RESUMEN
Background and purpose: Biomarkers for prediction of outcome in patients with pancreatic cancer are wanted in order to personalize the treatment. This study investigated the value of longitudinal diffusion-weighted magnetic resonance imaging (DWI) for prediction of overall survival (OS) in patients with locally advanced pancreatic cancer (LAPC) treated with stereotactic body radiotherapy (SBRT). Materials and methods: The study included 45 patients with LAPC who received 5 fractions of 10 Gy on a 1.5T MRI-Linac. DWI was acquired prior to irradiation at each fraction. The analysis included baseline values and time-trends of the apparent diffusion coefficient (ADC) and DWI parameters obtained using a decomposition method. A multivariable Cox proportional hazards model for OS was made using best-subset selection, using cross-validation based on Bootstrap. Results: The median OS from the first day of SBRT was 15.5 months (95% CI: 13.2-20.6), and the median potential follow-up time was 19.8 months. The best-performing multivariable model for OS included two decomposition-based DWI parameters: one baseline and one time-trend parameter. The C-Harrell index describing the model's discriminating power was 0.754. High baseline ADC values were associated with reduced OS, whereas no association between the ADC time-trend and OS was observed. Conclusion: Decomposition-based DWI parameters indicated value in the prediction of OS in LAPC. A DWI time-trend parameter was included in the best-performing model, indicating a potential benefit of acquiring longitudinal DWI during the SBRT course. These findings support both baseline and longitudinal DWI as candidate prognostic biomarkers, which may become tools for personalization of the treatment of patients with LAPC.
RESUMEN
In drug development, assessing the toxicity of candidate compounds is crucial for successfully transitioning from preclinical research to early-stage clinical trials. Drug safety is typically assessed using animal models with a manual histopathological examination of tissue sections to characterize the dose-response relationship of the compound - a time-intensive process prone to inter-observer variability and predominantly involving tedious review of cases without abnormalities. Artificial intelligence (AI) methods in pathology hold promise to accelerate this assessment and enhance reproducibility and objectivity. Here, we introduce TRACE, a model designed for toxicologic liver histopathology assessment capable of tackling a range of diagnostic tasks across multiple scales, including situations where labeled data is limited. TRACE was trained on 15 million histopathology images extracted from 46,734 digitized tissue sections from 157 preclinical studies conducted on Rattus norvegicus. We show that TRACE can perform various downstream toxicology tasks spanning histopathological response assessment, lesion severity scoring, morphological retrieval, and automatic dose-response characterization. In an independent reader study, TRACE was evaluated alongside ten board-certified veterinary pathologists and achieved higher concordance with the consensus opinion than the average of the pathologists. Our study represents a substantial leap over existing computational models in toxicology by offering the first framework for accelerating and automating toxicological pathology assessment, promoting significant progress with faster, more consistent, and reliable diagnostic processes.
RESUMEN
Early identification of drug toxicity is essential yet challenging in drug development. At the preclinical stage, toxicity is assessed with histopathological examination of tissue sections from animal models to detect morphological lesions. To complement this analysis, toxicogenomics is increasingly employed to understand the mechanism of action of the compound and ultimately identify lesion-specific safety biomarkers for which in vitro assays can be designed. However, existing works that aim to identify morphological correlates of expression changes rely on qualitative or semi-quantitative morphological characterization and remain limited in scale or morphological diversity. Artificial intelligence (AI) offers a promising approach for quantitatively modeling this relationship at an unprecedented scale. Here, we introduce GEESE, an AI model designed to impute morphomolecular signatures in toxicology data. Our model was trained to predict 1,536 gene targets on a cohort of 8,231 hematoxylin and eosin-stained liver sections from Rattus norvegicus across 127 preclinical toxicity studies. The model, evaluated on 2,002 tissue sections from 29 held-out studies, can yield pseudo-spatially resolved gene expression maps, which we correlate with six key drug-induced liver injuries (DILI). From the resulting 25 million lesion-expression pairs, we established quantitative relations between up and downregulated genes and lesions. Validation of these signatures against toxicogenomic databases, pathway enrichment analyses, and human hepatocyte cell lines asserted their relevance. Overall, our study introduces new methods for characterizing toxicity at an unprecedented scale and granularity, paving the way for AI-driven discovery of toxicity biomarkers.
RESUMEN
Incorporating selenium into high-surface-area carbon with hierarchical pores, derived from red kidney bean peels via simple carbonization/activation, yields a superior Li-Se battery cathode material. This method produces a carbon framework with 568 m2 g-1 surface area, significant pore volume, and improves the composite's electronic conductivity and stability by mitigating volume changes and reducing lithium polyselenide dissolution. The Se@ACRKB composite, containing 45 wt% selenium, shows high discharge capacities (609.13 mAh g-1 on the 2nd cycle, maintaining 470.76 mAh g-1 after 400 cycles at 0.2 C, and 387.58 mAh g-1 over 1000 cycles at 1 C). This demonstrates exceptional long-term stability and performance, also applicable to Na-Se batteries, with 421.36 mAh g-1 capacity after 200 cycles at 0.1 C. Our study showcases the potential of using sustainable materials for advanced battery technologies, emphasizing cost-effective and scalable solutions for energy storage.
