Validation and application of computer vision algorithms for video-based tremor analysis.

Tremor is one of the most common neurological symptoms. Its clinical and neurobiological complexity necessitates novel approaches for granular phenotyping. Instrumented neurophysiological analyses have proven useful, but are highly resource-intensive and lack broad accessibility. In contrast, bedside scores are simple to administer, but lack the granularity to capture subtle but relevant tremor features. We utilise the open-source computer vision pose tracking algorithm Mediapipe to track hands in clinical video recordings and use the resulting time series to compute canonical tremor features. This approach is compared to marker-based 3D motion capture, wrist-worn accelerometry, clinical scoring and a second, specifically trained tremor-specific algorithm in two independent clinical cohorts. These cohorts consisted of 66 patients diagnosed with essential tremor, assessed in different task conditions and states of deep brain stimulation therapy. We find that Mediapipe-derived tremor metrics exhibit high convergent clinical validity to scores (Spearman's ρ = 0.55-0.86, p≤ .01) as well as an accuracy of up to 2.60 mm (95% CI [-3.13, 8.23]) and ≤0.21 Hz (95% CI [-0.05, 0.46]) for tremor amplitude and frequency measurements, matching gold-standard equipment. Mediapipe, but not the disease-specific algorithm, was capable of analysing videos involving complex configurational changes of the hands. Moreover, it enabled the extraction of tremor features with diagnostic and prognostic relevance, a dimension which conventional tremor scores were unable to provide. Collectively, this demonstrates that current computer vision algorithms can be transformed into an accurate and highly accessible tool for video-based tremor analysis, yielding comparable results to gold standard tremor recordings.

Head movement dynamics in dystonia: a multi-centre retrospective study using visual perceptive deep learning.

Dystonia is a neurological movement disorder characterised by abnormal involuntary movements and postures, particularly affecting the head and neck. However, current clinical assessment methods for dystonia rely on simplified rating scales which lack the ability to capture the intricate spatiotemporal features of dystonic phenomena, hindering clinical management and limiting understanding of the underlying neurobiology. To address this, we developed a visual perceptive deep learning framework that utilizes standard clinical videos to comprehensively evaluate and quantify disease states and the impact of therapeutic interventions, specifically deep brain stimulation. This framework overcomes the limitations of traditional rating scales and offers an efficient and accurate method that is rater-independent for evaluating and monitoring dystonia patients. To evaluate the framework, we leveraged semi-standardized clinical video data collected in three retrospective, longitudinal cohort studies across seven academic centres. We extracted static head angle excursions for clinical validation and derived kinematic variables reflecting naturalistic head dynamics to predict dystonia severity, subtype, and neuromodulation effects. The framework was also applied to a fully independent cohort of generalised dystonia patients for comparison between dystonia sub-types. Computer vision-derived measurements of head angle excursions showed a strong correlation with clinically assigned scores. Across comparisons, we identified consistent kinematic features from full video assessments encoding information critical to disease severity, subtype, and effects of neural circuit interventions, independent of static head angle deviations used in scoring. Our visual perceptive machine learning framework reveals kinematic pathosignatures of dystonia, potentially augmenting clinical management, facilitating scientific translation, and informing personalized precision neurology approaches.

A unified framework for simplicial Kuramoto models.

Simplicial Kuramoto models have emerged as a diverse and intriguing class of models describing oscillators on simplices rather than nodes. In this paper, we present a unified framework to describe different variants of these models, categorized into three main groups: "simple" models, "Hodge-coupled" models, and "order-coupled" (Dirac) models. Our framework is based on topology and discrete differential geometry, as well as gradient systems and frustrations, and permits a systematic analysis of their properties. We establish an equivalence between the simple simplicial Kuramoto model and the standard Kuramoto model on pairwise networks under the condition of manifoldness of the simplicial complex. Then, starting from simple models, we describe the notion of simplicial synchronization and derive bounds on the coupling strength necessary or sufficient for achieving it. For some variants, we generalize these results and provide new ones, such as the controllability of equilibrium solutions. Finally, we explore a potential application in the reconstruction of brain functional connectivity from structural connectomes and find that simple edge-based Kuramoto models perform competitively or even outperform complex extensions of node-based models.

The neuron mixer and its impact on human brain dynamics.

