The startup trains embodied AI and world models using Medal’s dataset of 2 billion videos per year from 10 million monthly active users.

In a groundbreaking advancement that could revolutionize cancer prognosis and treatment, a team of researchers led by Liu, X., Dai, F., and Dai, J., has unveiled a sophisticated multimodal multitask deep learning system for grading management in non-small cell lung cancer (NSCLC). Published recently in Nature Communications, this innovative approach integrates diverse data modalities through an advanced artificial intelligence framework, demonstrating unprecedented accuracy and efficiency in tumor grading and patient stratification—key factors in tailoring personalized therapies and improving patient outcomes.

Non-small cell lung cancer remains one of the leading causes of cancer-related mortality worldwide, largely due to its heterogeneity and the challenge of precise tumor grading. Traditional grading methods rely heavily on histopathological analysis, which, while indispensable, faces limitations in capturing the complex biological and morphological variations across patients. This novel multimodal multitask network leverages deep learning’s capacity to extract nuanced features across multiple data types, including histology images, radiographic scans, and molecular profiles, effectively capturing the multidimensional nature of the disease.

The core innovation of this system lies in its multitask learning architecture, which simultaneously addresses multiple clinical endpoints—tumor grading, staging, and survival prediction—within a unified model. This approach contrasts with previous single-task models that focus on one predictive target, often neglecting the interdependencies among clinically significant tasks. By jointly optimizing these objectives, the model enhances its robustness and clinical utility, improving prognostic accuracy and enabling more informed decision-making.

From a technical perspective, the deep learning model employs convolutional neural networks (CNNs) tailored to each data modality. For imaging data, sophisticated CNN architectures with residual connections enable hierarchical feature extraction, while genomic and proteomic data are processed via fully connected layers, integrated through fusion mechanisms within the network. Crucially, the system employs attention mechanisms that dynamically weigh contributions from each data source, ensuring that the most relevant features guide the tumor grading process. This multimodal fusion enhances interpretability, as clinicians can trace back which data aspects influence the model’s predictions.

The training dataset comprises an extensive cohort of NSCLC patients, encompassing thousands of digitized histopathology slides, CT images, and matched molecular data. The researchers implemented rigorous preprocessing pipelines to normalize and augment the data, addressing variability inherent in clinical datasets. Importantly, domain adaptation techniques were employed to mitigate batch effects and imaging protocol heterogeneity, ensuring the model’s generalizability across different clinical settings and patient populations.

Performance metrics reported in the study underscore the system’s superiority over conventional grading methods and previous AI models. With an area under the receiver operating characteristic curve (AUC-ROC) exceeding 0.90 for tumor grading tasks, the model exhibits remarkable sensitivity and specificity. Notably, survival prediction tasks achieved similarly high accuracy, enabling stratification of patients into risk categories with potential to guide treatment intensification or de-escalation.

In addition to diagnostic accuracy, the system’s design emphasizes clinical workflow integration. The researchers developed an interactive platform allowing pathologists and oncologists to visualize model outputs alongside raw data, fostering collaborative interpretation. This interface incorporates uncertainty quantification and confidence scores, aiding clinicians in evaluating prediction reliability and assimilating AI-driven insights with clinical judgment.

The researchers also addressed ethical and regulatory considerations critical for clinical deployment. Data privacy was preserved through federated learning techniques during model training, circumventing the need for centralized data storage and enabling multi-institutional collaboration without compromising patient confidentiality. Moreover, the model’s transparent architecture and interpretability features facilitate compliance with emerging AI governance frameworks and regulatory approvals.

Beyond the immediate clinical implications, this study exemplifies the transformative potential of deep learning for precision medicine. By capturing tumor heterogeneity through multimodal data fusion, it paves the way for dynamic tumor characterization that adapts to evolving cancer biology over time, potentially informing adaptive therapy paradigms. Furthermore, the model’s multitask structure could be extended to other malignancies or complex diseases characterized by multifactorial phenotypes.

The authors highlight several avenues for future research, including prospective clinical trials to assess real-world efficacy and impact on treatment outcomes. Plans to incorporate longitudinal patient data hold promise for modeling disease progression and therapy resistance. Additionally, incorporating emerging data modalities such as circulating tumor DNA and spatial transcriptomics could further refine the grading system’s sensitivity and specificity.

