Data analytics for cardiac diseases.

In the present research we tackled the classification of seven genetic cardiac diseases and control subjects by using an extensive set of machine learning algorithms with their variations from simple K-nearest neighbor searching method to support vector machines. The research was based on calcium transient signals measured from induced pluripotent stem cell-derived cardiomyocytes. All in all, 55 different machine learning alternatives were used to model eight classes by applying the principle of 10-fold crossvalidation with the peak data of 1626 signals. The best classification accuracy of approximately 69% was given by random forests, which can be seen high enough here to show machine learning to be potential for the differentiation of the eight disease classes.

On computational classification of genetic cardiac diseases applying iPSC cardiomyocytes.

BACKGROUND: Cardiomyocytes differentiated from human induced pluripotent stem cells (iPSC-CMs) can be used to study genetic cardiac diseases. In patients these diseases are manifested e.g. with impaired contractility and fatal cardiac arrhythmias, and both of these can be due to abnormal calcium transients in cardiomyocytes. Here we classify different genetic cardiac diseases using Ca2+ transient data and different machine learning algorithms. METHODS: By studying calcium cycling of disease-specific iPSC-CMs and by using calcium transients measured from these cells it is possible to classify diseases from each other and also from healthy controls by applying machine learning computation on the basis of peak attributes detected from calcium transient signals. RESULTS: In the current research we extend our previous study having Ca-transient data from four different genetic diseases by adding data from two additional diseases (dilated cardiomyopathy and long QT Syndrome 2). We also study, in the light of the current data, possible differences and relations when machine learning modelling and classification accuracies were computed by using either leave-one-out test or 10-fold cross-validation. CONCLUSIONS: Despite more complex classification tasks compared to our earlier research and having more different genetic cardiac diseases in the analysis, it is still possible to attain good disease classification results. As excepted, leave-one-out test and 10-fold cross-validation achieved virtually equal results.

Analysis of Drug Effects on iPSC Cardiomyocytes with Machine Learning.

Patient-specific induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) offer an attractive experimental platform to investigate cardiac diseases and therapeutic outcome. In this study, iPSC-CMs were utilized to study their calcium transient signals and drug effects by means of machine learning, a central part of artificial intelligence. Drug effects were assessed in six iPSC-lines carrying different mutations causing catecholaminergic polymorphic ventricular tachycardia (CPVT), a highly malignant inherited arrhythmogenic disorder. The antiarrhythmic effect of dantrolene, an inhibitor of sarcoplasmic calcium release, was studied in iPSC-CMs after adrenaline, an adrenergic agonist, stimulation by machine learning analysis of calcium transient signals. First, beats of transient signals were identified with our peak recognition algorithm previously developed. Then 12 peak variables were computed for every identified peak of a signal and by means of this data signals were classified into different classes corresponding to those affected by adrenaline or, thereafter, affected by a drug, dantrolene. The best classification accuracy was approximately 79% indicating that machine learning methods can be utilized in analysis of iPSC-CM drug effects. In the future, data analysis of iPSC-CM drug effects together with machine learning methods can create a very valuable and efficient platform to individualize medication in addition to drug screening and cardiotoxicity studies.

Separation of HCM and LQT Cardiac Diseases with Machine Learning of Ca2+ Transient Profiles.

BACKGROUND: Modeling human cardiac diseases with induced pluripotent stem cells not only enables to study disease pathophysiology and develop therapies but also, as we have previously showed, it can offer a tool for disease diagnostics. We previously observed that a few genetic cardiac diseases can be separated from each other and healthy controls by applying machine learning to Ca2+ transient signals measured from iPSC-derived cardiomyocytes (CMs). OBJECTIVES: For the current research, 419 hypertrophic cardiomyopathy (HCM) transient signals and 228 long QT syndrome (LQTS) transient signals were measured. HCM signals included data recorded from iPSC-CMs carrying either α-tropomyosin, i.e., TPM1 (HCMT) or MYBPC3 or myosin-binding protein C (HCMM) mutation and LQTS signals included data recorded from iPSC-CMs carrying potassium voltage-gated channel subfamily Q member 1 (KCNQ1) mutation (long QT syndrome 1 [LQT1]) or KCNH2 mutation (long QT syndrome 2 [LQT2]). The main objective was to study whether and how effectively HCMM and HCMT can be separated from each other as well as LQT1 from LQT2. METHODS: After preprocessing those Ca2+ signals where we computed peak waveforms we then classified the two mutations of both disease pairs by using several different machine learning methods. RESULTS: We obtained excellent classification accuracies of 89% for HCM and even 100% for LQT at their best. CONCLUSION: The results indicate that the methods applied would be efficient for the identification of these genetic cardiac diseases.

