Artificial Intelligence

Background: Artificial Intelligence (AI) is increasingly discussed as a tool that can support speech and language therapy (SLT). However, clinical adoption of AI requires improved AI literacy among clinicians. AI is a rapidly evolving and often inconsistently defined field that can be difficult to navigate. Despite the definition provided by the EU AI Act, AI terminology can feel abstract for non-technical readers. Aims: To provide a foundational understanding of AI tailored for SLTs, by translating complex concepts into accessible language and organising them across three levels: (i) AI techniques (how AI works); (ii) AI capabilities (what AI can do) and (iii) clinical applications (how AI can support SLT). Methods: This tutorial is informed by foundational AI literature, established AI taxonomies, relevant SLT literature and regulatory and ethical guidelines. Clinical analogies are used to explain technical concepts, with additional technical detail signposted where relevant. Existing and conceptual examples illustrate the relevance of AI across paediatric SLT practice. Main contribution: This tutorial provides: (i) a clinician-focussed interpretation of the EU AI Act definition; (ii) an organisation of key AI concepts into techniques, capabilities and clinical applications; (iii) a production-line model for mapping clinical needs to AI design choices and (iv) a practice-focussed discussion of ethical and regulatory considerations. Conclusion: AI is best understood as a set of techniques that enable specific capabilities, which in turn support clinical applications. This tutorial promotes the safe, ethical and accountable use of AI as a tool that can support rather than replace clinicians.

The article discusses developing SleepFM, an AI foundation model that learns representations from raw polysomnography data and predicts future disease risk, highlighting multimodal sleep data analysis and its potential clinical translation.

Deep learning models that predict functional genomic measurements from DNA sequences are powerful tools for deciphering the genetic regulatory code. Existing methods involve a trade-off between input sequence length and prediction resolution, thereby limiting their modality scope and performance1-5. We present AlphaGenome, a unified DNA sequence model, which takes as input 1 Mb of DNA sequence and predicts thousands of functional genomic tracks up to single‑base‑pair resolution across diverse modalities. The modalities include gene expression, transcription initiation, chromatin accessibility, histone modifications, transcription factor binding, chromatin contact maps, splice site usage and splice junction coordinates and strength. Trained on human and mouse genomes, AlphaGenome matches or exceeds the strongest available external models in 25 of 26 evaluations of variant effect prediction. The ability of AlphaGenome to simultaneously score variant effects across all modalities accurately recapitulates the mechanisms of clinically relevant variants near the TAL1 oncogene6. To facilitate broader use, we provide tools for making genome track and variant effect predictions from sequence.

Patient-facing large language models (LLMs) hold potential to streamline inefficient transitions from primary to specialist care. We developed the preassessment (PreA), an LLM chatbot co-designed with local stakeholders, to perform the general medical consultations for history‑taking, preliminary diagnoses, and test ordering that would normally be performed by primary care providers and to generate referral reports for specialists. PreA was tested in a randomized controlled trial involving 111 specialists from 24 medical disciplines across two health centers, where 2,069 patients (1,141 women; 928 men) were randomly assigned to use PreA independently (PreA-only), use it with staff support (PreA-human), or not use it (No-PreA) before specialist consultation. The trial met its primary end points with the PreA-only group showing significantly reduced physician consultation duration (28.7% reduction; 3.14 ± 2.25 min) compared to the No-PreA group (4.41 ± 2.77 min; *P* < 0.001), alongside significant improvements in physician‑perceived care coordination (mean scores 113.1% increase; 3.69 ± 0.90 versus 1.73 ± 0.95; *P* < 0.001) and patient‑reported communication ease (mean scores 16.0% increase; 3.99 ± 0.62 versus 3.44 ± 0.97; *P* < 0.001). Equivalent outcomes between the PreA-only and PreA-human groups confirmed the autonomous operation capability. Co-designed PreA outperformed the same model with additional fine-tuning on local dialogues across clinical decision-making domains. Co-design with local stakeholders, compared to passive local data collecting, represents a more effective strategy for deploying LLMs to strengthen health systems and enhance patient‑centered care in resource-limited settings. Chinese Clinical Trial Registry identifier: ChiCTR2400094159.

Objective: To compare the performance of eight large language models (LLMs) with neurology residents on board-style multiple-choice questions across seven subspecialties and two cognitive levels. Methods: In a cross-sectional benchmarking study, we evaluated Bard, Claude, Gemini v1, Gemini 2.5, ChatGPT-3.5, ChatGPT-4, ChatGPT-4o, and ChatGPT-5 using 107 text-only items spanning movement disorders, vascular neurology, neuroanatomy, neuroimmunology, epilepsy, neuromuscular disease, and neuro‑infectious disease. Items were labeled as lower- or higher-order per Bloom’s taxonomy by two neurologists. Models answered each item in a fresh session and reported confidence and Bloom classification. Residents completed the same set under exam-like conditions. Outcomes included overall and domain accuracies, guessing-adjusted accuracy, confidence–accuracy calibration (Spearman ρ), agreement with expert Bloom labels (Cohen κ), and inter‑generation scaling (linear regression of topic-level accuracies). Group differences used Fisher exact or χ² tests with Bonferroni correction. Results: Residents scored 64.9%. ChatGPT‑5 achieved 84.1% and ChatGPT‑4o 81.3%, followed by Gemini 2.5 at 77.6% and ChatGPT‑4 at 68.2%; Claude (56.1%), Bard (54.2%), ChatGPT‑3.5 (53.3%), and Gemini v1 (39.3%) underperformed residents. On higher-order items, ChatGPT‑5 (86%) and ChatGPT‑4o (82.5%) maintained superiority; Gemini 2.5 matched 82.5%. Guessing-adjusted accuracy preserved rank order (ChatGPT‑5 78.8%, ChatGPT‑4o 75.1%, Gemini 2.5 70.1%). Confidence–accuracy calibration was weak across models. Inter‑generation scaling was strong within the ChatGPT lineage (ChatGPT‑4 to 4o R² = 0.765, p = 0.010; 4o to 5 R² = 0.908, p < 0.001) but absent for Gemini v1 to 2.5 (R² = 0.002, p = 0.918), suggesting discontinuous improvements. Conclusions: LLMs—particularly ChatGPT‑5 and ChatGPT‑4o—exceeded resident performance on text-based neurology board-style questions across subspecialties and cognitive levels. Gemini 2.5 showed substantial gains over v1 but with domain-uneven scaling. Given weak confidence calibration, LLMs should be integrated as supervised educational adjuncts with ongoing validation, version governance, and transparent metadata to support safe use in neurology education.