In summary, this landmark study ushers in a new era of AI-driven oncology, where deep learning models not only automate grading but also synthesize heterogeneous clinical data to empower personalized management in NSCLC. By bridging computational innovation and clinical applicability, it stands to significantly enhance diagnostic precision, treatment planning, and ultimately patient survival in a notoriously challenging cancer type.

As research continues to unravel the complexities of NSCLC, such sophisticated AI frameworks represent vital tools in the quest to deliver more effective, individualized care. The collaborative effort showcased by Liu, Dai, and colleagues sets a compelling precedent for harnessing multimodal data integration and multitask deep learning to tackle other pressing challenges in oncology and beyond.

---

Subject of Research: Multimodal multitask deep learning for tumor grading and management in non-small cell lung cancer

Article Title: Multimodal multitask deep learning for grading management system in non-small cell lung cancer

Article References: Liu, X., Dai, F., Dai, J. et al. Multimodal multitask deep learning for grading management system in non-small cell lung cancer. Nat Commun (2026). https://doi.org/10.1038/s41467-026-74474-2

Image Credits: AI Generated

In an unprecedented advancement for neurodegenerative disease research, a team led by Busquets, Li, Syed, and colleagues has unveiled iSCORE-PD, an innovative isogenic stem cell repository designed explicitly to accelerate Parkinson’s disease studies. Published in Nature Communications in 2026, this comprehensive collection represents a pivotal leap toward resolving the complex pathophysiology of Parkinson’s, a condition affecting millions worldwide yet still eluding curative therapies. The iSCORE-PD platform promises to redefine experimental paradigms by providing scientists with genetically matched cellular models that mirror the intricate genetic underpinnings of Parkinson’s disease.

Parkinson’s disease is characterized primarily by the progressive loss of dopaminergic neurons in the substantia nigra pars compacta, leading to debilitating motor symptoms such as tremors, rigidity, and bradykinesia, alongside a spectrum of non-motor complications. Despite extensive research, there remains a conspicuous gap between clinical observations and mechanistic understanding at the cellular and molecular levels. Existing animal models and conventional stem cell lines often fall short in recapitulating the exact human pathology. The iSCORE-PD initiative addresses this deficiency head-on by harnessing isogenic stem cell lines, which are genetically identical except for defined Parkinson’s-relevant mutations, thus enabling unprecedented control over genetic variables.

The technical foundation of iSCORE-PD relies on advanced genome editing methodologies, most notably CRISPR/Cas9 technology. By precisely introducing or correcting mutations in key genes implicated in Parkinson’s pathogenesis such as SNCA, LRRK2, PINK1, and PARKIN, researchers have generated isogenic pairs of induced pluripotent stem cells (iPSCs). These iPSC lines can subsequently be differentiated into dopaminergic neurons or other relevant cell types, facilitating side-by-side comparisons that isolate the effects of specific genetic alterations without the confounding noise of inter-individual genetic variability.

One of the remarkable features of the iSCORE-PD collection is its extensive genomic validation pipeline. Whole genome sequencing and transcriptomic analyses are systematically employed to ensure the integrity and authenticity of the edited lines, confirming that off-target effects are minimized and that the genetic landscape conforms precisely to intended edits. This meticulous validation is crucial for downstream experiments intending to dissect subtle cellular phenotypes attributable to Parkinson’s mutations with a high degree of confidence.

The availability of these isogenic cell lines dramatically enhances the fidelity of disease modeling. Scientists can now observe, in real time and within a human cellular context, the cascade of pathological events triggered by individual mutations. For instance, iSCORE-PD lines harboring the G2019S mutation in LRRK2 have already demonstrated distinct lysosomal dysfunction and mitochondrial impairment, hallmarks of Parkinson’s-related neurodegeneration. These insights provide vital clues about the molecular mechanisms that underpin neuronal vulnerability and death.