Detection of genetic cardiac diseases by Ca2+ transient profiles using machine learning methods.

Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) have revolutionized cardiovascular research. Abnormalities in Ca2+ transients have been evident in many cardiac disease models. We have shown earlier that, by exploiting computational machine learning methods, normal Ca2+ transients corresponding to healthy CMs can be distinguished from diseased CMs with abnormal transients. Here our aim was to study whether it is possible to separate different genetic cardiac diseases (CPVT, LQT, HCM) on the basis of Ca2+ transients using machine learning methods. Classification accuracies of up to 87% were obtained for these three diseases, indicating that Ca2+ transients are disease-specific. By including healthy controls in the classifications, the best classification accuracy obtained was still high: approximately 79%. In conclusion, we demonstrate as the proof of principle that the computational machine learning methodology appears to be a powerful means to accurately categorize iPSC-CMs and could provide effective methods for diagnostic purposes in the future.

Error-Correcting Output Codes in Classification of Human Induced Pluripotent Stem Cell Colony Images.

The purpose of this paper is to examine how well the human induced pluripotent stem cell (hiPSC) colony images can be classified using error-correcting output codes (ECOC). Our image dataset includes hiPSC colony images from three classes (bad, semigood, and good) which makes our classification task a multiclass problem. ECOC is a general framework to model multiclass classification problems. We focus on four different coding designs of ECOC and apply to each one of them k-Nearest Neighbor (k-NN) searching, naïve Bayes, classification tree, and discriminant analysis variants classifiers. We use Scaled Invariant Feature Transformation (SIFT) based features in classification. The best accuracy (62.4%) is obtained with ternary complete ECOC coding design and k-NN classifier (standardized Euclidean distance measure and inverse weighting). The best result is comparable with our earlier research. The quality identification of hiPSC colony images is an essential problem to be solved before hiPSCs can be used in practice in large-scale. ECOC methods examined are promising techniques for solving this challenging problem.

Machine Learning Approach to Automated Quality Identification of Human Induced Pluripotent Stem Cell Colony Images.

The focus of this research is on automated identification of the quality of human induced pluripotent stem cell (iPSC) colony images. iPS cell technology is a contemporary method by which the patient's cells are reprogrammed back to stem cells and are differentiated to any cell type wanted. iPS cell technology will be used in future to patient specific drug screening, disease modeling, and tissue repairing, for instance. However, there are technical challenges before iPS cell technology can be used in practice and one of them is quality control of growing iPSC colonies which is currently done manually but is unfeasible solution in large-scale cultures. The monitoring problem returns to image analysis and classification problem. In this paper, we tackle this problem using machine learning methods such as multiclass Support Vector Machines and several baseline methods together with Scaled Invariant Feature Transformation based features. We perform over 80 test arrangements and do a thorough parameter value search. The best accuracy (62.4%) for classification was obtained by using a k-NN classifier showing improved accuracy compared to earlier studies.

Signal analysis and classification methods for the calcium transient data of stem cell-derived cardiomyocytes.

Calcium cycling is crucial in the excitation-contraction coupling of cardiomyocytes, and therefore has a key role in cardiac functionality. Cardiac disorders and different drugs alter the calcium transients of cardiomyocytes and can cause serious dysfunction of the heart. New insights into this biochemical phenomena can be achieved by studying and analyzing calcium transients. Calcium transients of spontaneously beating human induced pluripotent stem cell-derived cardiomyocytes were recorded for a data set of 280 signals. Our objective was to develop and program procedures: (1) to automatically detect cycling peaks from signals and to classify the peaks of signals as either normal or abnormal, and (2) on the basis of the preceding peak detection results, to classify the entire signals into either a normal class or an abnormal class. We obtained a classification accuracy of approximately 80% compared to class decisions made separately by an experienced researcher, which is promising for the further development of an automatic classification approach. Automated classification software would be beneficial in the future for analyzing cardiomyocyte functionality on a large scale when screening for the adverse cardiac effects of new potential compounds, and also in future clinical applications.

On computation of calcium cycling anomalies in cardiomyocytes data.

