IEEE Access

Scientific image sequences often feature recurring structures with unknown appearance and dynamics. While traditional methods using predefined templates, noise models, or motion classes work well when targets are known, such assumptions can hinder exploratory analysis where discovering that knowledge is the primary goal. This work presents a self-supervised framework that maps noisy video into a discrete vocabulary of local states and their spatiotemporal relationships. The method operates without clean targets, pretrained representations, semantic labels, feature templates, or prescribed trajectories, assuming only that informative structures recur for a finite duration within a bounded spatial neighborhood. A temporally masked vector-quantized network predicts the missing central frame from its neighbors, while a masked Transformer refines token assignments by weighing encoder evidence against contextual compatibility. Project-specific token sets define the feature family, while a support model quantifies the temporal evidence for individual occurrences, facilitating optional conservative suppression. The framework is first demonstrated on synthetic videos of moving particles. Without access to clean frames or particle coordinates during learning and selection, the learned states recover compact, point-spread-function-like structures. Tests on low-signal particle-tracking benchmarks yield frame-wise component precision between 87.0% and 94.5%, with recall decreasing as particle density rises. An experimental example using E. coli further shows that discrete states can separate biological structures from illumination and acquisition artifacts. In all cases, the primary outputs are interpretable state, activity, support, and component maps; rendered images serve as diagnostics rather than optimization targets.

Parkinson disease (PD) affects speech production, motivating classification from repeated voice measurements. Record-wise evaluation can place recordings from the same participant in both training and test sets. We propose VoxNeuro, a novel subject-level framework combining Grassmann representations of repeated recordings with Euclidean mean-dispersion summaries through a provably positive-semidefinite kernel for support vector classification. Features undergo training-fold Gaussian rank normalization, and a supervised partial least squares subspace ensemble is activated when the feature count exceeds the number of training recordings. Two public cohorts were evaluated, UCI-489 (80 subjects) and PD-252 (252 subjects), each with three sustained-vowel recordings per subject. At the reference five-fold subject-disjoint partition, balanced accuracy was 87.5% and 81.6%, respectively, exceeding all five standard comparators under identical folds and normalization. PD-252 sensitivity was 89.9% and specificity was 73.4%. Across 20 additional subject partitions, PD-252 balanced accuracy averaged 78.4%, with a mean paired advantage of 3.2 percentage points over the highest-scoring standard comparator. VoxNeuro exceeded each standard comparator in at least 19 partitions for both balanced accuracy and macro-F1. Ablations favored rank normalization and the unaugmented supervised-subspace configuration on PD-252. VoxNeuro thus offers an effective and reproducible framework for subject-level analysis of repeated speech measurements in PD screening research.

Multiview data analysis has become increasingly important for integrating heterogeneous information from multiple data sources. Cooperative learning provides a flexible framework for multiview integration by bridging early and late fusion strategies. However, its use of squared loss makes it sensitive to outliers. In this study, we propose robust cooperative learning based on the soft-IPOD framework. The proposed method introduces an observation-specific outlier absorption vector into the cooperative learning objective, thereby reducing the influence of outlying observations while preserving information sharing across views. The resulting optimization problem can be formulated as a weighted Lasso-type problem using an augmented design matrix and can be naturally extended to more than two data views. Numerical simulations and a real-data application to labor onset prediction demonstrate the predictive performance of the proposed method in the presence of outliers.