RESUMEN
The field of computational pathology[1,2] has witnessed remarkable progress in the development of both task-specific predictive models and task-agnostic self-supervised vision encoders[3,4]. However, despite the explosive growth of generative artificial intelligence (AI), there has been limited study on building general purpose, multimodal AI assistants and copilots[5] tailored to pathology. Here we present PathChat, a vision-language generalist AI assistant for human pathology. We build PathChat by adapting a foundational vision encoder for pathology, combining it with a pretrained large language model and finetuning the whole system on over 456,000 diverse visual language instructions consisting of 999,202 question-answer turns. We compare PathChat against several multimodal vision language AI assistants and GPT4V, which powers the commercially available multimodal general purpose AI assistant ChatGPT-4[7]. PathChat achieved state-of-the-art performance on multiple-choice diagnostic questions from cases of diverse tissue origins and disease models. Furthermore, using open-ended questions and human expert evaluation, we found that overall PathChat produced more accurate and pathologist-preferable responses to diverse queries related to pathology. As an interactive and general vision-language AI Copilot that can flexibly handle both visual and natural language inputs, PathChat can potentially find impactful applications in pathology education, research, and human-in-the-loop clinical decision making.
RESUMEN
Experts discuss the challenges and opportunities of using artificial intelligence (AI) to study the evolution of cancer cells and their microenvironment, improve diagnosis, predict treatment response, and ensure responsible implementation in the clinic.
Asunto(s)
Inteligencia Artificial , Neoplasias , Microambiente Tumoral , Humanos , Neoplasias/terapia , Neoplasias/genética , Neoplasias/patologíaRESUMEN
Cytoplasmic mislocalization and aggregation of TDP-43 protein are hallmarks of amyotrophic lateral sclerosis (ALS) and are observed in the vast majority of both familial and sporadic cases. How these two interconnected processes are regulated on a molecular level, however, remains enigmatic. Genome-wide screens for modifiers of the ALS-associated genes TDP-43 and FUS have identified the phospholipase D (Pld) pathway as a key regulator of ALS-related phenotypes in the fruit fly Drosophila melanogaster [M. W. Kankel et al., Genetics 215, 747-766 (2020)]. Here, we report the results of our search for downstream targets of the enzymatic product of Pld, phosphatidic acid. We identify two conserved negative regulators of the cAMP/PKA signaling pathway, the phosphodiesterase dunce and the inhibitory subunit PKA-R2, as modifiers of pathogenic phenotypes resulting from overexpression of the Drosophila TDP-43 ortholog TBPH. We show that knockdown of either of these genes results in a mitigation of both TBPH aggregation and mislocalization in larval motor neuron cell bodies, as well as an amelioration of adult-onset motor defects and shortened lifespan induced by TBPH. We determine that PKA kinase activity is downstream of both TBPH and Pld and that overexpression of the PKA target CrebA can rescue TBPH mislocalization. These findings suggest a model whereby increasing cAMP/PKA signaling can ameliorate the molecular and functional effects of pathological TDP-43.
Asunto(s)
Proteínas Quinasas Dependientes de AMP Cíclico , AMP Cíclico , Proteínas de Unión al ADN , Proteínas de Drosophila , Drosophila melanogaster , Transducción de Señal , Animales , AMP Cíclico/metabolismo , Drosophila melanogaster/metabolismo , Proteínas Quinasas Dependientes de AMP Cíclico/metabolismo , Proteínas Quinasas Dependientes de AMP Cíclico/genética , Proteínas de Drosophila/metabolismo , Proteínas de Drosophila/genética , Proteínas de Unión al ADN/metabolismo , Proteínas de Unión al ADN/genética , Esclerosis Amiotrófica Lateral/metabolismo , Esclerosis Amiotrófica Lateral/genética , Humanos , Neuronas Motoras/metabolismoRESUMEN
Human tissue, which is inherently three-dimensional (3D), is traditionally examined through standard-of-care histopathology as limited two-dimensional (2D) cross-sections that can insufficiently represent the tissue due to sampling bias. To holistically characterize histomorphology, 3D imaging modalities have been developed, but clinical translation is hampered by complex manual evaluation and lack of computational platforms to distill clinical insights from large, high-resolution datasets. We present TriPath, a deep-learning platform for processing tissue volumes and efficiently predicting clinical outcomes based on 3D morphological features. Recurrence risk-stratification models were trained on prostate cancer specimens imaged with open-top light-sheet microscopy or microcomputed tomography. By comprehensively capturing 3D morphologies, 3D volume-based prognostication achieves superior performance to traditional 2D slice-based approaches, including clinical/histopathological baselines from six certified genitourinary pathologists. Incorporating greater tissue volume improves prognostic performance and mitigates risk prediction variability from sampling bias, further emphasizing the value of capturing larger extents of heterogeneous morphology.
