Investigating the Correspondence of Clinical Diagnostic Grouping With Underlying Neurobiological and Phenotypic Clusters Using Unsupervised Machine Learning

Attention-Like Multimodality Fusion With Data Augmentation for Diagnosis of Mental Disorders Using MRI

The globally rising prevalence of mental disorders leads to shortfalls in timely diagnosis and therapy to reduce patients' suffering. Facing such an urgent public health problem, professional efforts based on symptom criteria are seriously overstretched. Recently, the successful applications of computer-aided diagnosis approaches have provided timely opportunities to relieve the tension in healthcare services. Particularly, multimodal representation learning gains increasing attention thanks to the high temporal and spatial resolution information extracted from neuroimaging fusion. In this work, we propose an efficient multimodality fusion framework to identify multiple mental disorders based on the combination of functional and structural magnetic resonance imaging. A multioutput conditional generative adversarial network (GAN) is developed to address the scarcity of multimodal data for augmentation. Based on the augmented training data, the multiheaded gating fusion model is proposed for classification by extracting the complementary features across different modalities. The experiments demonstrate that the proposed model can achieve robust accuracies of 75.1 ± 1.5 %, 72.9 ± 1.1 %, and 87.2 ± 1.5 % for autism spectrum disorder (ASD), attention deficit/hyperactivity disorder, and schizophrenia, respectively. In addition, the interpretability of our model is expected to enable the identification of remarkable neuropathology diagnostic biomarkers, leading to well-informed therapeutic decisions.

One Size Does Not Fit All: Methodological Considerations for Brain-Based Predictive Modeling in Psychiatry

Psychiatric illnesses are heterogeneous in nature. No illness manifests in the same way across individuals, and no two patients with a shared diagnosis exhibit identical symptom profiles. Over the last several decades, group-level analyses of in vivo neuroimaging data have led to fundamental advances in our understanding of the neurobiology of psychiatric illnesses. More recently, access to computational resources and large, publicly available datasets alongside the rise of predictive modeling and precision medicine approaches have facilitated the study of psychiatric illnesses at an individual level. Data-driven machine learning analyses can be applied to identify disease-relevant biological subtypes, predict individual symptom profiles, and recommend personalized therapeutic interventions. However, when developing these predictive models, methodological choices must be carefully considered to ensure accurate, robust, and interpretable results. Choices pertaining to algorithms, neuroimaging modalities and states, data transformation, phenotypes, parcellations, sample sizes, and populations we are specifically studying can influence model performance. Here, we review applications of neuroimaging-based machine learning models to study psychiatric illnesses and discuss the effects of different methodological choices on model performance. An understanding of these effects is crucial for the proper implementation of predictive models in psychiatry and will facilitate more accurate diagnoses, prognoses, and therapeutics.

Machine learning approaches for parsing comorbidity/heterogeneity in antisociality and substance use disorders: A primer

By some accounts, as many as 93% of individuals diagnosed with antisocial personality disorder (ASPD) or psychopathy also meet criteria for some form of substance use disorder (SUD). This high level of comorbidity, combined with an overlapping biopsychosocial profile, and potentially interacting features, has made it difficult to delineate the shared/unique characteristics of each disorder. Moreover, while rarely acknowledged, both SUD and antisociality exist as highly heterogeneous disorders in need of more targeted parcellation. While emerging data-driven nosology for psychiatric disorders (e.g., Research Domain Criteria (RDoC), Hierarchical Taxonomy of Psychopathology (HiTOP)) offers the opportunity for a more systematic delineation of the externalizing spectrum, the interrogation of large, complex neuroimaging-based datasets may require data-driven approaches that are not yet widely employed in psychiatric neuroscience. With this in mind, the proposed article sets out to provide an introduction into machine learning methods for neuroimaging that can help parse comorbid, heterogeneous externalizing samples. The modest machine learning work conducted to date within the externalizing domain demonstrates the potential utility of the approach but remains highly nascent. Within the paper, we make suggestions for how future work can make use of machine learning methods, in combination with emerging psychiatric nosology systems, to further diagnostic and etiological understandings of the externalizing spectrum. Finally, we briefly consider some challenges that will need to be overcome to encourage further progress in the field.