Induced pluripotent stem cell (iPSC) lines derived from skin fibroblasts of patients suffering from cardiac disorders were differentiated to cardiomyocytes and used to generate a data set of Ca(2+) transients of 136 recordings. The objective was to separate normal signals for later medical research from abnormal signals. We constructed a signal analysis procedure to detect peaks representing calcium cycling in signals and another procedure to classify them into either normal or abnormal peaks. Using machine learning methods we classified signals into normal or abnormal signals on the basis of peak findings in them. We compared classification results obtained to those made visually by an expert biotechnologist who assessed the signals independent of the computer method. Classification accuracies of around 85% indicated high congruence between two modes denoting the high capability and usefulness of computer based processing for the present data.

Investigating local spatially-enhanced structural and textural descriptors for classification of iPSC colony images.

Induced pluripotent stem cells (iPSC) can be derived from fully differentiated cells of adult individuals and used to obtain any other cell type of the human body. This implies numerous prospective applications of iPSCs in regenerative medicine and drug development. In order to obtain valid cell culture, a quality control process must be applied to identify and discard abnormal iPSC colonies. Computer vision systems that analyze visual characteristics of iPSC colony health can be especially useful in automating and improving the quality control process. In this paper, we present an ongoing research that aims at the development of local spatially-enhanced descriptors for classification of iPSC colony images. For this, local oriented edges and local binary patterns are extracted from the detected colony regions and used to represent structural and textural properties of the colonies, respectively. We preliminary tested the proposed descriptors in classifying iPSCs colonies according to the degree of colony abnormality. The tests showed promising results for both, detection of iPSC colony borders and colony classification.

Machine learning approach to an otoneurological classification problem.

In this paper we applied altogether 13 classification methods to otoneurological disease classification. The main point was to use Half-Against-Half (HAH) architecture in classification. HAH structure was used with Support Vector Machines (SVMs), k-Nearest Neighbour (k-NN) method and Naïve Bayes (NB) methods. Furthermore, Multinomial Logistic Regression (MNLR) was tested for the dataset. HAH-SVM with the linear kernel achieved clearly the best accuracy being 76.9% which was a good result with the dataset tested. From the other classification methods HAH-k-NN with cityblock metric, HAH-NB and MNLR methods achieved above 60% accuracy. Around 77% accuracy is a good result compared to previous researches with the same dataset.

Half-Against-Half structure in classification of benthic macroinvertebrate images.

Benthic macroinvertebrates play a key role when water quality assessments are made. Benthic macroinvertebrates are difficult to identify and their identification need special expertise. Furthermore, manual identification is slow and expensive process. This paper concerns benthic macroinverte-brate classification when Half-Against-Half (HAH) structure was applied to Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA) and Minimum Mahalanobis Distance Classifier (MMDC) classifiers. Especially, LDA, QDA and MMDC classifiers were for first time applied with HAH structure to benthic macroinvertebrate classification. We performed thorough experiments altogether with ten methods. In the case of HAH-SVM we managed to improve classification results from the earlier research by using a different approach to class division problem. We obtained 96.1% classification accuracy with Radial Basis Function (RBF) kernel. Moreover, new variants of LDA, QDA and MMDC classification methods achieved 89.5% and 91.6% classification accuracies which can be considered as a good result in such a difficult classification task.

Prediction of a state of a subject on the basis of a stabilogram signal and video oculography test.

Postural stability decreases with ageing and may lead to accidental falls, isolation and a reduction in the quality of life. The age at the onset of postural derangement, its extent and the reason for deterioration are poorly known within an individual, but in general it becomes more severe with age. In order to prevent falls and avoid severe injuries the postural derangement has to be noticed by the person and the possible nursing personnel. In this work we propose such numerical features, which can discriminate the persons having good or poor postural stability. These features can also be utilized to measure the outcome and progression of balance training. With these postural stability algorithms providing stability features for a subject we managed to classify correctly the type of stance on the force platform in more than 80% of sixty subjects. We used k-nearest neighbor algorithm as an intuitive baseline method and compared its results with those of support vector machines and hidden Markov models.

Applying one-vs-one and one-vs-all classifiers in k-nearest neighbour method and support vector machines to an otoneurological multi-class problem.

We studied how the splitting of a multi-class classification problem into multiple binary classification tasks, like One-vs-One (OVO) and One-vs-All (OVA), affects the predictive accuracy of disease classes. Classifiers were tested with an otoneurological data using 10-fold cross-validation 10 times with k-Nearest Neighbour (k-NN) method and Support Vector Machines (SVM). The results showed that the use of multiple binary classifiers improves the classification accuracies of disease classes compared to one multi-class classifier. In general, OVO classifiers worked out better with this data than OVA classifiers. Especially, the OVO with k-NN yielded the highest total classification accuracies.