Although largely preventable, cardiovascular disease (CVD) is the leading cause of death worldwide. Synthetic data generation (SDG) of health data can facilitate data sharing for research and development of personalised cardiovascular disease prevention tools. Current state-of-the-art SDG methods are limited to cross-sectional data or focused on short-term or in-patient EHR applications lacking benchmarks for generalisability. Since prevention is a long-term longitudinal objective, these limitations must be addressed. We evaluated two tabular generative adversarial networks, TGAN and CTGAN, for privacy and fidelity on a structured 6-year longitudinal trial dataset within CVD prevention. We compared performance with a rudimentary perturbation method. In addition, we performed a variable-wise evaluation and an age-stratified analysis of the higher performing model to assess preservation of clinically relevant characteristics. Overall, CTGAN showed better fidelity than TGAN. The models performed similarly within privacy, showing better results than the perturbation method. CTGAN had lower performance in comparison to the perturbation method within numerical fidelity of continuous variables and overall temporal fidelity. Furthermore, CTGAN's performance was lower for younger age groups than older. CTGAN shows better than or equal performance within fidelity and privacy than TGAN and the perturbation method on a long-term longitudinal dataset within CVD prevention. To the best of our knowledge, this is the first study evaluating CTGAN on a longitudinal dataset and our findings suggest that handling of temporal characteristics and imbalanced subgroups need further refinement.

Discourse analysis can reliably predict real-world communicative success, yet is rarely implemented clinically, due to challenges in manually generating stimulus-specific main concept inventories (MCIs) and scoring patient narratives. We evaluated whether automatically derived macrolinguistic (main concept, sentiment) and microlinguistic (fluency, lexical, syntactic) features extracted from movie clip narrations can augment the clinical assessment of discourse in aphasia. The MCIs generated directly from each clip's video, audio and subtitles using a vision-language model approximated the main concepts healthy controls produced. Group classification (control: n=51; aphasia: n=54) based on the discourse features was excellent (AUC=0.95, 95% CI [0.90, 0.99]), with macrolinguistic features showing the largest group differences (main concept completeness: d=-1.84, 95% CI [-2.30, -1.38]) and strongest associations with aphasia severity (semantic distance to inventory: r=-0.79, 95% CI [-0.87, -0.66]). Findings support the feasibility and scalability of automated criterion-based discourse assessment using ecological stimuli, and highlight its potential for clinical implementation.

Tabular synthetic data generation (SDG) can facilitate research and development in the health domain, as access to health data is restricted. To accommodate for its structural complexity, such as multimodal distributions and fragmentation, transformer models provides an alternative to conventional SDG methods since they are better equipped to infer global spatial coherence due to textual context of tabular records. However, methods for optimising their adaption to tabular SDG are still largely unexplored. We contribute to the understanding of transformer-based SDG by evaluating fidelity and privacy of three prompting regimes, zero shot, one shot and few shot training, for tabular health SDG using small transformer models. We evaluated llama 3.2 (1B and 3B parameters) and Phi (1.3B and 2.7B parameters), on a real clinical dataset. Our results showed that zero shot learning was less prone to hallucinations and performed overall better for fidelity, but at the loss of privacy in comparison to the other regimes. Our results showed no differences between model sizes and families. Future studies should extend the analysis to other models and datasets, and explore strategies to reduce hallucinations.

Despite the use of technological countermeasures, studies show that the sensory conflict, and thus symptoms of motion sickness such as nausea, cannot be completely prevented while traveling as a car passenger. Accordingly, reliable prediction of motion sickness will represent a possible solution for intelligent vehicles to avoid the abortion of the drive, e.g., through adapted driving strategies or route planning. In previous works, the authors were able to demonstrate an application-oriented parameter set for the popular 6DOF-SVC model to estimate motion sickness incidence (MSI) in passengers with a lowered gaze on a non-driving-related task. In the current study, these parameters were validated and further optimized using five previous conducted motion sickness driving experiments. For the first time, it is also shown that the expected natural MSI fluctuations based on the driving dynamic profile, due to route characteristics or driving style, as well as the variation in experimental determined MSI is approximately 10%. Moreover, findings indicate that the participants appear to have an individual MSI threshold that correlates with their susceptibility to motion sickness, however, this threshold fluctuates too strongly across experiments to be suitable for practical application. As a result, the model can successfully estimate motion sickness risk across the test tracks, however, inaccuracies in the predicted MSI trajectories remain.

