Search

Understanding the performance and validity of clustering algorithms is both challenging and crucial, particularly when clustering must be done online. Until recently, most validation methods have relied on batch calculation and have required considerable human expertise in their interpretation. Improving real-time performance and interpretability of cluster validation, therefore, continues to be an important theme in unsupervised learning. Building upon previous work on incremental cluster validity indices (iCVIs), this paper introduces the Meta- iCVI as a tool for explainable and concise labeling of partition quality in online clustering. Leveraging a time-series classifier and data-fusion techniques, the Meta- iCVI combines the outputs of multiple iCVIs to produce a streaming label of either “over”, “under”, or “correctly” partitioned. Experiments were conducted on generalized synthetic and real-world data sets to demonstrate the efficacy and application of this method. Results of 100% accuracy were achieved in labeling partition quality on real-world data sets including MNIST and FLIR ADAS, demonstrating that the Meta- iCVI is a powerful and efficient tool for classifying partition quality in a variety of conditions. Its introduction should empower new and more efficient streaming clustering techniques. Additionally, we believe this to be the first implementation of an ensemble iCVI metric and the first time iCVI validation performance has been evaluated on randomized sample presentation.

In sequential estimation methods often used in oceanic and general climate calculations of the state and of forecasts, observations act mathematically and statistically as source or sink terms in conservation equations for heat, salt, mass, and momentum. These artificial terms obscure the inference of the system's variability or secular changes. Furthermore, for the purposes of calculating changes in important functions of state variables such as total mass and energy or volumetric current transports, results of both filter and smoother-based estimates are sensitive to misrepresentation of a large variety of parameters, including initial conditions, prior uncertainty covariances, and systematic and random errors in observations. Here, toy models of a coupled mass–spring oscillator system and of a barotropic Rossby wave system are used to demonstrate many of the issues that arise from such misrepresentations. Results from Kalman filter estimates and those from finite interval smoothing are analyzed. In the filter (and prediction) problem, entry of data leads to violation of conservation and other invariant rules. A finite interval smoothing method restores the conservation rules, but uncertainties in all such estimation results remain. Convincing trend and other time-dependent determinations in “reanalysis-like” estimates require a full understanding of models, observations, and underlying error structures. Application of smoother-type methods that are designed for optimal reconstruction purposes alleviate some of the issues.

Biomedical datasets distill many mechanisms of human diseases, linking diseases to genes and phenotypes (signs and symptoms of disease), genetic mutations to altered protein structures, and altered proteins to changes in molecular functions and biological processes. It is desirable to gain new insights from these data, especially with regard to the uncovering of hierarchical structures relating disease variants. However, analysis to this end has proven difficult due to the complexity of the connections between multi-categorical symbolic data. This article proposes symbolic tree adaptive resonance theory (START), with additional supervised, dual-vigilance (DV-START), and distributed dual-vigilance (DDV-START) formulations, for the clustering of multi-categorical symbolic data from biomedical datasets by demonstrating its utility in clustering variants of Charcot–Marie–Tooth disease using genomic, phenotypic, and proteomic data.

Disease phenotypes are characterized by signs (what a physician observes during the examination of a patient) and symptoms (the complaints of a patient to a physician). Large repositories of disease phenotypes are accessible through the Online Mendelian Inheritance of Man, Human Phenotype Ontology, and Orphadata initiatives. Many of the diseases in these datasets are neurologic. For each repository, the phenotype of neurologic disease is represented as a list of concepts of variable length where the concepts are selected from a restricted ontology. Visualizations of these concept lists are not provided. We address this limitation by using subsumption to reduce the number of descriptive features from 2,946 classes into thirty superclasses. Phenotype feature lists of variable lengths were converted into fixed-length vectors. Phenotype vectors were aggregated into matrices and visualized as heat maps that allowed side-by-side disease comparisons. Individual diseases (representing a row in the matrix) were visualized as word clouds. We illustrate the utility of this approach by visualizing the neuro-phenotypes of 32 dystonic diseases from Orphadata. Subsumption can collapse phenotype features into superclasses, phenotype lists can be vectorized, and phenotypes vectors can be visualized as heat maps and word clouds.