Functional Connectivity-Based Prediction of Autism on Site Harmonized ABIDE Dataset

OBJECTIVE

The larger sample sizes available from multi-site publicly available neuroimaging data repositories makes machine-learning based diagnostic classification of mental disorders more feasible by alleviating the curse of dimensionality. However, since multi-site data are aggregated post-hoc, i.e. they were acquired from different scanners with different acquisition parameters, non-neural inter-site variability may mask inter-group differences that are at least in part neural in origin. Hence, the advantages gained by the larger sample size in the context of machine-learning based diagnostic classification may not be realized.

METHODS

We address this issue using harmonization of multi-site neuroimaging data using the ComBat technique, which is based on an empirical Bayes formulation to remove inter-site differences in data distributions, to improve diagnostic classification accuracy. Specifically, we demonstrate this using ABIDE (Autism Brain Imaging Data Exchange) multi-site data for classifying individuals with Autism from healthy controls using resting state fMRI-based functional connectivity data.

RESULTS

Our results show that higher classification accuracies across multiple classification models can be obtained (especially for models based on artificial neural networks) from multi-site data post harmonization with the ComBat technique as compared to without harmonization, outperforming earlier results from existing studies using ABIDE. Furthermore, our network ablation analysis facilitated important insights into autism spectrum disorder pathology and the connectivity in networks shown to be important for classification covaried with verbal communication impairments in Autism.

CONCLUSION

Multi-site data harmonization using ComBat improves neuroimaging-based diagnostic classification of mental disorders.

SIGNIFICANCE

ComBat has the potential to make AI-based clinical decision-support systems more feasible in psychiatry.

Neurofunctional Segmentation Shifts in the Hippocampus

The hippocampus is one of the most phylogenetically preserved structures in the mammalian brain. Engaged in a host of diverse cognitive processes, there has been increasing interest in understanding how the hippocampus dynamically supports these functions. One of the lingering questions is how to reconcile the seemingly disparate cytoarchitectonic organization, which favors a dorsal-ventral layering, with the neurofunctional topography, which has strong support for longitudinal axis (anterior-posterior) and medial-lateral orientation. More recently, meta-analytically driven (e.g., big data) approaches have been employed, however, the question remains whether they are sensitive to important task-specific features such as context, cognitive processes recruited, or the type of stimulus being presented. Here, we used hierarchical clustering on functional magnetic resonance imaging (fMRI) data acquired from healthy individuals at 7T using a battery of tasks that engage the hippocampus to determine whether stimulus or task features influence cluster profiles in the left and right hippocampus. Our data suggest that resting state clustering appears to favor the cytoarchitectonic organization, while task-based clustering favors the neurofunctional clustering. Furthermore, encoding tasks were more sensitive to stimulus type than were recognition tasks. Interestingly, a face-name paired associate task had nearly identical clustering profiles for both the encoding and recognition conditions of the task, which were qualitatively morphometrically different than simple encoding of words or faces. Finally, corroborating previous research, the left hippocampus had more stable cluster profiles compared to the right hippocampus. Together, our data suggest that task-based and resting state cluster profiles are different and may account for the disparity or inconsistency in results across studies.

A Unified Framework for Inattention Estimation From Resting State Phase Synchrony Using Machine Learning

Inattention is one of the most significant clinical symptoms for evaluating attention deficit hyperactivity disorder (ADHD). Previous inattention estimations were performed using clinical scales. Recently, predictive models for inattention have been established for brain-behavior estimation using neuroimaging features. However, the performance of inattention estimation could be improved for conventional brain-behavior models with additional feature selection, machine learning algorithms, and validation procedures. This paper aimed to propose a unified framework for inattention estimation from resting state fMRI to improve the classical brain-behavior models. Phase synchrony was derived as raw features, which were selected with minimum-redundancy maximum-relevancy (mRMR) method. Six machine learning algorithms were applied as regression methods. 100 runs of 10-fold cross-validations were performed on the ADHD-200 datasets. The relevance vector machines (RVMs) based on the mRMR features for the brain-behavior models significantly improve the performance of inattention estimation. The mRMR-RVM models could achieve a total accuracy of 0.53. Furthermore, predictive patterns for inattention were discovered by the mRMR technique. We found that the bilateral subcortical-cerebellum networks exhibited the most predictive phase synchrony patterns for inattention. Together, an optimized strategy named mRMR-RVM for brain-behavior models was found for inattention estimation. The predictive patterns might help better understand the phase synchrony mechanisms for inattention.