A signal mixer facilitates rich computation, which has been the building block of modern telecommunication. This frequency mixing produces new signals at the sum and difference frequencies of input signals, enabling powerful operations such as heterodyning and multiplexing. Here, we report that a neuron is a signal mixer. We found through ex vivo and in vivo whole-cell measurements that neurons mix exogenous (controlled) and endogenous (spontaneous) subthreshold membrane potential oscillations, producing new oscillation frequencies, and that neural mixing originates in voltage-gated ion channels. Furthermore, we demonstrate that mixing is evident in human brain activity and is associated with cognitive functions. We found that the human electroencephalogram displays distinct clusters of local and inter-region mixing and that conversion of the salient posterior alpha-beta oscillations into gamma-band oscillations regulates visual attention. Signal mixing may enable individual neurons to sculpt the spectrum of neural circuit oscillations and utilize them for computational operations.

Functionally annotated electrophysiological neuromarkers of healthy ageing and memory function.

The unprecedented increase in life expectancy presents a unique opportunity and the necessity to explore both healthy and pathological aspects of ageing. Electroencephalography (EEG) has been widely used to identify neuromarkers of cognitive ageing due to its affordability and richness in information. However, despite the growing volume of data and methodological advancements, the abundance of contradictory and non-reproducible findings has hindered clinical translation. To address these challenges, our study introduces a comprehensive workflow expanding on previous EEG studies and investigates various static and dynamic power and connectivity estimates as potential neuromarkers of cognitive ageing in a large dataset. We also assess the robustness of our findings by testing their susceptibility to band specification. Finally, we characterise our findings using functionally annotated brain networks to improve their interpretability and multi-modal integration. Our analysis demonstrates the effect of methodological choices on findings and that dynamic rather than static neuromarkers are not only more sensitive but also more robust. Consequently, they emerge as strong candidates for cognitive ageing neuromarkers. Moreover, we were able to replicate the most established EEG findings in cognitive ageing, such as alpha oscillation slowing, increased beta power, reduced reactivity across multiple bands, and decreased delta connectivity. Additionally, when considering individual variations in the alpha band, we clarified that alpha power is characteristic of memory performance rather than ageing, highlighting its potential as a neuromarker for cognitive ageing. Finally, our approach using functionally annotated source reconstruction allowed us to provide insights into domain-specific electrophysiological mechanisms underlying memory performance and ageing. HIGHLIGHTS: We provide an open and reproducible pipeline with a comprehensive workflow to investigate static and dynamic EEG neuromarkers. Neuromarkers related to neural dynamics are sensitive and robust. Individualised alpha power characterises cognitive performance rather than ageing. Functional annotation allows cross-modal interpretation of EEG findings.

Kernel-based joint independence tests for multivariate stationary and non-stationary time series.

Multivariate time-series data that capture the temporal evolution of interconnected systems are ubiquitous in diverse areas. Understanding the complex relationships and potential dependencies among co-observed variables is crucial for the accurate statistical modelling and analysis of such systems. Here, we introduce kernel-based statistical tests of joint independence in multivariate time series by extending the d-variable Hilbert-Schmidt independence criterion to encompass both stationary and non-stationary processes, thus allowing broader real-world applications. By leveraging resampling techniques tailored for both single- and multiple-realization time series, we show how the method robustly uncovers significant higher-order dependencies in synthetic examples, including frequency mixing data and logic gates, as well as real-world climate, neuroscience and socio-economic data. Our method adds to the mathematical toolbox for the analysis of multivariate time series and can aid in uncovering high-order interactions in data.

Wavelet-Based Bracketing, Time-Frequency Beta Burst Detection: New Insights in Parkinson's Disease.