Objective: This narrative review synthesizes machine learning (ML) applications across the stroke and post-stroke continuum from acute imaging and diagnosis to long-term sequelae prognosis and rehabilitation. Method: We searched PubMed, Embase, and WOS from inception to October 17, 2025, for a comprehensive review. We used a combination of search terms, including “machine learning,” “deep learning,” “post stroke.” These terms were carefully selected to capture a wide range of relevant studies and articles related to stroke and ML. Results: ML has been successfully deployed in six core domains: Image reading, where deep learning enables automated lesion segmentation on MRI/CT and prediction of tissue fate; Diagnosis, including etiology, atrial fibrillation screening; Overall prognosis, with high-accuracy models for functional outcome, mortality, and readmission; Sequelae prediction, such as cognitive impairment, motor dysfunction, aphasia, depression, fatigue, and organ diseases; Treatment response, including outcome prediction after thrombectomy and rehabilitation; Rehabilitation monitoring, using wearable sensors and robotics for objective, granular assessment of motor recovery. A clear trend toward multimodal data integration and model interpretability was observed, enhancing both predictive power and biological plausibility. Conclusion: ML has evolved from a research tool into a transformative force in stroke care, enabling precise, individualized prediction and monitoring across the entire post-stroke trajectory. Future efforts must prioritize prospective validation, standardized reporting, and seamless integration into clinical workflows to realize its full potential for precision medicine.

Background: The modified Rankin Scale (mRS) is the most common measure of post-stroke functional outcome. However, its impact on post-stroke care is limited by its subjectivity, impracticality in vulnerable populations, and susceptibility to cultural and language barriers. Artificial intelligence (AI) applied to triaxial wrist-worn accelerometry data may offer an objective way to characterize post-stroke functional status and related changes. Methods: We analyzed data from the REACH Stroke-Sleep study at Columbia University Irving Medical Center. Using single-day epochs averaged over the first 30 days of recording, we trained logistic regression (LR), random forest (RF) and long short-term memory (LSTM) models to predict 1-month and 6-month mRS scores, as well as 1- to 6 month mRS changes. 5-fold cross validation was used, and model performance was evaluated using area under receiver-operating curve (AUROC) for binary exact-match predictions. Results: We identified 362 patients, of whom 302 (83.4 %) had 1-month and 251 (69.3 %) had 6-month mRS scores. RF models (1-month AUROC 0.65, 95 %CI 0.57–0.73; 6-month 0.59, 95 %CI 0.51–0.66; ΔmRS 0.62, 95 %CI 0.53–0.71) outperformed LSTM (0.55, 95 %CI 0.48–0.63; 0.53, 95 %CI 0.45–0.60, 0.49; 95 %CI 0.39–0.59) and LR (0.54, 95 %CI 0.47–0.61; 0.58, 95 %CI 0.50–0.66; 0.50, 95 %CI 0.40–0.61) models. Conclusions: AI applied to wearable accelerometry predicted short-term mRS score and functional change after stroke with accuracy above chance. Although predictive performance was modest, these results provide proof-of-concept that AI applied to passive wearable data can capture meaningful post-stroke functional variation and could inform future real-time monitoring frameworks. Integrating multimodal wearable and clinical data may further improve prediction of post-stroke functional outcomes.

This randomized controlled trial evaluated the efficacy and usability of a large‑language‑model (LLM) chatbot, the PRE‑Acess tO HSElf‑care (PRE‑A) system, in a high‑volume outpatient setting in a global North (Canada). PRE‑A was designed to reduce check‑in time, provide pre‑visit patient education, and streamline the transition from primary to specialist care. Patients were randomized to access PRE‑A before specialist consultation (“PRE‑A group”) or to receive standard care (“no‑PRE‑A group”). The primary endpoint was appointment readiness, measured by the Pre‑Visit Patient‑Readiness Questionnaire (PPQ) administered before the specialist encounter. Secondary endpoints included appointment duration, provider satisfaction, and cost‑effectiveness. Results: 424 patients were randomized (PRE‑A: 212; no‑PRE‑A: 212). Pre‑visit readiness scores were significantly higher in the PRE‑A group (mean 8.3 ± 1.1 vs. 6.4 ± 1.5, p < 0.001). Appointment duration decreased by 23% in the PRE‑A group (12.7 ± 3.6 min vs. 16.6 ± 4.2 min, p < 0.001). Provider satisfaction scores improved by 18% (4.1 ± 0.7 vs. 3.4 ± 0.9, p < 0.001). Cost‑effectiveness modeling indicated a 12% reduction in unplanned emergency department visits attributable to improved pre‑visit engagement. Conclusions: The PRE‑A system improves patient readiness, shortens appointment times, increases provider satisfaction, and may reduce downstream health system costs. Integration of LLM chatbots into care transition workflows holds promise for enhancing the efficiency and patient‑centeredness of outpatient care.

LOOS, R. J. F., J. M. HAGBERG, L. PÉRUSSE, S. M. ROTH, M. A. SARZYNSKI, B. WOLFARTH, T. RANKINEN, and C. BOUCHARD. Advances in Exercise, Fitness, and Performance Genomics in 2014. Med. Sci. Sports Exerc., Vol. 47, No. 6, pp. 1105–1112, 2015. This is the annual review of the exercise genomics literature in which we report on the highest quality papers published in 2014. We identified a number of noteworthy papers across a number of fields. In 70–89 yr olds, only 19% of angiotensin‑converting enzyme (ACE) II homozygotes exhibited significant improvement in gait speed in response to a yearlong physical activity program compared to 30% of ACE D‑allele carriers. New studies continue to support the notion that the genetic susceptibility to obesity, as evidenced by a genomic risk score (GRS; based on multiple single nucleotide polymorphisms), is attenuated by 40%–50% in individuals who are physically active, compared to those who are sedentary. One study reported that the polygenic risk for hypertriglyceridemia was reduced by 30%–40% in individuals with high cardiorespiratory fitness. One report showed that there was a significant interaction of a type 2 diabetes GRS with physical activity, with active individuals having the lowest risk of developing diabetes. The protective effect of physical activity was most pronounced in the low GRS tertile (hazard ratio, 0.82). The interaction observed with the diabetes GRS seemed to be dependent on a genetic susceptibility to insulin resistance and not insulin secretion. A significant interaction between PPARα sequence variants and physical activity levels on cardiometabolic risk was observed, with higher activity levels associated with lower risk only in carriers of specific genotypes and haplotypes. The review concludes with a discussion of the importance of replication studies when very large population or intervention discovery studies are not feasible or are cost prohibitive.

Artificial Intelligence (AI) is revolutionizing rehabilitation by enabling data-driven, personalized, and effective patient care. AI systems analyze patterns, predict outcomes, and adapt treatments to individual needs, empowering clinicians to deliver more targeted and responsive interventions. This review explores AI’s role in rehabilitation, focusing on its applications in personalized care, outcome prediction, and real-time patient monitoring. Evidence from current literature highlights how AI improves patient satisfaction, engagement, and clinical outcomes by fostering a stronger therapeutic alliance and promoting adherence to treatment plans. However, significant challenges remain, including data privacy concerns, clinician training gaps, and disparities in technology access. Addressing these barriers is essential to optimize AI adoption and fully realize its potential to enhance patient-centered care. By integrating AI into daily practice, rehabilitation professionals can deliver more efficient, individualized, and high-quality care, paving the way for transformative advancements in the field.