Detecting Risk Gene and Pathogenic Brain Region in EMCI Using a Novel GERF Algorithm Based on Brain Imaging and Genetic Data

Fusion analysis of disease-related multi-modal data is becoming increasingly important to illuminate the pathogenesis of complex brain diseases. However, owing to the small amount and high dimension of multi-modal data, current machine learning methods do not fully achieve the high veracity and reliability of fusion feature selection. In this paper, we propose a genetic-evolutionary random forest (GERF) algorithm to discover the risk genes and disease-related brain regions of early mild cognitive impairment (EMCI) based on the genetic data and resting-state functional magnetic resonance imaging (rs-fMRI) data. Classical correlation analysis method is used to explore the association between brain regions and genes, and fusion features are constructed. The genetic-evolutionary idea is introduced to enhance the classification performance, and to extract the optimal features effectively. The proposed GERF algorithm is evaluated by the public Alzheimer's Disease Neuroimaging Initiative (ADNI) database, and the results show that the algorithm achieves satisfactory classification accuracy in small sample learning. Moreover, we compare the GERF algorithm with other methods to prove its superiority. Furthermore, we propose the overall framework of detecting pathogenic factors, which can be accurately and efficiently applied to the multi-modal data analysis of EMCI and be able to extend to other diseases. This work provides a novel insight for early diagnosis and clinicopathologic analysis of EMCI, which facilitates clinical medicine to control further deterioration of diseases and is good for the accurate electric shock using transcranial magnetic stimulation.

Unsupervised Machine Learning to Identify Separable Clinical Alzheimer's Disease Sub-Populations

Heterogeneity among Alzheimer's disease (AD) patients confounds clinical trial patient selection and therapeutic efficacy evaluation. This work defines separable AD clinical sub-populations using unsupervised machine learning. Clustering (t-SNE followed by k-means) of patient features and association rule mining (ARM) was performed on the ADNIMERGE dataset from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Patient sociodemographics, brain imaging, biomarkers, cognitive tests, and medication usage were included for analysis. Four AD clinical sub-populations were identified using between-cluster mean fold changes [cognitive performance, brain volume]: cluster-1 represented least severe disease [+17.3, +13.3]; cluster-0 [-4.6, +3.8] and cluster-3 [+10.8, -4.9] represented mid-severity sub-populations; cluster-2 represented most severe disease [-18.4, -8.4]. ARM assessed frequently occurring pharmacologic substances within the 4 sub-populations. No drug class was associated with the least severe AD (cluster-1), likely due to lesser antecedent disease. Anti-hyperlipidemia drugs associated with cluster-0 (mid-severity, higher volume). Interestingly, antioxidants vitamin C and E associated with cluster-3 (mid-severity, higher cognition). Anti-depressants like Zoloft associated with most severe disease (cluster-2). Vitamin D is protective for AD, but ARM identified significant underutilization across all AD sub-populations. Identification and feature characterization of four distinct AD sub-population "clusters" using standard clinical features enhances future clinical trial selection criteria and cross-study comparative analysis.

Validation strategies for subtypes in psychiatry: A systematic review of research on autism spectrum disorder

Heterogeneity within autism spectrum disorder (ASD) is recognized as a challenge to both biological and psychological research, as well as clinical practice. To reduce unexplained heterogeneity, subtyping techniques are often used to establish more homogeneous subtypes based on metrics of similarity and dissimilarity between people. We review the ASD literature to create a systematic overview of the subtyping procedures and subtype validation techniques that are used in this field. We conducted a systematic review of 156 articles (2001-June 2020) that subtyped participants (range N of studies = 17-20,658), of which some or all had an ASD diagnosis. We found a large diversity in (parametric and non-parametric) methods and (biological, psychological, demographic) variables used to establish subtypes. The majority of studies validated their subtype results using variables that were measured concurrently, but were not included in the subtyping procedure. Other investigations into subtypes' validity were rarer. In order to advance clinical research and the theoretical and clinical usefulness of identified subtypes, we propose a structured approach and present the SUbtyping VAlidation Checklist (SUVAC), a checklist for validating subtyping results.