We used network analysis to identify subtypes of relapsing-remitting multiple sclerosis subjects based on their cumulative signs and symptoms. The electronic medical records of 113 subjects with relapsing-remitting multiple sclerosis were reviewed, signs and symptoms were mapped to classes in a neuro-ontology, and classes were collapsed into sixteen superclasses by subsumption. After normalization and vectorization of the data, bipartite (subject-feature) and unipartite (subject-subject) network graphs were created using NetworkX and visualized in Gephi. Degree and weighted degree were calculated for each node. Graphs were partitioned into communities using the modularity score. Feature maps visualized differences in features by community. Network analysis of the unipartite graph yielded a higher modularity score (0.49) than the bipartite graph (0.25). The bipartite network was partitioned into five communities which were named fatigue, behavioral, hypertonia/weakness, abnormal gait/sphincter, and sensory, based on feature characteristics. The unipartite network was partitioned into five communities which were named fatigue, pain, cognitive, sensory, and gait/weakness/hypertonia based on features. Although we did not identify pure subtypes (e.g., pure motor, pure sensory, etc.) in this cohort of multiple sclerosis subjects, we demonstrated that network analysis could partition these subjects into different subtype communities. Larger datasets and additional partitioning algorithms are needed to confirm these findings and elucidate their significance. This study contributes to the literature investigating subtypes of multiple sclerosis by combining feature reduction by subsumption with network analysis.

Although deep learning has been applied to the recognition of diseases and drugs in electronic health records and the biomedical literature, relatively little study has been devoted to the utility of deep learning for the recognition of signs and symptoms. The recognition of signs and symptoms is critical to the success of deep phenotyping and precision medicine. We have developed a named entity recognition model that uses deep learning to identify text spans containing neurological signs and symptoms and then maps these text spans to the clinical concepts of a neuro-ontology. We compared a model based on convolutional neural networks to one based on bidirectional encoder representation from transformers. Models were evaluated for accuracy of text span identification on three text corpora: physician notes from an electronic health record, case histories from neurologic textbooks, and clinical synopses from an online database of genetic diseases. Both models performed best on the professionally-written clinical synopses and worst on the physician-written clinical notes. Both models performed better when signs and symptoms were represented as shorter text spans. Consistent with prior studies that examined the recognition of diseases and drugs, the model based on bidirectional encoder representations from transformers outperformed the model based on convolutional neural networks for recognizing signs and symptoms. Recall for signs and symptoms ranged from 59.5% to 82.0% and precision ranged from 61.7% to 80.4%. With further advances in NLP, fully automated recognition of signs and symptoms in electronic health records and the medical literature should be feasible.

Abstract Background The use of blood biomarkers after mild traumatic brain injury (mTBI) has been widely studied. We have identified eight unresolved issues related to the use of five commonly investigated blood biomarkers: neurofilament light chain, ubiquitin carboxy-terminal hydrolase-L1, tau, S100B, and glial acidic fibrillary protein. We conducted a focused literature review of unresolved issues in three areas: mode of entry into and exit from the blood, kinetics of blood biomarkers in the blood, and predictive capacity of the blood biomarkers after mTBI. Findings Although a disruption of the blood brain barrier has been demonstrated in mild and severe traumatic brain injury, biomarkers can enter the blood through pathways that do not require a breach in this barrier. A definitive accounting for the pathways that biomarkers follow from the brain to the blood after mTBI has not been performed. Although preliminary investigations of blood biomarkers kinetics after TBI are available, our current knowledge is incomplete and definitive studies are needed. Optimal sampling times for biomarkers after mTBI have not been established. Kinetic models of blood biomarkers can be informative, but more precise estimates of kinetic parameters are needed. Confounding factors for blood biomarker levels have been identified, but corrections for these factors are not routinely made. Little evidence has emerged to date to suggest that blood biomarker levels correlate with clinical measures of mTBI severity. The significance of elevated biomarker levels thirty or more days following mTBI is uncertain. Blood biomarkers have shown a modest but not definitive ability to distinguish concussed from non-concussed subjects, to detect sub-concussive hits to the head, and to predict recovery from mTBI. Blood biomarkers have performed best at distinguishing CT scan positive from CT scan negative subjects after mTBI.