Studies have shown that beta band activity is not tonically elevated but comprises exaggerated phasic bursts of varying durations and magnitudes, for Parkinson's disease (PD) patients. Current methods for detecting beta bursts target a single frequency peak in beta band, potentially ignoring bursts in the wider beta band. In this study, we propose a new robust framework for beta burst identification across wide frequency ranges. Chronic local field potential at-rest recordings were obtained from seven PD patients implanted with Medtronic SenSight™ deep brain stimulation (DBS) electrodes. The proposed method uses wavelet decomposition to compute the time-frequency spectrum and identifies bursts spanning multiple frequency bins by thresholding, offering an additional burst measure, ∆f, that captures the width of a burst in the frequency domain. Analysis included calculating burst duration, magnitude, and ∆f and evaluating the distribution and likelihood of bursts between the low beta (13-20 Hz) and high beta (21-35 Hz). Finally, the results of the analysis were correlated to motor impairment (MDS-UPDRS III) med off scores. We found that low beta bursts with longer durations and larger width in the frequency domain (∆f) were positively correlated, while high beta bursts with longer durations and larger ∆f were negatively correlated with motor impairment. The proposed method, finding clear differences between bursting behavior in high and low beta bands, has clearly demonstrated the importance of considering wide frequency bands for beta burst behavior with implications for closed-loop DBS paradigms.

Machine versus physician-based programming of deep brain stimulation in isolated dystonia: A feasibility study.

BACKGROUND: Deep brain stimulation of the internal globus pallidus effectively alleviates dystonia motor symptoms. However, delayed symptom control and a lack of therapeutic biomarkers and a single pallidal sweetspot region complicates optimal programming. Postoperative management is complex, typically requiring multiple, lengthy follow-ups with an experienced physician - an important barrier to widespread adoption in medication-refractory dystonia patients. OBJECTIVE: Here we prospectively tested the best machine-predicted programming settings in a dystonia cohort treated with GPi-DBS against the settings derived from clinical long-term care in a specialised DBS centre. METHODS: Previously, we reconstructed an anatomical map of motor improvement probability across the pallidal region using individual stimulation volumes and clinical outcomes in dystonia patients. We used this to develop an algorithm that tests in silico thousands of putative stimulation settings in de novo patients after reconstructing an individual, image-based anatomical model of electrode positions, and suggests stimulation parameters with the highest likelihood of optimal symptom control. To test real-life application, our prospective study compared results in 10 patients against programming settings derived from long-term care. RESULTS: In this cohort, dystonia symptom reduction was observed at 74.9 ± 15.3% with C-SURF programming as compared to 66.3 ± 16.3% with clinical programming (p < 0.012). The average total electrical energy delivered (TEED) was similar for both the clinical and C-SURF programming (262.0 µJ/s vs. 306.1 µJ/s respectively). CONCLUSION: Our findings highlight the clinical potential of machine-based programming in dystonia, which could markedly reduce the programming burden in postoperative management.

Distinct phenotypes of stimulation-induced dysarthria represent different cortical networks in STN-DBS.

BACKGROUND: Deep brain stimulation of the subthalamic nucleus is an effective treatment of Parkinson's disease, yet it is often associated with a general deterioration of speech intelligibility. Clustering the phenotypes of dysarthria has been proposed as a strategy to tackle these stimulation-induced speech problems. METHODS: In this study, we examine a cohort of 24 patients to test the real-life application of the proposed clustering and attempt to attribute the clusters to specific brain networks with two different approaches of connectivity analysis. RESULTS: Both our data-driven and hypothesis-driven approaches revealed strong connections of variants of stimulation-induced dysarthria to brain regions that are known actors of motor speech control. We showed a strong connection between the spastic dysarthria type and the precentral gyrus and supplementary motor area, prompting a possible disruption of corticobulbar fibers. The connection between the strained voice dysarthria and more frontal areas hints toward a deeper disruption of the motor programming of speech production. CONCLUSIONS: These results provide insights into the mechanism of stimulation-induced dysarthria in deep brain stimulation of the subthalamic nucleus and may guide reprogramming attempts for individual Parkinson's patients based on pathophysiological understanding of the affected networks.

Machine Learning in Tremor Analysis: Critique and Directions.