Background: System One Model is designed for structured decision-making and can provide probabilistic outputs, but their performance in domain-specific physical therapy tasks has not been established. Objective: To evaluate the performance and potential utility of Jev, a System One Model, for professional knowledge-based selection tasks in physical therapy by comparing it with GPT-4o. Methods: A total of 380 multiple-choice questions from the 2024 and 2025 Korean Physical Therapist Licensing Examinations were independently submitted to Jev and GPT-4o. The outcomes included answer accuracy, subgroup performance, output validity, API response latency, token usage, estimated API cost, and Jev's prediction uncertainty. A retrospective probability-based model-cascading simulation was also performed. Results: Jev correctly answered 282 questions (74.2%), whereas GPT-4o answered 326 (85.8%), a difference of -11.6 percentage points. Jev produced valid structured responses for all questions, whereas GPT-4o produced three invalid responses. Jev's selected-answer probabilities discriminated correct from incorrect answers (AUROC = 0.866), and accuracy reached 99.3% when probabilities were greater than or equal to 0.90. The median API response latency was 0.94 s for Jev and 1.06 s for GPT-4o, with estimated costs of US$0.008 and US$0.161, respectively. At a retrospective cascading threshold of 0.70, accuracy was 86.6% with 175 simulated GPT-4o calls. Conclusion: Jev showed lower overall accuracy than GPT-4o but provided reliable structured outputs, informative prediction probabilities, lower latency, and lower estimated cost. Its probabilistic outputs warrant further prospective evaluation for selective model escalation and decision-support workflows in physical therapy.

The -Bonacci weighted sequence space {ell}(,) is introduced for [&ge;] 2, 1 [&le;] < {infty}, and > 0). It is proved to be a Banach space which is isometrically isomorphic to {ell}, with an unconditional Schauder basis. The space is also reflexive and uniformly convex. Its strict inclusion chain, continuous dual, and asymptotic truncation estimate of order - / are also established. Based on this framework, Scaled &#119898-BonacciNet is proposed as a lightweight CNN that rescales Fibonacci-Net channels using the dominant &#119898-Bonacci root and employs depth wise separable convolutions. Tribonacci-Net achieved 96.09% accuracy and a 0.9581 macro F1-score on the 15-class Brain MRI dataset using 0.85 million parameters and 0.366 GFLOPs. Pentanacci-Net attained 95.87% accuracy on the 44-class dataset while reducing parameters and FLOPs by 47.10% and 44.48%, respectively. On the CKD ultrasound dataset, Hexanacci-Net achieved 83.75% accuracy and a 0.7941 macro F1-score, with corresponding reductions of 50.37% and 45.10%. These findings support dominant-root scaling as a structured approach to lightweight CNN design.

Marker-based optical motion capture remains the gold standard for kinematic analysis across biomechanical research. However, when tracking dynamic movements, severe occlusions often create missing data and noisy signals that require time-consuming manual cleaning. Existing automated methods usually rely on predefined skeleton models or need massive training datasets for machine learning solutions. This study presents the AMP: an automated marker based motion capture processing pipeline. The AMP is a fully analytical approach that uses hierarchical algorithms including 1) rigid-body constraint, 2) Kalman Filter state estimation and 3) kinematic pattern match to label markers and fill data gaps without manual human intervention. First, the algorithm identifies and removes mislabeling markers before any data scrubbing. During the data cleaning process, our hierarchical algorithm uses a translation-based step to track segments even when there are insufficient donor markers to form a full rigid body. To reconstruct missing marker trajectories, the pipeline initially uses Piecewise Cubic Hermite Interpolating Polynomial (PCHIP) interpolation for short marker gaps. However, the PCHIP creates massive spatial errors during long occlusions, the system automatically switches to rigid-body tracking for gaps over 40 ms. Results showed the pipeline drastically reduced processing time compared to manual human expert methods by 73.5%. In addition, the AMP kept spatial errors within 10 mm of the manual ground truth. By using this combined hierarchical approach to match marker processing accuracy, the pipeline provides a fast, general-purpose tool that fully automates data cleaning and saves hours of manual effort.