The cytoskeletal protein tau is implicated in the pathogenesis of Alzheimer's disease which is characterized by intra-neuronal neurofibrillary tangles containing abnormally phosphorylated insoluble tau. Levels of soluble tau are elevated in the brain, the CSF, and the plasma of patients with Alzheimer's disease. To better understand the causes of these elevated levels of tau, we propose a three-compartment kinetic model (brain, CSF, and plasma). The model assumes that the synthesis of tau follows zero-order kinetics (uncorrelated with compartmental tau levels) and that the release, absorption, and clearance of tau is governed by first-order kinetics (linearly related to compartmental tau levels). Tau that is synthesized in the brain compartment can be released into the interstitial fluid, catabolized, or retained in neurofibrillary tangles. Tau released into the interstitial fluid can mix with the CSF and eventually drain to the plasma compartment. However, losses of tau in the drainage pathways may be significant. The kinetic model estimates half-life of tau in each compartment (552 h in the brain, 9.9 h in the CSF, and 10 h in the plasma). The kinetic model predicts that an increase in the neuronal tau synthesis rate or a decrease in tau catabolism rate best accounts for observed increases in tau levels in the brain, CSF, and plasma found in Alzheimer's disease. Furthermore, the model predicts that increases in brain half-life of tau in Alzheimer's disease should be attributed to decreased tau catabolism and not to increased tau synthesis. Most clearance of tau in the neuron occurs through catabolism rather than release to the CSF compartment. Additional experimental data would make ascertainment of the model parameters more precise.

Abstract Background Patient distances can be calculated based on signs and symptoms derived from an ontological hierarchy. There is controversy as to whether patient distance metrics that consider the semantic similarity between concepts can outperform standard patient distance metrics that are agnostic to concept similarity. The choice of distance metric can dominate the performance of classification or clustering algorithms. Our objective was to determine if semantically augmented distance metrics would outperform standard metrics on machine learning tasks. Methods We converted the neurological findings from 382 published neurology cases into sets of concepts with corresponding machine-readable codes. We calculated patient distances by four different metrics (cosine distance, a semantically augmented cosine distance, Jaccard distance, and a semantically augmented bipartite distance). Semantic augmentation for two of the metrics depended on concept similarities from a hierarchical neuro-ontology. For machine learning algorithms, we used the patient diagnosis as the ground truth label and patient findings as machine learning features. We assessed classification accuracy for four classifiers and cluster quality for two clustering algorithms for each of the distance metrics. Results Inter-patient distances were smaller when the distance metric was semantically augmented. Classification accuracy and cluster quality were not significantly different by distance metric. Conclusion Although semantic augmentation reduced inter-patient distances, we did not find improved classification accuracy or improved cluster quality with semantically augmented patient distance metrics when applied to a dataset of neurology patients. Further work is needed to assess the utility of semantically augmented patient distances.