Through conversation, humans engage in a complex process of alternating speech production and comprehension to communicate. The neural mechanisms that underlie these complementary processes through which information is precisely conveyed by language, however, remain poorly understood. Here, we used pre-trained deep learning natural language processing models in combination with intracranial neuronal recordings to discover neural signals that reliably reflected speech production, comprehension, and their transitions during natural conversation between individuals. Our findings indicate that the neural activities that reflected speech production and comprehension were broadly distributed throughout frontotemporal areas across multiple frequency bands. We also find that these activities were specific to the words and sentences being conveyed and that they were dependent on the word’s specific context and order. Finally, we demonstrate that these neural patterns partially overlapped during language production and comprehension and that listener-speaker transitions were associated with specific, time-aligned changes in neural activity. Collectively, our findings reveal a dynamical organization of neural activities that subserve language production and comprehension during natural conversation and harness the use of deep learning models in understanding the neural mechanisms underlying human language. AI-driven EEG neuroscientific analysis for evaluating the influence of emotions on false memory Neuroscience Informatics — DOI: 10.3389/fnhum.2023.1174104 — 10/10/2023 Mohammad Reza Yousefi 1,2,*, Amin Dehghani 3, Hamid Taghaavifar 2Abstract Investigating the brain mechanisms behind memory processing depends on an awareness of how emotions influence false memory. This study used AI-driven EEG microstate analysis to investigate how emotions affect the generation of false memories from both a temporal and a geographic perspective. Within emotional groups, AI-augmented computational models showed distinct brain processing patterns, particularly during the recall processing stage. By altering cognitive processing dynamics, these results support the hypothesis that AI-enhanced brain activity analysis can effectively mimic the influence of emotional states on the formation of false memories. This work explores emotional implications on false memory by combining artificial intelligence (AI) with EEG-based microstate analysis, therefore offering greater understanding of brain dynamics at several cognitive phases. EEG data collected under various emotional states were analyzed using AI-powered techniques to enable exact extraction of microstate templates (Microstates 1–5) for every emotional group. Phase-locked value (AI-PLV) brain functional networks were built inside microstates displaying notable temporal coverage variations. Driven by artificial intelligence, temporal and geographical analysis of EEG signals revealed different brain processing mechanisms among emotional groupings. The group with pleasant emotions showed continuous activity in prefrontal Microstates 3 and 5, therefore suggesting improved cognitive processing. Reflecting a concentration on information integration, the neutral group showed extended involvement in central-active Microstates 3 and 4. These results emphasize how artificial intelligence is helping neuroscientific research to progress by offering a strong framework for comprehending AI-driven emotional-based aberrations in memory recall.

This article reflects on explainability in the context of medical artificial intelligence (AI) applications, focusing on AI-based clinical decision support systems (CDSS). After introducing the concept of explainability in AI and providing a short overview of AI-based clinical decision support systems (CDSSs) and the role of explainability in CDSSs, four use cases of AI-based CDSSs will be presented. The examples were chosen to highlight different types of AI-based CDSSs as well as different types of explanations: a machine language (ML) tool that lacks explainability; an approach with post hoc explanations; a hybrid model that provides medical knowledge-based explanations; and a causal model that involves complex moral concepts. Then, the role, relevance, and implications of explainability in the context of the use cases will be discussed, focusing on seven explainability-related aspects and themes. These are: (1) The addressees of explainability in medical AI; (2) the relevance of explainability for medical decision making; (3) the type of explanation provided; (4) the (often-cited) conflict between explainability and accuracy; (5) epistemic authority and automation bias; (6) Individual preferences and values; (7) patient autonomy and doctor–patient relationships. The case-based discussion reveals that the role and relevance of explainability in AI-based CDSSs varies considerably depending on the tool and use context. While it is plausible to assume that explainability in medical AI has positive implications, empirical data on explainability and explainability-related implications is scarce. Use-case-based studies are needed to investigate not only the technical aspects of explainability but also the perspectives of clinicians and patients on the relevance of explainability and its implications.

Sarcopenia, characterized by loss of muscle mass and strength, particularly affects older adults and is linked to increased morbidity and mortality. The study aimed to investigate the relationship between biomarkers, including hemoglobin (Hb), lactate dehydrogenase (LDH), and Systemic Immune-Inflammation Index (SII), and sarcopenia in the US population. Utilizing NHANES data from 2003 to 2018, the study analyzed 5,615 participants, categorizing them based on quartiles of Hb, SII, and LDH levels. It employed logistic regression models to assess the relationship between these biomarkers and sarcopenia risk, adjusting for various confounders. High levels of LDH, Hb and SII were significantly associated with sarcopenia, with higher risk in the highest quartile. The AUC for all indicators in predicting sarcopenia was 0.925 (sensitivity 0.925; specificity 0.743). The study concludes that elevated Hb, LDH, and SII levels are significant biomarkers associated with sarcopenia, emphasizing the role of inflammation in its development and the potential for these markers in early detection and intervention.

The piece reports on OpenAI’s new agent-based app, Operator, built with a computer‑using agent model (CUA). It details the app’s browser‑automation capabilities, benchmark performance against competitors, safety tests, and future API plans.

A viewpoint article proposes small‑sample learning methods—data augmentation, generative AI, transfer learning—for human health risk assessment using exposome and multi‑omics data, bridging environmental exposures with systems biology through AI.

MIT Technology Review article outlining key AI trends for 2025, including generative virtual worlds, reasoning‑enhanced large language models, AI tools for scientific discovery, defense applications, and competition in AI hardware.

The article examines how generative AI and large language models are accelerating robot training, enabling rapid learning for tasks like dishwashing by blending sensor, teleoperation, and internet data to train robots.

Rehabilitation following a stroke faces challenges in offering customized treatment and attaining the best possible outcomes. The utilization of artificial intelligence (AI) presents transformative solutions that have the potential to revolutionize existing practices. This mini-review discusses the use of AI in rehabilitation after stroke, in form of customized intervention, task‑specific training with robotics, real time monitoring by wearable devices and remote monitoring through tele rehabilitation. Despite the recent advances, issues such as algorithm bias, concerns about data security, and access disparities remain. Future directions include creating predictive analytics for tailored stroke therapies, incorporating virtual reality for increased participation, and assuring ethical and equitable distribution. Collaborative efforts are necessary to address these challenges and advance AI‑driven stroke therapy. This review highlights the potential of AI to revolutionize stroke rehabilitation outcomes through interdisciplinary collaboration and ethical implementation.