Tremor is the most frequent human movement disorder, and its diagnosis is based on clinical assessment. Yet finding the accurate clinical diagnosis is not always straightforward. Fine-tuning of clinical diagnostic criteria over the past few decades, as well as device-based qualitative analysis, has resulted in incremental improvements to diagnostic accuracy. Accelerometric assessments are commonplace, enabling clinicians to capture high-resolution oscillatory properties of tremor, which recently have been the focus of various machine-learning (ML) studies. In this context, the application of ML models to accelerometric recordings provides the potential for less-biased classification and quantification of tremor disorders. However, if implemented incorrectly, ML can result in spurious or nongeneralizable results and misguided conclusions. This work summarizes and highlights recent developments in ML tools for tremor research, with a focus on supervised ML. We aim to highlight the opportunities and limitations of such approaches and provide future directions while simultaneously guiding the reader through the process of applying ML to analyze tremor data. We identify the need for the movement disorder community to take a more proactive role in the application of these novel analytical technologies, which so far have been predominantly pursued by the engineering and data analysis field. Ultimately, big-data approaches offer the possibility to identify generalizable patterns but warrant meaningful translation into clinical practice. © 2023 The Authors. Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Smartphone video nystagmography using convolutional neural networks: ConVNG.

BACKGROUND: Eye movement abnormalities are commonplace in neurological disorders. However, unaided eye movement assessments lack granularity. Although videooculography (VOG) improves diagnostic accuracy, resource intensiveness precludes its broad use. To bridge this care gap, we here validate a framework for smartphone video-based nystagmography capitalizing on recent computer vision advances. METHODS: A convolutional neural network was fine-tuned for pupil tracking using > 550 annotated frames: ConVNG. In a cross-sectional approach, slow-phase velocity of optokinetic nystagmus was calculated in 10 subjects using ConVNG and VOG. Equivalence of accuracy and precision was assessed using the "two one-sample t-test" (TOST) and Bayesian interval-null approaches. ConVNG was systematically compared to OpenFace and MediaPipe as computer vision (CV) benchmarks for gaze estimation. RESULTS: ConVNG tracking accuracy reached 9-15% of an average pupil diameter. In a fully independent clinical video dataset, ConVNG robustly detected pupil keypoints (median prediction confidence 0.85). SPV measurement accuracy was equivalent to VOG (TOST p < 0.017; Bayes factors (BF) > 24). ConVNG, but not MediaPipe, achieved equivalence to VOG in all SPV calculations. Median precision was 0.30°/s for ConVNG, 0.7°/s for MediaPipe and 0.12°/s for VOG. ConVNG precision was significantly higher than MediaPipe in vertical planes, but both algorithms' precision was inferior to VOG. CONCLUSIONS: ConVNG enables offline smartphone video nystagmography with an accuracy comparable to VOG and significantly higher precision than MediaPipe, a benchmark computer vision application for gaze estimation. This serves as a blueprint for highly accessible tools with potential to accelerate progress toward precise and personalized Medicine.

Transcranial ultrasound stimulation to human middle temporal complex improves visual motion detection and modulates electrophysiological responses.

BACKGROUND: Transcranial ultrasound stimulation (TUS) holds promise as a novel technology for non-invasive neuromodulation, with greater spatial precision than other available methods and the ability to target deep brain structures. However, its safety and efficacy for behavioural and electrophysiological modulation remains controversial and it is not yet clear whether it can be used to manipulate the neural mechanisms supporting higher cognitive function in humans. Moreover, concerns have been raised about a potential TUS-induced auditory confound. OBJECTIVES: We aimed to investigate whether TUS can be used to modulate higher-order visual function in humans in an anatomically-specific way whilst controlling for auditory confounds. METHODS: We used participant-specific skull maps, functional localisation of brain targets, acoustic modelling and neuronavigation to guide TUS delivery to human visual motion processing cortex (hMT+) whilst participants performed a visual motion detection task. We compared the effects of hMT+ stimulation with sham and control site stimulation and examined EEG data for modulation of task-specific event-related potentials. An auditory mask was applied which prevented participants from distinguishing between stimulation and sham trials. RESULTS: Compared with sham and control site stimulation, TUS to hMT+ improved accuracy and reduced response times of visual motion detection. TUS also led to modulation of the task-specific event-related EEG potential. The amplitude of this modulation correlated with the performance benefit induced by TUS. No pathological changes were observed comparing structural MRI obtained before and after stimulation. CONCLUSIONS: The results demonstrate for the first time the precision, efficacy and safety of TUS for stimulation of higher-order cortex and cognitive function in humans whilst controlling for auditory confounds.

Prediction of hospital-onset COVID-19 infections using dynamic networks of patient contact: an international retrospective cohort study.