Deep multi-kernel auto-encoder network for clustering brain functional connectivity data

In this study, we propose a deep-learning network model called the deep multi-kernel auto-encoder clustering network (DMACN) for clustering functional connectivity data for brain diseases. This model is an end-to-end clustering algorithm that can learn potentially advanced features and cluster disease categories. Unlike other auto-encoders, DMACN has an added self-expression layer and standard back-propagation is used to learn the features that are beneficial for clustering brain functional connectivity data. In the self-expression layer, the kernel matrix is constructed to extract effective features and a new loss function is proposed to constrain the clustering portion, which enables the training of a deep neural learning network that tends to cluster. To test the performance of the proposed algorithm, we applied the end-to-end deep unsupervised clustering algorithm to brain connectivity data. We then conducted experiments based on four public brain functional connectivity data sets and our own functional connectivity data set. The DMACN algorithm yielded good results in various evaluations compared with the existing clustering algorithm for brain functional connectivity data, the deep auto-encoder clustering algorithm, and several other relevant clustering algorithms. The deep-learning-based clustering algorithm has great potential for use in the unsupervised recognition of brain diseases.

Current challenges and possible future developments in personalized psychiatry with an emphasis on psychotic disorders

A personalized medicine approach seems to be particularly applicable to psychiatry. Indeed, considering mental illness as deregulation, unique to each patient, of molecular pathways, governing the development and functioning of the brain, seems to be the most justified way to understand and treat disorders of this medical category. In order to extract correct information about the implicated molecular pathways, data can be drawn from sampling phenotypic and genetic biomarkers and then analyzed by a machine learning algorithm. This review describes current difficulties in the field of personalized psychiatry and gives several examples of possibly actionable biomarkers of psychotic and other psychiatric disorders, including several examples of genetic studies relevant to personalized psychiatry. Most of these biomarkers are not yet ready to be introduced in clinical practice. In a next step, a perspective on the path personalized psychiatry may take in the future is given, paying particular attention to machine learning algorithms that can be used with the goal of handling multidimensional datasets.

Multi-Level Clustering of Dynamic Directional Brain Network Patterns and Their Behavioral Relevance

Dynamic functional connectivity (DFC) obtained from resting state functional magnetic resonance imaging (fMRI) data has been shown to provide novel insights into brain function which may be obscured by static functional connectivity (SFC). Further, DFC, and by implication how different brain regions may engage or disengage with each other over time, has been shown to be behaviorally relevant and more predictive than SFC of behavioral performance and/or diagnostic status. DFC is not a directional entity and may capture neural synchronization. However, directional interactions between different brain regions is another putative mechanism by which neural populations communicate. Accordingly, static effective connectivity (SEC) has been explored as a means of characterizing such directional interactions. But investigation of its dynamic counterpart, i.e., dynamic effective connectivity (DEC), is still in its infancy. Of particular note are methodological insufficiencies in identifying DEC configurations that are reproducible across time and subjects as well as a lack of understanding of the behavioral relevance of DEC obtained from resting state fMRI. In order to address these issues, we employed a dynamic multivariate autoregressive (MVAR) model to estimate DEC. The method was first validated using simulations and then applied to resting state fMRI data obtained in-house (N = 21), wherein we performed dynamic clustering of DEC matrices across multiple levels [using adaptive evolutionary clustering (AEC)] – spatial location, time, and subjects. We observed a small number of directional brain network configurations alternating between each other over time in a quasi-stable manner akin to brain microstates. The dominant and consistent DEC network patterns involved several regions including inferior and mid temporal cortex, motor and parietal cortex, occipital cortex, as well as part of frontal cortex. The functional relevance of these DEC states were determined using meta-analyses and pertained mainly to memory and emotion, but also involved execution and language. Finally, a larger cohort of resting-state fMRI and behavioral data from the Human Connectome Project (HCP) (N = 232, Q1–Q3 release) was used to demonstrate that metrics derived from DEC can explain larger variance in 70 behaviors across different domains (alertness, cognition, emotion, and personality traits) compared to SEC in healthy individuals.