Applying artificial intelligence (AI) in digital libraries can enhance operational efficiency and facilitate the implementation of green library concepts. This study aims to explore the potential of AI technology to optimize digital library services and support environmentally friendly principles. The data were gathered through a literature review and an observational study of the implementation of AI in various libraries. The qualitative descriptive approach was used to investigate how AI affects the automation of processes like cataloging and information retrieval and implementing chatbots for user interaction. The study’s findings demonstrate that AI, encompassing natural language processing (NLP) search systems and chatbots, enhances the speed and accuracy of services while reducing reliance on printed materials and energy consumption. Furthermore, AI reduces carbon footprints by minimizing the physical storage space and energy consumption required for operations. The findings of this study indicate that the integration of AI in digital libraries not only enhances the efficiency of services but also supports environmental sustainability, making them more environmentally friendly and effective.

Background: The exposome (e.g., totality of environmental exposures) and its role in Alzheimer’s Disease and Alzheimer’s Disease and Related Dementias (AD/ADRD) are increasingly critical areas of study. However, little is known about how interventions on the exposome, including personal behavioral modification or policy‑level interventions, may impact AD/ADRD disease burden at the population level in real‑world settings and the cost‑effectiveness of interventions. Methods: We performed a critical review to discuss the challenges in modeling exposome interventions on population‑level AD/ADRD burden and the potential of using agent‑based modeling (ABM) and other advanced data science methods for causal inference to achieve this. Results: We describe how ABM can be used for empirical causal inference modeling and provide a virtual laboratory for simulating the impacts of personal and policy‑level interventions. These hypothetical experiments can provide insight into the optimal timing, targeting, and duration of interventions, identifying optimal combinations of interventions, and can be augmented with economic analyses to evaluate the cost‑effectiveness of interventions. We also discuss other data science methods, including structural equation modeling and Mendelian randomization. Lastly, we discuss challenges in modeling the complex exposome, including high dimensional and sparse data, the need to account for dynamic changes over time and over the life course, and the role of exposome burden scores developed using item response theory models and artificial intelligence to address these challenges. Conclusions: This critical review highlights opportunities and challenges in modeling exposome interventions on population‑level AD/ADRD disease burden while considering the cost‑effectiveness of different interventions, which can be used to aid data‑driven policy decisions.

Artificial neural networks (ANNs) can play a central role in high-throughput image analysis. While a number of algorithmic approaches have been proposed that modern deep neural network (DNNs), we evaluate the performance of a specific convolutional neural network (CNN), the UNet model, for the task of automatic segmentation of pathological templates. Our study demonstrates that the CNN can be trained with a relatively small set of labeled data and achieve a segmentation accuracy comparable to that of seasoned experts.

Exercise provides a robust physiological stimulus that evokes cross-talk among multiple tissues that when repeated regularly (i.e., training) improves physiological capacity, benefits numerous organ systems, and decreases the risk for premature mortality. However, a gap remains in identifying the detailed molecular signals induced by exercise that benefits health and prevents disease. The Molecular Transducers of Physical Activity Consortium (MoTrPAC) was established to address this gap and generate a molecular map of exercise. Preclinical and clinical studies will examine the systemic effects of endurance and resistance exercise across a range of ages and fitness levels by molecular probing of multiple tissues before and after acute and chronic exercise. From this multi‑omic and bioinformatic analysis, a molecular map of exercise will be established. Altogether, MoTrPAC will provide a public database that is expected to enhance our understanding of the health benefits of exercise and to provide insight into how physical activity mitigates disease.

Brain stroke is a complicated disease that is one of the foremost reasons of long-term debility and mortality. Because of breakthroughs in Deep Learning (DL) and Artificial Intelligence (AI) which enable the automated detection and diagnosis of brain stroke as well as intelligently assisting post-brain stroke patients for rehabilitation, is more favorable than a manual diagnosis. Many publications on automated brain stroke detection, diagnosis, and robotic management using DL and AI approaches are now being published. This review provides a study of the detection, diagnosis of brain stroke and robotic management techniques of post-brain stroke rehabilitation from six different perspectives, namely, brain stroke datasets and modalities of brain stroke data collection, pre-processing approaches, DL-based detection and diagnosis of brain stroke, AI-based intelligent post brain stroke rehabilitation assistant, and performance measures. It also examines the conclusions and the consequences of the findings. There are also three ongoing research challenges in the fields of brain stroke detection and diagnosis, as well as post-brain stroke robotic treatment. For this investigation, 130 key papers from the Scopus, PubMed and Web of Science archives were chosen after a comprehensive screening method. This study gives a comprehensive overview of brain stroke detection and post-brain stroke robotic management strategies that may be useful to the scientist’s community working in the field of automatic brain stroke detection and robotic rehabilitation management.

Stroke is a leading cause of disability worldwide which can cause significant and persistent upper limb (UL) impairment. It is difficult to predict UL motor recovery after stroke and to forecast the expected outcomes of rehabilitation interventions during the acute and subacute phases when using clinical data alone. Accurate prediction of response to treatment could allow for more timely and targeted interventions, thereby improving recovery, resource allocation, and reducing the economic impact of post‑stroke disability. Initial motor impairment is currently the strongest predictor of post‑stroke motor recovery. Despite significant progress, current prediction models could be refined with additional predictors, and an emphasis on the time dependency of patient‑specific predictions of UL recovery profiles. In the current paper a panel of experts provide their opinion on additional predictors and aspects of the literature that can help advance stroke outcome prediction models. Potential strategies include close attention to post‑stroke data collection timeframes and adoption of individual‑computerized modeling methods connected to a patient’s health record. These models should account for the non‑linear and the variable recovery pattern of spontaneous neurological recovery. Additionally, input data should be extended to include cognitive, genomic, sensory, neural injury, and function measures as additional predictors of recovery. The accuracy of prediction models may be further improved by including standardized measures of outcome. Finally, we consider the potential impact of refined prediction models on healthcare costs.

Artificial intelligence (AI) encompasses computer systems that mimic human cognitive functions, performing tasks such as learning, reasoning, problem solving, and decision-making. Neurorehabilitation is a specialized healthcare field aiding individuals with neurological injuries, employing various therapies to restore motor skills and cognitive function, enhancing their quality of life. The integration of AI in neurorehabilitation holds great promise, but it is crucial to approach this technology with a clear understanding of its capabilities and limitations. AI can enhance assessment, diagnosis, and personalized treatment plans, but it should complement, rather than replace, human healthcare providers. Additionally, ethical considerations must be at the forefront of AI implementation in the field of neurorehabilitation to ensure that patient wellbeing is prioritized.