Beyond mechanistic studies, iSCORE-PD serves as an invaluable platform for drug discovery and therapeutic screening. High-throughput assays conducted on these standardized cellular models allow for robust evaluation of candidate compounds’ efficacy and toxicity. The isogenic background ensures that drug responses can be directly attributed to genetic contributors and not extraneous genomic differences, thereby accelerating the identification of promising therapeutic leads with increased translational potential.

Moreover, the iSCORE-PD repository supports integrative omics approaches. By coupling proteomics, metabolomics, and epigenomic profiling with the genetically defined cell lines, researchers can generate comprehensive molecular atlases elucidating the multifaceted alterations engendered by Parkinson’s-associated mutations. Such data-rich resources augment our understanding of disease heterogeneity and may reveal novel biomarkers for early diagnosis and progression monitoring.

The initiative also pioneers robust methodologies for modeling non-cell autonomous effects in Parkinson’s pathology. By co-culturing isogenic dopaminergic neurons with glial cells or other brain cell types derived from similarly engineered iPSC lines, investigators can explore the cell–cell interactions and inflammatory pathways that contribute to disease propagation. These complex in vitro systems increasingly resemble the in vivo brain microenvironment, bridging translational gaps.

In addition to its scientific hinge, iSCORE-PD represents a significant step forward in reproducibility and collaborative science. By providing open-access to a rigorously characterized and genetically standardized stem cell collection, the project promotes harmonization of experimental models across laboratories worldwide. This standardization mitigates discrepancies often observed in Parkinson’s research due to genetic heterogeneity of patient-derived cells and disparate culture conditions.

Ethical considerations have been front and center in the development and dissemination of the iSCORE-PD lines. The team has implemented stringent protocols to ensure donor anonymity, informed consent, and compliance with international guidelines for stem cell research. This ethical diligence is fundamental to fostering public trust and enabling broad scientific utilization with responsibility.

The iSCORE-PD research consortium envisions future expansion of the collection to encompass additional mutations, gene variants, and epigenetic modifications implicated in Parkinson’s disease, thereby creating an ever-more comprehensive toolkit for the field. This evolving repository will facilitate the study of gene–environment interactions and provide models for the sporadic forms of Parkinson’s that constitute the majority of cases.

Importantly, the development of iSCORE-PD aligns synergistically with advances in single-cell sequencing and live-cell imaging technologies. These combined approaches empower real-time monitoring of neuronal dynamics, synaptic connectivity, and intracellular trafficking in disease-relevant contexts. Such capabilities promise to unravel the temporal progression of neuronal dysfunction in unprecedented detail.

The translational impact of iSCORE-PD extends beyond fundamental research, holding profound implications for personalized medicine. Patient-specific isogenic lines can be derived, enabling evaluation of individualized drug responses and prediction of disease progression trajectories. This precision medicine paradigm could revolutionize clinical management and therapeutic development for Parkinson’s patients.

Finally, iSCORE-PD serves as a testament to the power of interdisciplinary collaboration, integrating expertise in stem cell biology, genome editing, neuroscience, bioinformatics, and clinical neurology. This concerted effort exemplifies how cutting-edge technologies, when harnessed collectively, can illuminate complex diseases and hasten the arrival of effective treatments.

In sum, the iSCORE-PD isogenic stem cell collection stands as a groundbreaking resource that transforms the landscape of Parkinson’s disease research. By providing genetically precise, highly validated cellular models, it enables unparalleled mechanistic insights, drug discovery opportunities, and personalized medicine applications. As the Parkinson’s field grapples with its most urgent challenges, iSCORE-PD promises to be a beacon guiding innovative interventions and ultimately improving patient outcomes worldwide.

---

Subject of Research: Parkinson’s disease; isogenic stem cell models; neurodegeneration; genome editing; induced pluripotent stem cells.

Article Title: iSCORE-PD: an isogenic stem cell collection to research Parkinson’s disease.

Article References:

Busquets, O., Li, H., Syed, K.M. et al. iSCORE-PD: an isogenic stem cell collection to research Parkinson’s disease.
Nat Commun (2026). https://doi.org/10.1038/s41467-026-74355-8

Image Credits: AI Generated

Across the expansive military installations of the United States, where substantial tracts of restricted-access natural lands persist, an evolving challenge has taken shape: how to effectively conserve endangered species amid overlapping and often conflicting recovery plans. These lands, which serve dual purposes of defense readiness and environmental stewardship, present a unique ecological puzzle. Amid this complexity, researchers at Virginia Tech’s College of Natural Resources and Environment have embarked on pioneering projects funded by two $2 million grants, aiming to redefine conservation strategies on military properties by emphasizing ecosystem-wide management rather than isolated species-specific interventions.