Medical ultrasound is a portable, non-invasive, and cost effective imaging modality that is particularly well suited for resource-limited settings. Optic nerve sheath diameter (ONSD) measurement from ultrasound is used as a point-of-care assessment tool for intracranial pressure (ICP), which is associated with several neurological conditions. However, valid measurement depends on selecting a frame in which the optic nerve sheath is clearly visible. Manual selection requires a high level of expertise. To enable the use of ONSD in low resource settings by medical personnel of all levels, we present a sparsely supervised AI framework that scores each frame of an ONSD ultrasound video and selects the optimal frame for ONSD assessment. Per-frame embeddings from an ultrasound foundation model (USFM) are passed to a lightweight temporal head and trained using Gaussian soft labels, which assign graded targets around labeled key frames, rather than hard binary (0/1) per-frame targets. We evaluate the model performance using subject-level cross-validation on 18 subjects and 323 ultrasound videos spanning nine acquisition sweep types. Top-k frame selection is used as a metric for direct comparison against the baselines. While only requiring sparse labels, our method can identify an optimal frame in 82.2% of the evaluated sweeps, exceeding the strongest training-free baseline (49.2%) and hard-label classification (71.9%).

Automated surveillance systems can prove to be extremely useful in detecting outbreaks of respiratory diseases that need further epidemiological investigation. In order for the system to be effective, careful attention has to be paid to the baseline used, absence of the observations, and the use of proper statistical methods. As a solution, the paper describes EpiAlert, a weekly surveillance system based on CDC FluView ILINet. The system is able to transform the number of ILI visits and the total number of visits into a ILI prevalence for the region, create the first reference from the three previous years using a +/- 2-week interval, transform the rate value, and apply two different detectors - CUSUM and EWMA. CUSUM is set at k=0.5 and h=5, whereas EWMA uses {lambda}=0.2 and H=2.7 values. Among the 55 districts, the system gives 45,138 pieces of information about the disease by transforming 44,297 pieces of data into alerts. It is found that there are 2,820 alerts in total (out of which 2,711 alerts used CUSUM, 402 used EWMA, and 293 used both detectors). Additionally, the experiments show that the sensitivity of the device depends greatly on the detector. The latest statistics can be tracked in the report for week 37 of 2026 that recently concluded in West.

Synthetic medical data are commonly generated from manually chosen prompts, inverse frequency sampling, or fixed subgroup balancing rules. These strategies increase sample count but do not answer a more consequential question: which missing observations would most improve the current model? We introduce GAP DREAM, a closed-loop framework that treats a medical image generator as a queryable source of candidate evidence rather than an unlimited source of augmentation. At each round, GAP DREAM maps model failures into an evidence gap score that combines scarcity, predictive uncertainty, local representation density, and gradient novelty. A query policy translates high-value gaps into structured clinical requests, a modality specific generator proposes candidate imagetext pairs, and an independent selector retains only diagnostically consistent, diverse, non-duplicative candidates. The downstream model is then updated and the gap map recomputed. The central objective is therefore not image realism in isolation, but reduction of measurable evidence gaps under a fixed synthetic budget. Across a multi-modality protocol spanning long-tailed chest radiography, dermoscopy, clinical dermatology, and histopathology, GAPDREAM reaches stronger held-out real-data performance with substantially fewer generated samples than random,class-balanced, uncertainty-only, and static targeted augmentation. Results show gains in CXRLT tail mAP, HAM10000 macroAUROC, Came lyon17 out-of-domain accuracy, and worst group dermatology accuracy while reducing redundant synthetic samples. GAPDREAM reframes medical generation as active evidence acquisition: generate the examples the learner needs, not simply more examples of what is rare.

