-aided diagnostic system uses novel features that have not yet been explored for Hashimoto's thyroiditis diagnosis. Even though the accuracy is only 80%, the presented preliminary results are encouraging to warrant analysis of more such powerful features on larger databases.

Computed tomography carotid wall plaque characterization using a combination of discrete wavelet transform and texture features: A pilot study

In 30% of stroke victims, the cause of stroke has been found to be the stenosis caused by plaques in the carotid artery. Early detection of plaque and subsequent classification of the same into symptomatic and asymptomatic can help the clinicians to choose only those patients who are at a higher risk of stroke for risky surgeries and stenosis treatments. Therefore, in this work, we have proposed a non-invasive computer-aided diagnostic technique to classify the detected plaque into the two classes. Computed tomography (CT) images of the carotid artery images were used to extract Local Binary Pattern (LBP) features and wavelet energy features. Significant features were then used to train and test several supervised learning algorithm based classifiers. The Support Vector Machine (SVM) classifier with various kernel configurations was evaluated using LBP and wavelet features. The SVM classifier presented the highest accuracy of 88%, sensitivity of 90.2%, and specificity of 86.5% for radial basis function (RBF) kernel function. The CT images of the carotid artery provide unique 3D images of the artery and plaque that could be used for calculating percentage of stenosis. Our proposed technique enables automatic classification of plaque into asymptomatic and symptomatic with high accuracy, and hence, it can be used for deciding the course of treatment. We have also proposed a single-valued integrated index (Atheromatic Index) using the significant features which can provide a more objective and faster prediction of the class.

Cost-effective and non-invasive automated benign and malignant thyroid lesion classification in 3D contrast-enhanced ultrasound using combination of wavelets and textures: a class of ThyroScan™ algorithms

Ultrasound has great potential to aid in the differential diagnosis of malignant and benign thyroid lesions, but interpretative pitfalls exist and the accuracy is still poor. To overcome these difficulties, we developed and analyzed a range of knowledge representation techniques, which are a class of ThyroScan™ algorithms from Global Biomedical Technologies Inc., California, USA, for automatic classification of benign and malignant thyroid lesions. The analysis is based on data obtained from twenty nodules (ten benign and ten malignant) taken from 3D contrast-enhanced ultrasound images. Fine needle aspiration biopsy and histology confirmed malignancy. Discrete Wavelet Transform (DWT) and texture algorithms are used to extract relevant features from the thyroid images. The resulting feature vectors are fed to three different classifiers: K-Nearest Neighbor (K-NN), Probabilistic Neural Network (PNN), and Decision Tree (DeTr). The performance of these classifiers is compared using Receiver Operating Characteristic (ROC) curves. Our results show that combination of DWT and texture features coupled with K-NN resulted in good performance measures with the area of under the ROC curve of 0.987, a classification accuracy of 98.9%, a sensitivity of 98%, and a specificity of 99.8%. Finally, we have proposed a novel integrated index called Thyroid Malignancy Index (TMI), which is made up of texture features, to diagnose benign or malignant nodules using just one index. We hope that this TMI will help clinicians in a more objective detection of benign and malignant thyroid lesions.

Automated identification of diabetes type-2 subjects with and without neuropathy using eigenvalues

Diabetes is a disorder of metabolism and has been a leading healthcare burden throughout the world. The most typical form of diabetes is type-2 diabetes. It is commonly developed in adults of age 40 and older. The purpose of this study is to identify the plantar pressure distribution in normal subjects, diabetic type-2 subjects with neuropathy, and diabetic type-2 subjects without neuropathy. Foot scan images were obtained using the F-Scan (Tekscan USA) in-shoe measurement system. The eigenvalues were evaluated from principal-component analysis after performing continuous wavelets transformation (CWT). The eigenvalues of CWT in regions 5 and 7 had shown excellent p values of more than 95 per cent confidence level when subjected to an analysis-of-variance test. These parameters were then presented to an artificial neural network (ANN) and a Gaussian mixture model (GMM) for automatic classification. The results show that the ANN classifier performs better than the GMM and is able to identify the unknown class with a sensitivity of 100 per cent and a specificity of 72 per cent.