Validation is one of the most important aspects of clustering, particularly when the user is designing a trustworthy or explainable system. However, most clustering validation approaches require batch calculation. This is an important gap because of the value of clustering in real-time data streaming and other online learning applications. Therefore, interest has grown in providing online alternatives for validation. This paper extends the incremental cluster validity index (iCVI) family by presenting incremental versions of Calinski-Harabasz (iCH), Pakhira-Bandyopadhyay-Maulik (iPBM), WB index (iWB), Silhouette (iSIL), Negentropy Increment (iNI), Representative Cross Information Potential (irCIP), Representative Cross Entropy (irH), and Conn_Index (iConn_Index). This paper also provides a thorough comparative study of correct, under- and over-partitioning on the behavior of these iCVIs, the Partition Separation (PS) index as well as four recently introduced iCVIs: incremental Xie-Beni (iXB), incremental Davies-Bouldin (iDB), and incremental generalized Dunn's indices 43 and 53 (iGD43 and iGD53). Experiments were carried out using a framework that was designed to be as agnostic as possible to the clustering algorithms. The results on synthetic benchmark data sets showed that while evidence of most under-partitioning cases could be inferred from the behaviors of the majority of these iCVIs, over-partitioning was found to be a more challenging problem, detected by fewer of them. Interestingly, over-partitioning, rather then under-partitioning, was more prominently detected on the real-world data experiments within this study. The expansion of iCVIs provides significant novel opportunities for assessing and interpreting the results of unsupervised lifelong learning in real-time, wherein samples cannot be reprocessed due to memory and/or application constraints.

We present a framework for an explainable and statistically validated ensemble clustering model applied to Traumatic Brain Injury (TBI). The objective of our analysis is to identify patient injury severity subgroups and key phenotypes that delineate these subgroups using varied clinical and computed tomography data. Explainable and statistically-validated models are essential because a data-driven identification of subgroups is an inherently multidisciplinary undertaking. In our case, this procedure yielded six distinct patient subgroups with respect to mechanism of injury, severity of presentation, anatomy, psychometric, and functional outcome. This framework for ensemble cluster analysis fully integrates statistical methods at several stages of analysis to enhance the quality and the explainability of results. This methodology is applicable to other clinical data sets that exhibit significant heterogeneity as well as other diverse data science applications in biomedicine and elsewhere.

Traumatic brain injury (TBI) imposes a significant economic and social burden. The diagnosis and prognosis of mild TBI, also called concussion, is challenging. Concussions are common among contact sport athletes. After a blow to the head, it is often difficult to determine who has had a concussion, who should be withheld from play, if a concussed athlete is ready to return to the field, and which concussed athlete will develop a post-concussion syndrome. Biomarkers can be detected in the cerebrospinal fluid and blood after traumatic brain injury and their levels may have prognostic value. Despite significant investigation, questions remain as to the trajectories of blood biomarker levels over time after mild TBI. Modeling the kinetic behavior of these biomarkers could be informative. We propose a one-compartment kinetic model for S100B, UCH-L1, NF-L, GFAP, and tau biomarker levels after mild TBI based on accepted pharmacokinetic models for oral drug absorption. We approximated model parameters using previously published studies. Since parameter estimates were approximate, we did uncertainty and sensitivity analyses. Using estimated kinetic parameters for each biomarker, we applied the model to an available post-concussion biomarker dataset of UCH-L1, GFAP, tau, and NF-L biomarkers levels. We have demonstrated the feasibility of modeling blood biomarker levels after mild TBI with a one compartment kinetic model. More work is needed to better establish model parameters and to understand the implications of the model for diagnostic use of these blood biomarkers for mild TBI.

This paper presents a distributed optimal model reference adaptive control approach for solving containment control of heterogeneous multi-agent systems (MASs) with non-autonomous leaders. First, a fully distributed adaptive observer is designed to provide for each agent the desired reference trajectory by estimating the convex hull spanned by leaders. The distributed observer dynamics serves as a reference model for each follower to synchronize. The global communication graph information or the leader dynamics is not required to design the observer. In contrast to existing model reference adaptive controllers (MRAC) for single-agent systems and containment control solutions for MASs, the proposed MRAC approach imposes optimality and presents a distributed adaptive optimal solution to the containment control problem. To impose optimality, a performance function is defined based on the adaptive observers' states as well as the followers' local measurements. It is shown that considering non-autonomous leaders in this optimal control problem leads to solving inhomogeneous algebraic Riccati equations (AREs), instead of normal AREs in standard optimal control problems. To obviate the requirement of knowing the agents' dynamics, an off-policy reinforcement learning approach implemented on an actor-critic structure is utilized for solving the inhomogeneous ARE. A simulation example is conducted to illustrate the effectiveness of the presented method.