Historically, wildlife management on military bases has heavily relied on individual species’ recovery plans, a methodology that becomes problematic when multiple endangered species inhabit the same territory. Each species’ plan often prescribes distinct conservation actions, potentially leading to redundant efforts, increased management complexity, and sometimes conflicting directives. Recognizing these inefficiencies, the Virginia Tech research initiative seeks to develop integrated strategies that optimize land management actions to benefit diverse species assemblages simultaneously. This integrative approach has the potential not only to streamline operational practices on military lands but also to serve as a scalable model for ecosystems worldwide grappling with multifaceted conservation demands.

One cornerstone of this research is led by Assistant Professor Elizabeth Hunter, who operates within the U.S. Geological Survey’s Virginia Cooperative Fish and Wildlife Research Unit. Hunter’s project area in southeastern Georgia—including large military bases characterized by fire-managed ecosystems—examines the application of prescribed burning as a habitat management tool. Prescribed fire has traditionally been tailored around a handful of focal species, such as the red-cockaded woodpecker, leaving many other species with differing ecological needs inadequately addressed. Hunter’s work challenges this paradigm by employing cutting-edge monitoring technologies to survey a wide spectrum of taxa, from avian and mammalian species to amphibians, reptiles, insects, and crucial pollinators, aiming to capture a comprehensive biological response to various fire regimes.

By integrating extensive biodiversity data into a sophisticated multi-objective optimization framework, Hunter’s team can evaluate how different prescribed burning strategies affect numerous species simultaneously. This analytical tool allows for the assessment of trade-offs and synergies in conservation outcomes, addressing complex ecological interactions. The methodology asks crucial questions such as whether it is possible to adjust burn frequency, intensity, and spatial patterns to enhance habitats for multiple species concurrently. Such questions are vital for military land managers, who must balance ecosystem health with operational imperatives. Collaborations with organizations like the Orianne Society and the National Council for Air and Stream Improvement anchor this work in both academic and applied conservation contexts.

Parallel to Hunter’s fire ecology focus, Associate Professor Haldre Rogers directs research tackling the inefficiencies inherent in the current species-by-species recovery plan framework under the U.S. Endangered Species Act. Rogers’ project, focused on the Marianas Islands, explores whether ecosystem-wide habitat management can serve as an effective surrogate for the management of multiple individually listed species. Given that military lands often harbor dozens of endangered species with overlapping but separate recovery plans, the question arises: can habitat proxies be identified that allow for effective ecosystem-level conservation without the cumbersome requirement of separate interventions and monitoring for each species?

Rogers utilizes innovative ecosystem hypergraph modeling to encapsulate the intricate web of species interactions, habitat connectivity, and life-history traits, thereby identifying clusters of species that might respond similarly to shared management actions. This approach strives to uncover ecological congruencies and potential conflicts, aiming to inform management choices that maximize conservation benefits while minimizing inadvertent harm to co-occurring species. The project acknowledges inevitable ecological trade-offs but aspires to equip land managers with a landscape-scale understanding that promotes more efficient and informed decision-making.

The ecological importance of military installations in the United States cannot be overstated. Many contain well-preserved habitats that have been safeguarded from urban development, offering refugia for imperiled species. Activities such as prescribed burning and restricted access have inadvertently preserved biodiversity hotspots. At the same time, these installations must fulfill critical defense and training missions, underscoring the need for a delicate balance between operational readiness and environmental stewardship. The innovations proposed by Hunter, Rogers, and their teams promise a new frontier in reconciling these dual mandates through science-driven, scalable conservation models.