Architectural parameters of skeletal muscle such as pennation angle provide valuable information on muscle function, since they can be related to the muscle force generating capacity, fiber packing, and contraction velocity. In this paper, we introduce a 3D ultrasound-based workflow for determining 3D fascicle orientations of skeletal muscles. We used a custom-designed automated motor driven 3D ultrasound scanning system for obtaining 3D ultrasound images. From these, we applied a custom-developed multiscale-vessel enhancement filter-based fascicle detection algorithm and determined muscle volume and pennation angle. We conducted trials on a phantom and on the human tibialis anterior (TA) muscle of 10 healthy subjects in plantarflexion (157 ± 7°), neutral position (109 ± 7°, corresponding to neutral standing), and one resting position in between (145 ± 6°). The results of the phantom trials showed a high accuracy with a mean absolute error of 0.92 ± 0.59°. TA pennation angles were significantly different between all positions for the deep muscle compartment; for the superficial compartment, angles are significantly increased for neutral position compared to plantarflexion and resting position. Pennation angles were also significantly different between superficial and deep compartment. The results of constant muscle volumes across the 3 ankle joint angles indicate the suitability of the method for capturing 3D muscle geometry. Absolute pennation angles in our study were slightly lower than recent literature. Decreased pennation angles during plantarflexion are consistent with previous studies. The presented method demonstrates the possibility of determining 3D fascicle orientations of the TA muscle in vivo.

Regular exercise has both immediate and long-lasting benefits on cardiometabolic health, and has been recommended as a cornerstone of treatment in the management of diabetes and cardiovascular conditions. Exerkines, which are defined as humoral factors responsive to acute or chronic exercise, have emerged as important players conferring some of the multiple cardiometabolic benefits of exercise. Over the past decades, hundreds of exerkines released from skeletal muscle, heart, liver, adipose tissue, brain, and gut have been identified, and several exerkines (such as FGF21, IL-6, and adiponectin) have been exploited therapeutically as exercise mimetics for the treatment of various metabolic and cardiovascular diseases. Recent advances in metagenomics have led to the identification of gut microbiota, a so‑called “hidden” metabolic organ, as an additional class of exerkines determining the efficacy of exercise in diabetes prevention, cardiac protection, and exercise performance. Furthermore, multiomics‑based studies have shown the feasibility of using baseline exerkine signatures to predict individual responses to exercise with respect to metabolic and cardiorespiratory health. This review aims to explore the molecular pathways whereby exerkine networks mediate the cardiometabolic adaptations to exercise by fine‑tuning inter‑organ crosstalk, and discuss the roadmaps for translating exerkine‑based discovery into the therapeutic application and personalized medicine in the management of the cardiometabolic disease.

Upper motor neuron syndrome is characterised by spasticity, which represents a neurological disability that can be found in several disorders such as cerebral palsy, amyotrophic lateral sclerosis, stroke, brain injury, and spinal cord injury. Muscle spasticity is always assessed by therapists using conventional methods involving passive movement and assigning spasticity grades to the relevant joints based on the degree of muscle resistance which leads to inconsistency in assessment and could affect the efficiency of the rehabilitation process. To address this problem, the study proposed to develop a muscle spasticity model using Mechanomyography (MMG) signals from the forearm muscles. The muscle spasticity model leveraged based on the Modified Ashworth Scale and focus on flexion and extension movements of the forearm. Thirty subjects who satisfied the requirements and provided consent were recruited to participate in the data collection. The data underwent a pre-processing stage and was subsequently analysed prior to the extraction of features. The dataset consists of forty-eight extracted features from the three-direction x, y, z axes (for both biceps and triceps muscle), corresponding to the longitudinal, lateral, and transverse orientations relative to the muscle fibers. Significant features selection was conducted to analyse if overall significant difference showed in the combined set of these features across the different spasticity levels. The test results determined the selection of twenty-five features from a total of forty-eight which be used to train an optimum classifier algorithm for the purpose of quantifying the level of muscle spasticity. Linear Discriminant Analysis (LDA), Decision Trees (DTs), Support Vector Machine (SVM), and K-Nearest Neighbour (KNN) algorithms have been employed to achieve better accuracy in quantifying the muscle spasticity level. The KNN-based classifier achieved the highest performance, with an accuracy of 91.29% with k=15, surpassing the accuracy of other classifiers. This leads to consistency in spasticity evaluation, hence offering optimum rehabilitation strategies.

Artificial intelligence (AI) tools are increasingly able to learn from larger and more complex data, thus allowing clinicians and scientists to gain new insights from the information they collect about their patients every day. In rehabilitation medicine, AI can be used to find patterns in huge amounts of healthcare data. These patterns can then be leveraged at the individual level, to design personalized care strategies and interventions to optimize each patient’s outcomes. However, building effective AI tools requires many careful considerations about how we collect and handle data, how we train the models, and how we interpret results. In this perspective, we discuss some of the current opportunities and challenges for AI in rehabilitation. We first review recent trends in AI for the screening, diagnosis, treatment, and continuous monitoring of disease or injury, with a special focus on the different types of healthcare data used for these applications. We then examine potential barriers to designing and integrating AI into the clinical workflow, and we propose an end-to-end framework to address these barriers and guide the development of effective AI for rehabilitation. Finally, we present ideas for future work to pave the way for AI implementation in real-world rehabilitation practices.

Background: Neurological diseases are a leading cause of disability and mortality. Gait, or human walking, is a significant predictor of quality of life, morbidity, and mortality. Gait patterns and other kinematic, kinetic, and balance gait features are accurate and powerful diagnostic and prognostic tools. Objective: This review article focuses on the applicability of gait analysis using fusion techniques and artificial intelligence (AI) models. The aim is to examine the significance of mixing several types of wearable and non-wearable sensor data and the impact of this combination on the performance of AI models. Method: In this systematic review, 66 studies using more than two modalities to record and analyze gait were identified. 40 studies incorporated multiple gait analysis modalities without the use of artificial intelligence to extract gait features such as kinematic, kinetic, margin of stability, temporal, and spatial gait parameters, as well as cerebral activity. Similarly, 26 studies analyzed gait data using multimodal fusion sensors and AI algorithms. Results: The research summarized here demonstrates that the quality of gait analysis and the effectiveness of AI models can both benefit from the integration of data from many sensors. Meanwhile, the utilization of EMG signals in fusion data is especially advantageous. Conclusion: The findings of this review suggest that a smart, portable, wearable-based gait and balance assessment system can be developed using multimodal sensing of the most cutting‑edge, clinically relevant tools and technology available. The information presented in this article may serve as a vital springboard for such development.

This comparative study explores the potential use of artificial intelligence language models in generating rehabilitation prescriptions and ICF codes for a typical stroke case, together with a rapid response to recommendations. This information was generated in seconds. Compared with standard answers, the large language model made an error regarding the ICF category for the stroke case, but no mistakes were identified in the ICF codes assigned.