Background and purpose: Accurate identification of cardiac MRI volume orientations is essential for reliable image interpretation and for enabling downstream automated analysis pipelines. However, orientation labels and commonly assigned manually, making the process time-consuming and prone to variability. Recent advances in vision-language models offers new opportunities for automated label generation by jointly modeling visual and semantic information. Methods: In this study, we systematically compare a conventional end-to-end transformer-based captioning model with a CLIPCap architecture for automated cardiac MRI view labelling. Both models were evaluated on a dataset of 1184 de-identified cardiac MR images spanning across six clinically relevant orientations: short axis, two-, three-, four-chamber views, left and right ventricular outflow tract views. The caption transformer was trained from scratch, whereas CLIPCap leveraged pretrained CLIP image embeddings combined with lightweight mapping network and a frozen language model. Results: Experimental results demonstrate a substantial performance advantage from CLIPCap, which achieved an overall accuracy of 98% compared to 68% for the caption transformer. CLIPCap consistently delivered high precision and recall across all orientation classes, including those of high clinical importance, while the end to end transformer exhibited unstable performance and complete failure in certain views. Conclusion: These findings highlight the benefits of leveraging large scale multimodal pretrained model for medical image labeling tasks, particularly in data limited settings. The study suggests that CLIPCap provides a more robust and clinically reliable solution for automated cardiac MRI orientation labeling, supporting its integration into efficient and scalable clinical workflows.

Background: Lean Healthcare has been adopted as an approach to reducing waste, enhancing process efficiency, and optimizing clinical and administrative workflows. However, empirical evaluations aimed at establishing a causal relationship between Lean Healthcare interventions and outcomes remain limited. This evidence gap is particularly pronounced in the context of large-scale implementations in low- and middle-income countries, where more efficient allocation could yield substantial improvements in population health. Methods: We conducted an impact evaluation study employing a staggered difference-in-differences approach to estimate the effects of the Lean Healthcare in Emergency Departments (ED) initiative in the Brazilian context. The intervention was implemented between January 2017 and December 2025 across 309 hospitals. We analyzed hospital-level indicators of the number of hospitalizations and the average length of stay (in days) among patients admitted through urgent or emergency pathways. Results: Our estimates suggest that the intervention was associated with an average increase of 38.6 admissions per hospital per month (p = 0.033), corresponding to approximately a 5.9% increase relative to the monthly admission volume. We also observed a reduction of 0.184 days in the average length of stay (p = 0.058), equivalent to an approximately 3.0% decrease in hospitalization time. These effects exhibited distinct temporal dynamics: the effect on the number of hospitalizations emerged gradually and increased with the duration of the intervention, whereas the reduction in length of stay became evident earlier. Conclusion: This study provides evidence of a causal effect of Lean Healthcare implementation in emergency departments on operational performance. The simultaneous increase in admissions and reduction in length of stay suggests improved bed productivity and effective capacity without expanding physical infrastructure. These findings support the use of flow-management methodologies as an effective strategy for mitigating emergency department overcrowding while increasing service capacity.