BACKGROUND: Real-time prediction is key to prevention and control of infections associated with health-care settings. Contacts enable spread of many infections, yet most risk prediction frameworks fail to account for their dynamics. We developed, tested, and internationally validated a real-time machine-learning framework, incorporating dynamic patient-contact networks to predict hospital-onset COVID-19 infections (HOCIs) at the individual level. METHODS: We report an international retrospective cohort study of our framework, which extracted patient-contact networks from routine hospital data and combined network-derived variables with clinical and contextual information to predict individual infection risk. We trained and tested the framework on HOCIs using the data from 51 157 hospital inpatients admitted to a UK National Health Service hospital group (Imperial College Healthcare NHS Trust) between April 1, 2020, and April 1, 2021, intersecting the first two COVID-19 surges. We validated the framework using data from a Swiss hospital group (Department of Rehabilitation, Geneva University Hospitals) during a COVID-19 surge (from March 1 to May 31, 2020; 40 057 inpatients) and from the same UK group after COVID-19 surges (from April 2 to Aug 13, 2021; 43 375 inpatients). All inpatients with a bed allocation during the study periods were included in the computation of network-derived and contextual variables. In predicting patient-level HOCI risk, only inpatients spending 3 or more days in hospital during the study period were examined for HOCI acquisition risk. FINDINGS: The framework was highly predictive across test data with all variable types (area under the curve [AUC]-receiver operating characteristic curve [ROC] 0·89 [95% CI 0·88-0·90]) and similarly predictive using only contact-network variables (0·88 [0·86-0·90]). Prediction was reduced when using only hospital contextual (AUC-ROC 0·82 [95% CI 0·80-0·84]) or patient clinical (0·64 [0·62-0·66]) variables. A model with only three variables (ie, network closeness, direct contacts with infectious patients [network derived], and hospital COVID-19 prevalence [hospital contextual]) achieved AUC-ROC 0·85 (95% CI 0·82-0·88). Incorporating contact-network variables improved performance across both validation datasets (AUC-ROC in the Geneva dataset increased from 0·84 [95% CI 0·82-0·86] to 0·88 [0·86-0·90]; AUC-ROC in the UK post-surge dataset increased from 0·49 [0·46-0·52] to 0·68 [0·64-0·70]). INTERPRETATION: Dynamic contact networks are robust predictors of individual patient risk of HOCIs. Their integration in clinical care could enhance individualised infection prevention and early diagnosis of COVID-19 and other nosocomial infections. FUNDING: Medical Research Foundation, WHO, Engineering and Physical Sciences Research Council, National Institute for Health Research (NIHR), Swiss National Science Foundation, and German Research Foundation.

Relative, local and global dimension in complex networks.

Dimension is a fundamental property of objects and the space in which they are embedded. Yet ideal notions of dimension, as in Euclidean spaces, do not always translate to physical spaces, which can be constrained by boundaries and distorted by inhomogeneities, or to intrinsically discrete systems such as networks. To take into account locality, finiteness and discreteness, dynamical processes can be used to probe the space geometry and define its dimension. Here we show that each point in space can be assigned a relative dimension with respect to the source of a diffusive process, a concept that provides a scale-dependent definition for local and global dimension also applicable to networks. To showcase its application to physical systems, we demonstrate that the local dimension of structural protein graphs correlates with structural flexibility, and the relative dimension with respect to the active site uncovers regions involved in allosteric communication. In simple models of epidemics on networks, the relative dimension is predictive of the spreading capability of nodes, and identifies scales at which the graph structure is predictive of infectivity. We further apply our dimension measures to neuronal networks, economic trade, social networks, ocean flows, and to the comparison of random graphs.

Listening to Mental Health Crisis Needs at Scale: Using Natural Language Processing to Understand and Evaluate a Mental Health Crisis Text Messaging Service.