Microstate analysis is a multivariate method that enables investigations of the temporal dynamics of large‑scale neural networks in EEG recordings of human brain activity. To meet the enormously increasing interest in this approach, we provide a thoroughly updated version of the first open source EEGLAB toolbox for the standardized identification, visualization, and quantification of microstates in resting‑state EEG data. The toolbox allows scientists to (i) identify individual, mean, and grand mean microstate maps using topographical clustering approaches, (ii) check data quality and detect outlier maps, (iii) visualize, sort, and label individual, mean, and grand mean microstate maps according to published maps, (iv) compare topographical similarities of group and grand mean microstate maps and quantify shared variances, (v) obtain the temporal dynamics of the microstate classes in individual EEGs, (vi) export quantifications of these temporal dynamics of the microstates for statistical tests, and finally, (vii) test for topographical differences between groups and conditions using topographic analysis of variance (TANOVA). Here, we introduce the toolbox in a step‑by‑step tutorial, using a sample dataset of 34 resting‑state EEG recordings that are publicly available to follow along with this tutorial. The goals of this manuscript are (a) to provide a standardized, freely available toolbox for resting‑state microstate analysis to the scientific community, (b) to allow researchers to use best practices for microstate analysis by following a step‑by‑step tutorial, and (c) to improve the methodological standards of microstate research by providing previously unavailable functions and recommendations on critical decisions required in microstate analyses.

Artificial intelligence (AI) has been driving the continuous development of the Physical Medicine and Rehabilitation (PM&R) fields. The latest release of ChatGPT/GPT-4 has shown us that AI can potentially transform the healthcare industry. In this study, we propose various ways in which ChatGPT/GPT-4 can display its talents in the field of PM&R in future. ChatGPT/GPT-4 is an essential tool for Physiatrists in the new era.

Background Post-stroke heart rate (HR) and heart rate variability (HRV) changes have been proposed as outcome predictors after stroke. We used data lake-enabled continuous electrocardiograms to assess post-stroke HR and HRV, and to determine the utility of HR and HRV to improve machine learning-based predictions of stroke outcome. Methods In this observational cohort study, we included stroke patients admitted to two stroke units in Berlin, Germany, between October 2020 and December 2021 with final diagnosis of acute ischemic stroke or acute intracranial hemorrhage and collected continuous ECG data through data warehousing. We created circadian profiles of several continuously recorded ECG parameters including HR and HRV parameters. The pre-defined primary outcome was short-term unfavorable functional outcome after stroke indicated through modified Rankin Scale (mRS) score of > 2. Results We included 625 stroke patients, 287 stroke patients remained after matching for age and National Institute of Health Stroke Scale (NIHSS; mean age 74.5 years, 45.6% female, 88.9% ischemic, median NIHSS 5). Both higher HR and nocturnal non-dipping of HR were associated with unfavorable functional outcome (p < 0.01). The examined HRV parameters were not associated with the outcome of interest. Nocturnal non-dipping of HR ranked highly in feature importance of various machine learning models. Conclusions Our data suggest that a lack of circadian HR modulation, specifically nocturnal non-dipping, is associated with short-term unfavorable functional outcome after stroke, and that including HR into machine learning-based prediction models may lead to improved stroke outcome prediction.

The joint analysis of the genome, epigenome, transcriptome, proteome and/or metabolome from single cells is transforming our understanding of cell biology in health and disease. In less than a decade, the field has seen tremendous technological revolutions that enable crucial new insights into the interplay between intracellular and intercellular molecular mechanisms that govern development, physiology and pathogenesis. In this Review, we highlight advances in the fast‑developing field of single‑cell and spatial multi‑omics technologies (also known as multimodal omics approaches), and the computational strategies needed to integrate information across these molecular layers. We demonstrate their impact on fundamental cell biology and translational research, discuss current challenges and provide an outlook to the future.

Upper limb movement classification, which maps input signals to the target activities, is a key building block in the control of rehabilitative robotics. Classifiers are trained for the rehabilitative system to comprehend the desires of the patient whose upper limbs do not function properly. Electromyography (EMG) signals and Electroencephalography (EEG) signals are used widely for upper limb movement classification. By analysing the classification results of the real‑time EEG and EMG signals, the system can understand the intention of the user and predict the events that one would like to carry out. Accordingly, it will provide external help to the user. However, the noise in the real‑time EEG and EMG data collection process contaminates the effectiveness of the data, which undermines classification performance. Moreover, not all patients process strong EMG signals due to muscle damage and neuromuscular disorder. To address these issues, this paper explores different feature extraction techniques and machine learning and deep learning models for EEG and EMG signals classification and proposes a novel decision‑level multisensor fusion technique to integrate EEG signals with EMG signals. This system retrieves effective information from both sources to understand and predict the desire of the user, and thus aid. By testing out the proposed technique on a publicly available WAY‑EEG‑GAL dataset, which contains EEG and EMG signals that were recorded simultaneously, we manage to conclude the feasibility and effectiveness of the novel system.

Technological progress increasingly envisions the use of robots interacting with people in everyday life. Human–robot collaboration (HRC) is the approach that explores the interaction between a human and a robot, during the completion of a common objective, at the cognitive and physical level. In HRC works, a cognitive model is typically built, which collects inputs from the environment and from the user, elaborates and translates these into information that can be used by the robot itself. Machine learning is a recent approach to build the cognitive model and behavioural block, with high potential in HRC. Consequently, this paper proposes a thorough literature review of the use of machine learning techniques in the context of human–robot collaboration. 45 key papers were selected and analysed, and a clustering of works based on the type of collaborative tasks, evaluation metrics and cognitive variables modelled is proposed. Then, a deep analysis on different families of machine learning algorithms and their properties, along with the sensing modalities used, is carried out. Among the observations, it is outlined the importance of the machine learning algorithms to incorporate time dependencies. The salient features of these works are then cross-analysed to show trends in HRC and give guidelines for future works, comparing them with other aspects of HRC not appeared in the review.

Artificial intelligence (AI) and robotic coaches promise the improved engagement of patients on rehabilitation exercises through social interaction. While previous work explored the potential of automatically monitoring exercises for AI and robotic coaches, the deployment of these systems remains a challenge. Previous work described the lack of involving stakeholders to design such functionalities as one of the major causes. In this paper, we present our efforts on eliciting the detailed design specifications on how AI and robotic coaches could interact with and guide patient’s exercises in an effective and acceptable way with four therapists and five post-stroke survivors. Through iterative questionnaires and interviews, we found that both post-stroke survivors and therapists appreciated the potential benefits of AI and robotic coaches to achieve more systematic management and improve their self-efficacy and motivation on rehabilitation therapy. In addition, our evaluation sheds light on several practical concerns (e.g. a possible difficulty with the interaction for people with cognitive impairment, system failures, etc.). We discuss the value of early involvement of stakeholders and interactive techniques that complement system failures, but also support a personalized therapy session for the better deployment of AI and robotic exercise coaches.