Their research partnership traverses multiple disciplines and institutions, spanning Virginia Tech’s Department of Fish and Wildlife Conservation and Statistics, collaborations with Rice University, Brown University, MIT, and contributions from postdoctoral scholars. This transdisciplinary approach enriches the projects with a fusion of ecological expertise, advanced computational modeling, and applied management perspectives. By leveraging emerging technologies and sophisticated analytical tools, these projects are poised to generate transformative insights relevant to conservation practitioners, policymakers, and military land managers alike.

In essence, these Virginia Tech-led studies embody a paradigm shift: transitioning from fragmented, species-specific management plans to holistic ecosystem-based strategies is both a necessity and an opportunity. This shift acknowledges the complex interdependencies among species and their habitats, calling for innovative methodologies that embrace ecological complexity rather than oversimplify it. For military lands, where the intersection of conservation and operational use is most pronounced, such research can catalyze both more effective biodiversity protection and enhanced land-use efficiency.

By integrating large-scale biological monitoring with computational optimization and ecosystem modeling, this research presents a robust framework for addressing the mounting challenges of endangered species management amid competing land uses. The implications extend far beyond military properties, offering a strategic blueprint adaptable to diverse ecosystems facing overlapping conservation mandates. The collective efforts spearheaded by Hunter and Rogers represent a significant leap forward in safeguarding the nation’s biological heritage through science-guided stewardship, setting the stage for broader applications in the global conservation community.

As endangered species faces continue to escalate globally, the insights arising from military installation ecosystems can inform best practices worldwide. The integration of multi-species data into management decisions, the recognition of ecosystem processes over isolated species needs, and the embracing of technological and analytical innovation underscore the future of conservation science. In a world of finite conservation resources and competing priorities, finding operationally feasible yet ecologically sound approaches is paramount.

These projects, supported by the Strategic Environmental Research and Development Program, exemplify such an approach, bridging theoretical science and practical land management. Their success will not only enhance the resilience of endangered species populations on military lands but also serve as a clarion call to redefine conservation paradigms universally. Ultimately, their work invites a transformative dialogue around how humanity navigates the complex stewardship of nature amid modern societal demands.

---

Subject of Research:
Endangered species management on military lands through ecosystem-based strategies and multi-species conservation optimization.

Article Title:
Rethinking Conservation: Integrated Ecosystem Management for Endangered Species on Military Installations

News Publication Date:
Not provided

Web References:

• Virginia Tech College of Natural Resources and Environment: https://cnre.vt.edu/
• Department of Fish and Wildlife Conservation: https://fishwild.vt.edu/
• Orianne Society: https://www.oriannesociety.org/
• National Council for Air and Stream Improvement: https://ncasi.org/

Image Credits:
Photo courtesy of Elizabeth Hunter.

Keywords:
Endangered species, Extinction, Wildlife management, Conservation biology, Ecosystem management, Prescribed fire, Habitat connectivity, Multi-species optimization, Military land conservation, Ecosystem hypergraphs.

Emerging insights into autism spectrum disorder (ASD) have been propelled forward by an innovative study examining cortical development dynamics using multiple mouse models. This work delves deeply into the transcriptional landscapes shaped by mutations implicated in ASD, revealing a complex interplay of stage-specific and genotype-specific molecular signatures. Employing cutting-edge network colocalization analyses and protein–protein interaction mapping, the study illuminates how biological processes converge and diverge during neurodevelopment in mutant genotypes.

Fundamentally, the research builds upon gene set enrichment analysis (GSEA), which previously highlighted overlapping pathways across various ASD-related mutations. The researchers expanded on this by applying network colocalization analysis to protein interaction networks, allowing functionally related proteins to be grouped into clusters through sophisticated averaging of random walks. This methodological advancement enables a nuanced assessment of expression changes aggregated at the systems-level rather than focusing solely on individual genes.

At the heart of their findings is the identification of distinct yet overlapping patterns of molecular dysregulation during critical developmental stages, particularly embryonic day 14.5 (E14.5), postnatal day 4 (P4), and postnatal day 14 (P14). At E14.5, the study reveals that mutations such as those in HnrnpU and Kdm6b robustly influence pathways tied to the actin cytoskeleton, cell proliferation, and division within excitatory neurons. Notably, these effects also extend, to a lesser degree, to Cul3 mutants, aligning with earlier evidence suggesting these genes’ involvement in neurodevelopmental regulation.