The current mental health crisis is a growing public health issue requiring a large-scale response that cannot be met with traditional services alone. Digital support tools are proliferating, yet most are not systematically evaluated, and we know little about their users and their needs. Shout is a free mental health text messaging service run by the charity Mental Health Innovations, which provides support for individuals in the UK experiencing mental or emotional distress and seeking help. Here we study a large data set of anonymised text message conversations and post-conversation surveys compiled through Shout. This data provides an opportunity to hear at scale from those experiencing distress; to better understand mental health needs for people not using traditional mental health services; and to evaluate the impact of a novel form of crisis support. We use natural language processing (NLP) to assess the adherence of volunteers to conversation techniques and formats, and to gain insight into demographic user groups and their behavioural expressions of distress. Our textual analyses achieve accurate classification of conversation stages (weighted accuracy = 88%), behaviours (1-hamming loss = 95%) and texter demographics (weighted accuracy = 96%), exemplifying how the application of NLP to frontline mental health data sets can aid with post-hoc analysis and evaluation of quality of service provision in digital mental health services.

Informing antimicrobial management in the context of COVID-19: understanding the longitudinal dynamics of C-reactive protein and procalcitonin.

BACKGROUND: To characterise the longitudinal dynamics of C-reactive protein (CRP) and Procalcitonin (PCT) in a cohort of hospitalised patients with COVID-19 and support antimicrobial decision-making. METHODS: Longitudinal CRP and PCT concentrations and trajectories of 237 hospitalised patients with COVID-19 were modelled. The dataset comprised of 2,021 data points for CRP and 284 points for PCT. Pairwise comparisons were performed between: (i) those with or without significant bacterial growth from cultures, and (ii) those who survived or died in hospital. RESULTS: CRP concentrations were higher over time in COVID-19 patients with positive microbiology (day 9: 236 vs 123 mg/L, p < 0.0001) and in those who died (day 8: 226 vs 152 mg/L, p < 0.0001) but only after day 7 of COVID-related symptom onset. Failure for CRP to reduce in the first week of hospital admission was associated with significantly higher odds of death. PCT concentrations were higher in patients with COVID-19 and positive microbiology or in those who died, although these differences were not statistically significant. CONCLUSIONS: Both the absolute CRP concentration and the trajectory during the first week of hospital admission are important factors predicting microbiology culture positivity and outcome in patients hospitalised with COVID-19. Further work is needed to describe the role of PCT for co-infection. Understanding relationships of these biomarkers can support development of risk models and inform optimal antimicrobial strategies.

Repurposed floxacins targeting RSK4 prevent chemoresistance and metastasis in lung and bladder cancer.

Lung and bladder cancers are mostly incurable because of the early development of drug resistance and metastatic dissemination. Hence, improved therapies that tackle these two processes are urgently needed to improve clinical outcome. We have identified RSK4 as a promoter of drug resistance and metastasis in lung and bladder cancer cells. Silencing this kinase, through either RNA interference or CRISPR, sensitized tumor cells to chemotherapy and hindered metastasis in vitro and in vivo in a tail vein injection model. Drug screening revealed several floxacin antibiotics as potent RSK4 activation inhibitors, and trovafloxacin reproduced all effects of RSK4 silencing in vitro and in/ex vivo using lung cancer xenograft and genetically engineered mouse models and bladder tumor explants. Through x-ray structure determination and Markov transient and Deuterium exchange analyses, we identified the allosteric binding site and revealed how this compound blocks RSK4 kinase activation through binding to an allosteric site and mimicking a kinase autoinhibitory mechanism involving the RSK4's hydrophobic motif. Last, we show that patients undergoing chemotherapy and adhering to prophylactic levofloxacin in the large placebo-controlled randomized phase 3 SIGNIFICANT trial had significantly increased (P = 0.048) long-term overall survival times. Hence, we suggest that RSK4 inhibition may represent an effective therapeutic strategy for treating lung and bladder cancer.

HCGA: Highly comparative graph analysis for network phenotyping.

Networks are widely used as mathematical models of complex systems across many scientific disciplines. Decades of work have produced a vast corpus of research characterizing the topological, combinatorial, statistical, and spectral properties of graphs. Each graph property can be thought of as a feature that captures important (and sometimes overlapping) characteristics of a network. In this paper, we introduce HCGA, a framework for highly comparative analysis of graph datasets that computes several thousands of graph features from any given network. HCGA also offers a suite of statistical learning and data analysis tools for automated identification and selection of important and interpretable features underpinning the characterization of graph datasets. We show that HCGA outperforms other methodologies on supervised classification tasks on benchmark datasets while retaining the interpretability of network features. We exemplify HCGA by predicting the charge transfer in organic semiconductors and clustering a dataset of neuronal morphology images.