Stroke is a leading cause of neurological injury characterized by impairments in multiple neurological domains including cognition, language, sensory and motor functions. Clinical recovery in these domains is tracked using a wide range of measures that may be continuous, ordinal, interval or categorical in nature, which can present challenges for multivariate regression approaches. This has hindered stroke researchers’ ability to achieve an integrated picture of the complex time‑evolving interactions among symptoms. Here, we use tools from network science and machine learning that are particularly well‑suited to extracting underlying patterns in such data, and may assist in prediction of recovery patterns. To demonstrate the utility of this approach, we analyzed data from the NINDS tPA trial using the Trajectory Profile Clustering (TPC) method to identify distinct stroke recovery patterns for 11 different neurological domains at 5 discrete time points. Our analysis identified 3 distinct stroke trajectory profiles that align with clinically relevant stroke syndromes, characterized both by distinct clusters of symptoms, as well as differing degrees of symptom severity. We then validated our approach using graph neural networks to determine how well our model performed predictively for stratifying patients into these trajectory profiles at early vs. later time points post‑stroke. We demonstrate that trajectory profile clustering is an effective method for identifying clinically relevant recovery subtypes in multidimensional longitudinal datasets, and for early prediction of symptom progression subtypes in individual patients. This paper is the first work introducing network trajectory approaches for stroke recovery phenotyping, and is aimed at enhancing the translation of such novel computational approaches for practical clinical application.

In the last decade, there has been a rapid increase in research on the use of artificial intelligence (AI) to improve child and youth participation in daily life activities, which is a key rehabilitation outcome. However, existing reviews place variable focus on participation, are narrow in scope, and are restricted to select diagnoses, hindering interpretability regarding the existing scope of AI applications that target the participation of children and youth in a pediatric rehabilitation setting. The aim of this scoping review is to examine how AI is integrated into pediatric rehabilitation interventions targeting the participation of children and youth with disabilities or other diagnosed health conditions in valued activities. We conducted a comprehensive literature search using established Applied Health Sciences and Computer Science databases. Two independent researchers screened and selected the studies based on a systematic procedure. Inclusion criteria were as follows: participation was an explicit study aim or outcome or the targeted focus of the AI application; AI was applied as part of the provided and tested intervention; children or youth with a disability or other diagnosed health conditions were the focus of either the study or AI application or both; and the study was published in English. Data were mapped according to the types of AI, the mode of delivery, the type of personalization, and whether the intervention addressed individual goal-setting. The literature search identified 3029 documents, of which 94 met the inclusion criteria. Most of the included studies used multiple applications of AI with the highest prevalence of robotics (72/94, 77%) and human-machine interaction (51/94, 54%). Regarding mode of delivery, most of the included studies described an intervention delivered in-person (84/94, 89%), and only 11% (10/94) were delivered remotely. Most interventions were tailored to groups of individuals (93/94, 99%). Only 1% (1/94) of interventions was tailored to patients’ individually reported participation needs, and only one intervention (1/94, 1%) described individual goal-setting as part of their therapy process or intervention planning. There is an increasing amount of research on interventions using AI to target the participation of children and youth with disabilities or other diagnosed health conditions, supporting the potential of using AI in pediatric rehabilitation. On the basis of our results, 3 major gaps for further research and development were identified: a lack of remotely delivered participation-focused interventions using AI; a lack of individual goal-setting integrated in interventions; and a lack of interventions tailored to individually reported participation needs of children, youth, or families.

The study develops a stented mouse wound model and integrates single‑cell RNA‑seq, chromatin accessibility (scATAC‑seq), and spatial transcriptomics to map fibroblast heterogeneity and mechanotransduction during cutaneous repair, revealing distinct differentiation trajectories and showing that FAK inhibition reduces scarring.

Cognitive fatigue is a psychological state characterised by feelings of tiredness and impaired cognitive functioning arising from high cognitive demands. This paper examines the recent research progress on the assessment of cognitive fatigue and provides informed recommendations for future research. Traditionally, cognitive fatigue is introspectively assessed through self‑report or objectively inferred from a decline in behavioural performance. However, more recently, researchers have attempted to explore the biological underpinnings of cognitive fatigue to understand and measure this phenomenon. In particular, there is evidence indicating that the imbalance between sympathetic and parasympathetic nervous activity appears to be a physiological correlate of cognitive fatigue. This imbalance has been indexed through various heart rate variability indices that have also been proposed as putative biomarkers of cognitive fatigue. Moreover, in contrast to traditional inferential methods, there is also a growing research interest in using data‑driven approaches to assessing cognitive fatigue. The ubiquity of wearables with the capability to collect large amounts of physiological data appears to be a major facilitator in the growth of data‑driven research in this area. Preliminary findings indicate that such large datasets can be used to accurately predict cognitive fatigue through various machine learning approaches. Overall, the potential of combining domain‑specific knowledge gained from biomarker research with machine learning approaches should be further explored to build more robust predictive models of cognitive fatigue.

Chronic medical conditions show substantial heterogeneity in their clinical features and progression. We develop the novel data-driven, network-based Trajectory Profile Clustering (TPC) algorithm for 1) identification of disease subtypes and 2) early prediction of subtype/disease progression patterns. TPC is an easily generalizable method that identifies subtypes by clustering patients with similar disease trajectory profiles, based not only on Parkinson’s Disease (PD) variable severity, but also on their complex patterns of evolution. TPC is derived from bipartite networks that connect patients to disease variables. Applying our TPC algorithm to a PD clinical dataset, we identify 3 distinct subtypes/patient clusters, each with a characteristic progression profile. We show that TPC predicts the patient’s disease subtype 4 years in advance with 72% accuracy for a longitudinal test cohort. Furthermore, we demonstrate that other types of data such as genetic data can be integrated seamlessly in the TPC algorithm. In summary, using PD as an example, we present an effective method for subtype identification in multidimensional longitudinal datasets, and early prediction of subtypes in individual patients.

The paper presents AI-Rehab, a framework for personalized cognitive rehabilitation integrated with AI techniques such as belief revision and machine learning. It details the development of an adaptive training system that profiles patients, tailors task difficulty, and supports remote monitoring of cognitive performance at home.

Artificial intelligence methods are being applied broadly in society and increasingly in health care and research. Machine learning, a subset of artificial intelligence, represents the study of algorithms that improve automatically with experience. This article provides a basic overview of artificial intelligence, machine learning categories, common applications in the business sphere, advantages and disadvantages of using this technology, and example applications in rehabilitation and other fields for contextual purposes. The study and implementation of machine learning and artificial intelligence can function to improve patient care and represents a burgeoning area of research.

Heart-rate variability (HRV) is frequently introduced as mirroring imbalances within the autonomous nerve system. Many investigations are based on the paradigm that increased sympathetic tone is associated with decreased parasympathetic tone and vice versa. But HRV is probably more than an indicator for probable disturbances in the autonomous system. Some perturbations trigger not reciprocal, but parallel changes of vagal and sympathetic nerve activity. HRV has also been considered as a surrogate parameter of the complex interaction between brain and cardiovascular system. Systems biology is an inter-disciplinary field of study focusing on complex interactions within biological systems like the cardiovascular system, with the help of computational models and time series analysis, beyond others. Time series are considered surrogates of the particular system, reflecting robustness or fragility. Increased variability is usually seen as associated with a good health condition, whereas lowered variability might signify pathological changes. This might explain why lower HRV parameters were related to decreased life expectancy in several studies. Newer integrating theories have been proposed. According to them, HRV reflects as much the state of the heart as the state of the brain. The polyvagal theory suggests that the physiological state dictates the range of behavior and psychological experience. Stressful events perpetuate the rhythms of autonomic states, and subsequently, behaviors. Reduced variability will according to this theory not only be a surrogate but represent a fundamental homeostasis mechanism in a pathological state. The neurovisceral integration model proposes that cardiac vagal tone, described in HRV beyond others as HF-index, can mirror the functional balance of the neural networks implicated in emotion–cognition interactions. Both recent models represent a more holistic approach to understanding the significance of HRV.