Aim. To evaluate the accuracy, run-to-run consistency, and agreement of large language models on the Ukrainian Integrated Licensing Examination Krok 2 for the Specialty of Physical Therapy. Materials and Methods. Five cloud-based LLMs (ChatGPT-5.5, Claude 5 Sonnet, Gemini 3.5 Flash, Grok 4, DeepSeek-V3) were evaluated using 150 Ukrainian-language multiple-choice questions from the official Krok 2 database. Each model completed two independent runs with identical inputs. Accuracy (with 95% confidence intervals, Wilson method), run-to-run consistency, and agreement were assessed. Statistical analysis included Cohen's {kappa}, Cochran's Q and pairwise McNemar tests with Holm-Bonferroni correction. Results. All models achieved >80% accuracy in both runs, exceeding the passing threshold of [&ge;]64%. In the first run, accuracy ranged from 80.00% (DeepSeek-V3) to 93.33% (Gemini 3.5 Flash); in the second run, from 82.67% (Claude 5 Sonnet) to 88.67% (Gemini 3.5 Flash). In the first run significant differences were observed between Gemini 3.5 Flash and Grok 4 (93.33% vs 84.00%; p=0.0022) and DeepSeek-V3 (93.33% vs 80.00%; p=0.0001). There were no statistically significant differences between models in the second run. R2R consistency exceeded 90% for all models with the highest in ChatGPT-5.5 (98.67%) and lowest in DeepSeek-V3 (90.00%). Cohen's {kappa} ranged from 0.64 to 0.94, indicating substantial to almost perfect agreement. The stability of LLMs' response correctness across two runs ranged from 77.33% (DeepSeek-V3) to 88.00% (Gemini 3.5 Flash). Conclusions. All LLMs demonstrated high performance on Ukrainian-language Krok 2 examination tasks. Gemini 3.5 Flash showed the highest accuracy in both runs and response correctness stability while ChatGPT-5.5 exhibited the greatest response consistency and agreement. Although the accuracy and stability of LLM responses may be interrelated, they represent distinct aspects of model performance. This highlights the need for a comprehensive approach to evaluating LLMs. The findings indicate the potential of LLMs as an auxiliary tool in the professional practice of physical therapy specialists.

Biofilms are widely distributed in both natural and engineered systems and play a fundamental role in microbial ecology and biotechnological applications. When grown on agar substrates, biofilms often exhibit macroscopic structural features due to mechanical instabilities driven by matrix production. These wrinkles encode relevant information about biofilm growth and structural development and have even been shown to possess some functional roles. Despite their relevance, currently available approaches for wrinkle detection from images rely on manual annotation, which is time-consuming, not scalable, and strongly dependent on the annotator. In the present study, we developed an end-to-end machine learning pipeline, named AutoWrinkleID, that starts from bright field images of biofilms and performs wrinkle identification, characterisation, and quantification across multiple bacterial species and strains. The name reflects the automated identification and subsequent characterisation and quantification of biofilm wrinkles. The annotations used to generate the masks, namely images containing only the wrinkle structures, were obtained both through manual labelling and using constitutive fluorescence and motility fluorescence. Interestingly, we found that models trained using only manual annotations perform worse than those trained using motility fluorescence markers. A comparison between motility and constitutive fluorescence masks further indicates that motility based annotations consistently outperform constitutive fluorescence across all strains, strongly suggesting that motile phenotypes are tightly associated with wrinkle structures. Additional computational analysis is conducted on the same dataset to assess the robustness and limitations of the results as a function of training dataset. We also observed that Sholl analysis, originally developed for the study of neuronal dendrites and later applied to wrinkle analysis, can effectively characterize wrinkle patterns, although its applicability is strongly limited by wrinkle shape and by the quality of the masks.

Age and sex estimation from dental radiographs supports forensic identification and age-dependent legal assessment. Multi-input models may combine global and regional representations, but when these inputs are derived from the same radiograph, they may repeat anatomical information rather than provide complementary signals. We developed HMA-Net, a hierarchical multi-stream anatomical network, together with a task-specific input audit that evaluates whether each candidate representation improves at least one prediction task without materially degrading the other. The model was developed using a retrospective single-centre cohort of 4,293 panoramic radiographs from individuals aged 3-18 years and combined a panoramic-image stream with graph-linked tooth representations. On the held-out test set, the final panoramic-plus-permanent-tooth model achieved a mean absolute age error of 0.51 years and a sex-classification accuracy of 85.2%; 99.4% of age estimates were within 2 years of chronological age. Across three independent training runs, removing the jaw-region stream reduced mean absolute age error from 0.68 +/- 0.01 to 0.52 +/- 0.01 years and increased sex accuracy from 81.3+/- 1.7% to 85.8 +/- 0.8%. Adding 20 deciduous-tooth classes provided no age benefit and reduced sex accuracy by 7.9% relative to the permanent-only representation. The mandibular first molars received the greatest tooth-level attention, and their occlusion impaired both prediction tasks. These findings show that anatomically plausible inputs do not necessarily provide incremental predictive value. Task-specific input auditing can improve joint model performance, reduce input redundancy and preserve anatomically traceable predictions. More broadly, this framework provides a testable approach for selecting correlated inputs in medical-imaging systems in which regional views re-present anatomy already contained in the whole image. Keywords: Forensic odontology; Dental age estimation; Sex classification; Panoramic radiography; Multi-input learning; Anatomical input auditing