Non-linear analysis of body responses to functional electrical stimulation on hemiplegic subjects

Functional electrical stimulation (FES) is a method of applying low-level electrical currents to restore or improve body functions lost through nervous system impairment. FES is applied to peripheral nerves that control specific muscles or muscle groups. Application of advanced signal computing techniques to the medical field has helped to achieve practical solutions to the health care problems accurately. The physiological signals are essentially non-stationary and may contain indicators of current disease, or even warnings about impending diseases. These indicators may be present at all times or may occur at random on the timescale. However, to study and pinpoint these subtle changes in the voluminous data collected over several hours is tedious. These signals, e.g. walking-related accelerometer signals, are not simply linear and involve non-linear contributions. Hence, non-linear signal-processing methods may be useful to extract the hidden complexities of the signal and to aid physicians in their diagnosis. In this work, a young female subject with major neuromuscular dysfunction of the left lower limb, which resulted in an asymmetric hemiplegic gait, participated in a series of FES-assisted walking experiments. Two three-axis accelerometers were attached to her left and right ankles and their corresponding signals were recorded during FES-assisted walking. The accelerometer signals were studied in three directions using the Hurst exponent H, the fractal dimension (FD), the phase space plot, and recurrence plots (RPs). The results showed that the H and FD values increase with increasing FES, indicating more synchronized variability due to FES for the left leg (paralysed leg). However, the variation in the normal right leg is more chaotic on FES.

Computer-based detection of diabetes retinopathy stages using digital fundus images

Diabetes mellitus is a heterogeneous clinical syndrome characterized by hyperglycaemia and the long-term complications are retinopathy, neuropathy, nephropathy, and cardiomyopathy. It is a leading cause of blindness. Diabetic retinopathy is the progressive pathological alterations in the retinal microvasculature, leading to areas of retinal nonperfusion, increased vascular permeability, and the pathological proliferation of retinal vessels. Hence, it is beneficial to have regular cost-effective eye screening for diabetes subjects. Nowadays, different stages of diabetes retinopathy are detected by retinal examination using indirect biomicroscopy by senior ophthalmologists. In this work, morphological image processing and support vector machine (SVM) techniques were used for the automatic diagnosis of eye health. In this study, 331 fundus images were analysed. Five groups were identified: normal retina, mild non-proliferative diabetic retinopathy, moderate non-proliferative diabetic retinopathy, severe non-proliferative diabetic retinopathy, and proliferative diabetic retinopathy. Four salient features blood vessels, microaneurysms, exudates, and haemorrhages were extracted from the raw images using image-processing techniques and fed to the SVM for classification. A sensitivity of more than 82 per cent and specificity of 86 per cent was demonstrated for the system developed.

Automatic identification of epileptic electroencephalography signals using higher-order spectra

Epilepsy is a pathological condition characterized by the spontaneous and unforeseeable occurrence of seizures, during which the perception or behaviour of patients is disturbed. An automatic early detection of the seizure onsets would help the patients and observers to take appropriate precautions. Various methods have been proposed to predict the onset of seizures based on electroencephalography (EEG) recordings. The use of non-linear features motivated by the higher-order spectra (HOS) has been reported to be a promising approach to differentiate between normal, background (pre-ictal), and epileptic EEG signals. In this work, these features are used to train both a Gaussian mixture model classifier and a support vector machine classifier. Results show that the classifiers were able to achieve 93.11 per cent and 92.67 per cent classification accuracy respectively, with selected HOS-based features. About 2 h of EEG recordings from ten patients were used in this study.