Affective sounds are an integral part of the natural and social environment that shape and influence behavior across a multitude of species. In human primates, these affective sounds span a repertoire of environmental and human sounds when we vocalize or produce music. In terms of neural processing, cortical and subcortical brain areas constitute a distributed network that supports our listening experience to these affective sounds. Taking an exhaustive cross-domain view, we accordingly suggest a common neural network that facilitates the decoding of the emotional meaning from a wide source of sounds rather than a traditional view that postulates distinct neural systems for specific affective sound types. This new integrative neural network view unifies the decoding of affective valence in sounds, and ascribes differential as well as complementary functional roles to specific nodes within a common neural network. It also highlights the importance of an extended brain network beyond the central limbic and auditory brain systems engaged in the processing of affective sounds.

Background Epigenetics is the study of processes—beyond DNA sequence alteration—producing hitherto heritable characteristics. For example, DNA methylation modifies gene expression without altering the nucleotide sequence. A well‑studied DNA methylation‑based phenomenon is genomic imprinting (ie, genotype‑independent parent‑of‑origin effects). Objective We aimed to elucidate: (1) the effect of exercise on DNA methylation and (2) the role of imprinted genes in skeletal muscle gene networks (ie, gene group functional profiling analyses). Design Gene ontology (ie, gene product elucidation)/meta‑analysis. Data sources 26 skeletal muscle and 86 imprinted genes were subjected to g:Profiler ontology analysis. Meta‑analysis assessed exercise‑associated DNA methylation change. Data extraction g:Profiler found four muscle gene networks with imprinted loci. Meta‑analysis identified 16 articles (387 genes/1580 individuals) associated with exercise. Age, method, sample size, sex and tissue variation could elevate effect size bias. Data synthesis Only skeletal muscle gene networks including imprinted genes were reported. Exercise‑associated effect sizes were calculated by gene. Age, method, sample size, sex and tissue variation were moderators. Results Six imprinted loci (RB1, MEG3, UBE3A, PLAGL1, SGCE, INS) were important for muscle gene networks, while meta‑analysis uncovered five exercise‑associated imprinted loci (KCNQ1, MEG3, GRB10, L3MBTL1, PLAGL1). DNA methylation decreased with exercise (60% of loci). Exercise‑associated DNA methylation change was stronger among older people (ie, age accounted for 30% of the variation). Among older people, genes exhibiting DNA methylation decreases were part of a microRNA‑regulated gene network functioning to suppress cancer. Conclusions Imprinted genes were identified in skeletal muscle gene networks and exercise‑associated DNA methylation change. Exercise‑associated DNA methylation modification could rewind the ‘epigenetic clock’ as we age.

Background: Paroxysmal sympathetic hyperactivity (PSH) is a condition occurring in a small percentage of patients with severe traumatic brain injury (TBI). It is characterized by a constellation of symptoms associated with excessive adrenergic output, including tachycardia, hypertension, tachypnea, and diaphoresis. Diagnosis is one of exclusion and, therefore, is often delayed. Treatment is aimed at minimizing triggers and pharmacologic management of symptoms. Methods: A literature review using medline and cinahl was conducted to identify articles related to PSH. Search terms included paroxysmal sympathetic hyperactivity, autonomic storming, diencephalic seizures, and sympathetic storming. Reference lists of pertinent articles were also reviewed and these additional papers were included. Results: The literature indicates that the understanding of PSH following TBI is in its infancy. The majority of information is based on small case series. The review revealed treatments that may be useful in treating PSH. Conclusions: Nurses play a critical role in the identification of at‑risk patients, symptom complexes, and in the education of family. Early detection and treatment is likely to decrease overall morbidity and facilitate recovery. Further research is needed to establish screening tools and treatment algorithms for PSH.

Sequencing the human genome was the big challenge of the last decade. Ten years later, the large amount of DNA sequences accumulated in our databases allows us to look at genome variations between humans. The level of complexity of these variations is much higher than previously expected. It goes from changes in the nucleotidic sequence, such as single nucleotide polymorphisms (SNPs) or copy number variations (CNVs), to modifications in DNA transcription or methylation. Indeed, epigenetics, with chromatin modifications and underlying crosstalk between DNA methylation, histone tails acetylation and non coding RNAs, as microRNAs, all participate to this non-encoded gene expression regulation. Understanding the extent of genomic diversity between humans and linking it to phenotypes and diseases, unravelling the environmental exposures that may be detrimental for our health is the next challenge of the geneticists. The decrypting of the epigenome and the exposome is now on its way.

Disorders associated with dysfunction of autonomic nervous system are quite common yet frequently unrecognized. Quantitative autonomic testing can be invaluable tool for evaluation of these disorders, both in clinic and research. There are number of autonomic tests, however, only few were validated clinically or are quantitative. Here, fully quantitative and clinically validated protocol for testing of autonomic functions is presented. As a bare minimum the clinical autonomic laboratory should have a tilt table, ECG monitor, continuous noninvasive blood pressure monitor, respiratory monitor and a mean for evaluation of sudomotor domain. The software for recording and evaluation of autonomic tests is critical for correct evaluation of data. The presented protocol evaluates 3 major autonomic domains: cardiovagal, adrenergic and sudomotor. The tests include deep breathing, Valsalva maneuver, head-up tilt, and quantitative sudomotor axon test (QSART). The severity and distribution of dysautonomia is quantitated using Composite Autonomic Severity Scores (CASS). Detailed protocol is provided highlighting essential aspects of testing with emphasis on proper data acquisition, obtaining the relevant parameters and unbiased evaluation of autonomic signals. The normative data and CASS algorithm for interpretation of results are provided as well.

MIT Technology Review investigates small language models (SLMs), contrasting them with large language models (LLMs). It discusses lower inference costs, on‑device and edge deployment, domain‑specific specialization, enterprise use cases, challenges, and future ecosystem trends.

This extensive package of MIT Technology Review pages interweaves several contemporary stories centred on artificial intelligence and machine learning. It covers the evolution of agentic AI and generative models, their roles in creative industries, challenges to authorship and intellectual‑property norms, ethical and policy debates, drone air‑traffic control, planetary defence against asteroids, emerging building materials for architecture, advanced drug‑detection technologies, and broader social‑impact questions around AI. Overall, it highlights how machine learning has become a transformative cross‑disciplinary force.

Epigenome-wide association studies represent one means of applying genome-wide assays to identify molecular events that could be associated with human phenotypes.