Artificial intelligence as an emerging technology in the current care of neurological disorders

BACKGROUND

Artificial intelligence (AI) has influenced all aspects of human life and neurology is no exception to this growing trend. The aim of this paper is to guide medical practitioners on the relevant aspects of artificial intelligence, i.e., machine learning, and deep learning, to review the development of technological advancement equipped with AI, and to elucidate how machine learning can revolutionize the management of neurological diseases. This review focuses on unsupervised aspects of machine learning, and how these aspects could be applied to precision neurology to improve patient outcomes. We have mentioned various forms of available AI, prior research, outcomes, benefits and limitations of AI, effective accessibility and future of AI, keeping the current burden of neurological disorders in mind.

DISCUSSION

The smart device system to monitor tremors and to recognize its phenotypes for better outcomes of deep brain stimulation, applications evaluating fine motor functions, AI integrated electroencephalogram learning to diagnose epilepsy and psychological non-epileptic seizure, predict outcome of seizure surgeries, recognize patterns of autonomic instability to prevent sudden unexpected death in epilepsy (SUDEP), identify the pattern of complex algorithm in neuroimaging classifying cognitive impairment, differentiating and classifying concussion phenotypes, smartwatches monitoring atrial fibrillation to prevent strokes, and prediction of prognosis in dementia are unique examples of experimental utilizations of AI in the field of neurology. Though there are obvious limitations of AI, the general consensus among several nationwide studies is that this new technology has the ability to improve the prognosis of neurological disorders and as a result should become a staple in the medical community.

CONCLUSION

AI not only helps to analyze medical data in disease prevention, diagnosis, patient monitoring, and development of new protocols, but can also assist clinicians in dealing with voluminous data in a more accurate and efficient manner.

Artificial intelligence for understanding concussion: Retrospective cluster analysis on the balance and vestibular diagnostic data of concussion patients

OBJECTIVES

We propose a bottom-up, machine-learning approach, for the objective vestibular and balance diagnostic data of concussion patients, to provide insight into the differences in patients' phenotypes, independent of existing diagnoses (unsupervised learning).

METHODS

Diagnostic data from a battery of validated balance and vestibular assessments were extracted from the database of the Swiss Concussion Center. The desired number of clusters within the patient database was estimated using Calinski-Harabasz criteria. Complex (self-organizing map, SOM) and standard (k-means) clustering tools were used, and the formed clusters were compared.

RESULTS

A total of 96 patients (81.3% male, age (median [IQR]): 25.0[10.8]) who were expected to suffer from sports-related concussion or post-concussive syndrome (52[140] days between diagnostic testing and the concussive episode) were included. The cluster evaluation indicated dividing the data into two groups. Only the SOM gave a stable clustering outcome, dividing the patients in group-1 (n = 38) and group-2 (n = 58). A large significant difference was found for the caloric summary score for the maximal speed of the slow phase, where group-1 scored 30.7% lower than group-2 (27.6[18.2] vs. 51.0[31.0]). Group-1 also scored significantly lower on the sensory organisation test composite score (69.0[22.3] vs. 79.0[10.5]) and higher on the visual acuity (-0.03[0.33] vs. -0.14[0.12]) and dynamic visual acuity (0.38[0.84] vs. 0.20[0.20]) tests. The importance of caloric, SOT and DVA, was supported by the PCA outcomes. Group-1 tended to report headaches, blurred vision and balance problems more frequently than group-2 (>10% difference).

CONCLUSION

The SOM divided the data into one group with prominent vestibular disorders and another with no clear vestibular or balance problems, suggesting that artificial intelligence might help improve the diagnostic process.

Identifying neuropsychiatric disorders using unsupervised clustering methods: Data and code

This article provides data for five different neuropsychiatric disorders-Attention Deficit Hyperactivity Disorder, Alzheimer's Disease, Autism Spectrum Disorder, Post-Traumatic Stress Disorder, and Post-Concussion Syndrome-along with healthy controls. The data includes clinical diagnostic labels, phenotypic variables, and resting-state functional magnetic resonance imaging connectivity features obtained from individuals. In addition, it provides the source MATLAB codes used for data analyses. Three existing clustering methods have been incorporated into the provided code, which do not require a priori specification of the number of clusters. A genetic algorithm based feature selection method has also been included to find the relevant subset of features and clustering the subset of data simultaneously. Findings from this data set and further detailed interpretations are available in our recent research study (Zhao et al., 2017) [1]. This contribution is a valuable asset for performing unsupervised machine learning on fMRI data to investigate the correspondence of clinical diagnostic grouping with the underlying neurobiological/phenotypic clusters.