Background: Accurate patient understanding of lung cancer screening (LCS) results is critical for engagement and follow-up adherence, given the currently low screening uptake. While artificial intelligence (AI) systems show promise in facilitating patient communication, its effective use in clinical settings requires appropriate invocation of imaging tools and self-regulation by deferring certain questions to physicians. To this end, we developed and evaluated a planner-executor style multimodal agentic system to answer simulated patient questions about LCS CT results. Methods: This retrospective study utilized 116 LCS CT reports and images collected from a tertiary academic hospital. The system employed a multi-agent architecture (planner and executor) to mimic clinical reasoning. We developed a large language model (LLM)-based question generation framework to prepare a comprehensive set of 699 simulated patient questions, categorized as answerable and defer-to-doctor to assess self-regulation. Performance was evaluated on tool-calling, self-regulation accuracy (ability to correctly defer out-of-scope questions to a human provider) assessed by LLM-as-a-judge, and clinical quality assessed by a reader performance study with four physicians evaluating a subset of 100 responses. Results: The system demonstrated high overall tool-calling accuracy of 93.1% (646/694; 95% CI: 91.2%, 95.0%) and self-regulation accuracy of 92.6% (462/499; 90.2%, 94.8%). For answerable questions, the percentage of responses receiving perfect 5.0 scores across all readers included 78.9% (180/228) for clinical accuracy (inter-rater agreement: Gwet's AC2=0.92), and 73.2% (167/228) for patient understandability (0.86). For defer-to-doctor questions, the percentage of responses included 77.3% (133/172) for clinical accuracy (0.80) and 70.3% (121/172) for patient understandability (0.78). Conclusions: The planner-executor style multi-modal agentic system reliably answers patient-specific questions about LCS. Its high tool-calling and self-regulation performance demonstrates its potential as a safe and effective digital communication facilitator in LCS programs.

Measurements of human movement are critical for evaluating patients with neuromuscular diseases. However, current clinical tools are limited in sensitivity and can fail to detect the full spectrum of movement changes patients may exhibit. These limitations can obscure early signs of decline, delaying interventions that could preserve function and independence. OpenCap, a smartphone application that performs video-based markerless motion capture, can help address this gap by enabling scalable, out-of-lab measurements of human movement. Here, we present a method for both detecting disease and quantifying the severity of impaired movement from smartphone videos of patient motion. We used OpenCap to measure kinematics from 337 individuals with diagnoses spanning 12 rare, genetically-defined neuromuscular diseases and 78 age-matched healthy controls. Our dataset is the largest existing publicly available resource of body segment kinematics for this class of diseases. These data were collected in neuromuscular clinics during regular clinic visits as well as at conferences and community events across the United States. We trained deep learning models on these kinematic time series to differentiate between participants with neuromuscular disease and controls (AUROC = 0.97 [0.93-1.00]). We also developed a movement impairment severity score that correlates with both participant ratings of their own functional abilities ({rho} = -0.81, p < 0.001) and existing clinical tests of movement capabilities such as run speed ({rho} = -0.83, p < 0.001). Our methods for detecting disease and deriving quantitative biomarkers of movement impairment from smartphone videos have the potential to complement clinical assessments, inform screening protocols, and improve clinical trial endpoints.