Delving deeper, the analysis highlights a provocative role of WNT signaling, especially pronounced in HnrnpU and Kdm6b variants. WNT pathways are critical for neural patterning and synaptogenesis, and their disruption could underpin some ASD phenotypes. This novel perspective accentuates the shared mechanistic underpinnings of ASD mutations converging on fundamental developmental pathways, stressing the importance of timing within neurodevelopmental trajectories.

Postnatally, at P4 and P14, the transcriptional alterations shift noticeably towards synaptic functions, ion transport, and metabolic processes—markers of neuronal specialization and circuit maturation. The data demonstrate how the influence of mutations evolves alongside neural differentiation and maturation. Intriguingly, Kmt5b mutants exhibited an inverse pattern at P4, suggesting either a compensatory or opposing regulatory role within the shared pathways affected across mutants. This counterintuitive finding raises the possibility of complex feedback or adaptive mechanisms operating in ASD-linked genetic variants.

By P14, metabolic processes prominently emerge in the affected molecular networks, especially in Bckdk, HnrnpU, and Trip12 mutants, indicating that energy metabolism perturbations may be integral to ASD pathogenesis. These results underscore the temporal dynamics of how genetic mutations impinge upon neurodevelopment, transitioning from early cytoskeletal and proliferative effects to later synaptic and metabolic modulation.

The study also innovatively explores how differentially expressed genes from each genotype localize within the interactome network and assesses similarity across developmental stages and mutations. This reveals a compelling developmental-stage-dependent pattern where transcriptional signatures cluster more closely within each stage but become more genetically distinct by P14. Such divergence suggests that while ASD mutations initially impact common biological processes, their influence fragments into distinct genotype-specific programs as development proceeds.

These findings have broad implications for understanding ASD heterogeneity and the challenges of developing universal therapeutic interventions. The discovery of both shared and unique pathway disruptions at different developmental times suggests treatment strategies may need to be tailored to both genotype and developmental timing to maximize efficacy. Moreover, the role of metabolic dysregulation invites further exploration into metabolic interventions or dietary modulation as adjunctive strategies.

Critically, the network colocalization analysis used represents a powerful approach to dissecting complex genetic disorders. By moving beyond simple gene expression differences and integrating system-level interactions, researchers can better capture the multifaceted nature of neurodevelopmental perturbations characteristic of ASD. This systems biology approach is likely to prove invaluable in future studies dissecting the molecular bases of brain disorders.

The comprehensive nature of this study, which integrates multiple ASD-relevant mouse models, enables the identification of convergent and divergent mechanisms with precision. It highlights how individual genes implicated in autism can distinctly alter transcriptional landscapes depending on context, highlighting the multifactorial dimension of ASD. Additionally, it underscores that the timing of gene dysregulation is paramount, with early embryonic disruptions cascading into diverse neurodevelopmental consequences.

In sum, this landmark investigation offers a rich, dynamic map of the molecular trajectories altered by ASD mutations across cortical development. Its integration of genomic, proteomic, and network analyses provides a blueprint for future research aimed at unraveling the complexity of neurodevelopmental disorders. By elucidating how multiple risk genes influence overlapping yet unique biological processes during critical developmental windows, the study charts a path toward precision medicine in autism spectrum disorder.

As the field progresses, leveraging such multi-dimensional analytical frameworks will be essential for translating genetic discoveries into mechanistic insights and targeted therapies. This work also highlights the ongoing need to examine neurodevelopment through temporal and molecular lenses simultaneously, enriching our understanding of how genetic diversity shapes brain development and function. Ultimately, investigations like these nurture hope for deciphering the intricate biology underlying ASD and devising effective interventions tailored to individual developmental trajectories.

Subject of Research: Cortical development dynamics and molecular mechanisms in autism spectrum disorder mouse models.

Article Title: Cortical development dynamics across autism spectrum disorder mouse models.

Article References:
Schwarz, L.A., Dotter, C.P., Isaev, S. et al. Cortical development dynamics across autism spectrum disorder mouse models. Nature (2026). https://doi.org/10.1038/s41586-026-10679-1

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41586-026-10679-1