Analysis of body responses to an accelerating platform by the largest-Lyapunov-exponent method

Various disciplines have benefited from the advent of high-performance computing in achieving practical solutions to their problems, and the area of health care is no exception to this. Non-linear signal-processing tools have been developed to understand the hidden complexity of the time series, and these will help clinicians in diagnosis and treatment. Postural study helps the elderly and people with a balancing problem due to various pathological conditions. In elderly subjects, falls are common and may result in injury. Correct postural balance is basic to well-being and it influences our daily life significantly. These postural signals are non-stationary; they may appear to be random in the time scale and it is difficult to observe the subtle changes for the human observer. Hence, more hidden information can be obtained from the signal using non-linear parameters. In this paper, ten young normal subjects are subjected to the balancing platform whose acceleration is gradually increased from 1 m/s2 to 5 m/s2 to study the postural response. The ankle front-back acceleration and ankle pitch angular velocity sensor data were studied using the largest Lyapunov exponent (LLE). The results show that for higher acceleration of the platform the ankle movement follows a particular rhythm, resulting in a lower Lyapunov exponent. During lower acceleration of the balancing platform, this value is higher because of the random movement of the ankle. In this work, the pattern of the body response was studied using LLE values for different accelerations using ankle data as the base signal for the normal subjects.

Computer-based analysis of cardiac state using entropies, recurrence plots and Poincare geometry

Heart rate variability refers to the regulation of the sinoatrial node, the natural pacemaker of the heart by the sympathetic and parasympathetic branches of the autonomic nervous system. Heart rate variability is important because it provides a window to observe the heart's ability to respond to normal regulatory impulses that affect its rhythm. A computer-based intelligent system for analysis of cardiac states is very useful in diagnostics and disease management. Parameters are extracted from the heart rate signals and analysed using computers for diagnostics. This paper describes the analysis of normal and seven types of cardiac abnormal signals using approximate entropy (ApEn), sample entropy (SampEn), recurrence plots and Poincare plot patterns. Ranges of these parameters for various cardiac abnormalities are presented with an accuracy of more than 95%. Among the two entropies, ApEn showed better performance for all the cardiac abnormalities. Typical Poincare and recurrence plots are shown for various cardiac abnormalities.

Cardiac state diagnosis using higher order spectra of heart rate variability

Heart rate variability (HRV) refers to the regulation of the sinoatrial node, the natural pacemaker of the heart, by the sympathetic and parasympathetic branches of the autonomic nervous system. Heart rate variability analysis is an important tool to observe the heart's ability to respond to normal regulatory impulses that affect its rhythm. A computer-based intelligent system for analysis of cardiac states is very useful in diagnostics and disease management. Like many bio-signals, HRV signals are nonlinear in nature. Higher order spectral analysis (HOS) is known to be a good tool for the analysis of nonlinear systems and provides good noise immunity. In this work, we studied the HOS of the HRV signals of normal heartbeat and seven classes of arrhythmia. We present some general characteristics for each of these classes of HRV signals in the bispectrum and bicoherence plots. We also extracted features from the HOS and performed an analysis of variance (ANOVA) test. The results are very promising for cardiac arrhythmia classification with a number of features yielding a p-value < 0.02 in the ANOVA test.

Lead acetate induced cytotoxicity in male germinal cells of Swiss mice

Single intraperitoneal injection of lead acetate (200 mg/kg b.w) to Swiss mice stimulated testicular weight loss with a constant increase in the incidence of abnormal sperm population and decrease in the total sperm count. Testicular ascorbic acid also declined significantly during the post-treatment phase with significant rise in Lipid Peroxidation Potential (LPP) of the tissue. Elevated LPP is indicative of oxidative stress in treated mice testes. The possible role of lead-induced oxidative stress in culminating increased sperm abnormality and decreased sperm count have been discussed. Further, possible antioxidative role of testicular ascorbic acid in minimizing oxidative stress in lead-treated mice has been demonstrated.

In vivo lipid peroxidation responses of tissues in lead-treated Swiss mice

A single intraperitoneal injection of lead acetate (200 mg/kg body weight) increased the lipid peroxidation potential (LPP) measured as thiobarbituric acid-reactive substance (TBA-RS) in different tissues of Swiss mice. All the tissues taken for experimentation, generated significantly higher amount of TBA-RS in lead-treated mice when compared with the respective control value. However, none of the tissues could correspond to the control value after the lapse of four weeks post-treatment. Possibilities of differential responsiveness of tissues to generate lipid peroxides in lead-treated mice have been discussed.