Asunto(s)
Imagenología Tridimensional , Neoplasias de la Próstata , Aprendizaje Automático Supervisado , Humanos , Masculino , Aprendizaje Profundo , Imagenología Tridimensional/métodos , Pronóstico , Neoplasias de la Próstata/patología , Neoplasias de la Próstata/diagnóstico por imagen , Microtomografía por Rayos X/métodosRESUMEN
PURPOSE: Stereotactic body radiotherapy (SBRT) has emerged as a promising new modality for locally advanced pancreatic cancer (LAPC). The current study evaluated the efficacy and toxicity of SBRT in patients with LAPC (NCT03648632). METHODS: This prospective single institution phase II study recruited patients with histologically or cytologically proven adenocarcinoma of the pancreas after more than two months of combination chemotherapy with no sign of progressive disease. Patients were prescribed 50-60 Gy in 5-8 fractions. Patients were initially treated on a standard linac (n = 4). Since 2019, patients were treated using online magnetic resonance (MR) image-guidance on a 1.5 T MRI-linac, where the treatment plan was adapted to the anatomy of the day. The primary endpoint was resection rate. RESULTS: Twenty-eight patients were enrolled between August 2018 and March 2022. All patients had non-resectable disease at time of diagnosis. Median follow-up from inclusion was 28.3 months (95 % CI 24.0-NR). Median progression-free and overall survival from inclusion were 7.8 months (95 % CI 5.0-14.8) and 16.5 months (95 % CI 10.7-22.6), respectively. Six patients experienced grade III treatment-related adverse events (jaundice, nausea, vomiting and/or constipation). One of the initial four patients receiving treatment on a standard linac experienced a grade IV perforation of the duodenum. Six patients (21 %) underwent resection. A further one patient was offered resection but declined. CONCLUSION: This study demonstrates that SBRT in patients with LAPC was associated with promising overall survival and resection rates. Furthermore, SBRT was safe and well tolerated, with limited severe toxicities.
Asunto(s)
Neoplasias Pancreáticas , Radiocirugia , Humanos , Radiocirugia/métodos , Radiocirugia/efectos adversos , Neoplasias Pancreáticas/radioterapia , Neoplasias Pancreáticas/patología , Neoplasias Pancreáticas/diagnóstico por imagen , Masculino , Femenino , Anciano , Persona de Mediana Edad , Estudios Prospectivos , Anciano de 80 o más Años , Radioterapia Guiada por Imagen/métodos , Adenocarcinoma/patología , Adenocarcinoma/radioterapia , Adenocarcinoma/cirugía , Adenocarcinoma/diagnóstico por imagen , Imagen por Resonancia Magnética/métodosRESUMEN
Artificial intelligence (AI) has been commoditized. It has evolved from a specialty resource to a readily accessible tool for cancer researchers. AI-based tools can boost research productivity in daily workflows, but can also extract hidden information from existing data, thereby enabling new scientific discoveries. Building a basic literacy in these tools is useful for every cancer researcher. Researchers with a traditional biological science focus can use AI-based tools through off-the-shelf software, whereas those who are more computationally inclined can develop their own AI-based software pipelines. In this article, we provide a practical guide for non-computational cancer researchers to understand how AI-based tools can benefit them. We convey general principles of AI for applications in image analysis, natural language processing and drug discovery. In addition, we give examples of how non-computational researchers can get started on the journey to productively use AI in their own work.
Asunto(s)
Inteligencia Artificial , Neoplasias , Humanos , Descubrimiento de Drogas/métodos , Programas Informáticos , Investigadores , Procesamiento de Lenguaje Natural , Procesamiento de Imagen Asistido por Computador/métodos , Investigación Biomédica/métodosRESUMEN
Despite increasing numbers of regulatory approvals, deep learning-based computational pathology systems often overlook the impact of demographic factors on performance, potentially leading to biases. This concern is all the more important as computational pathology has leveraged large public datasets that underrepresent certain demographic groups. Using publicly available data from The Cancer Genome Atlas and the EBRAINS brain tumor atlas, as well as internal patient data, we show that whole-slide image classification models display marked performance disparities across different demographic groups when used to subtype breast and lung carcinomas and to predict IDH1 mutations in gliomas. For example, when using common modeling approaches, we observed performance gaps (in area under the receiver operating characteristic curve) between white and Black patients of 3.0% for breast cancer subtyping, 10.9% for lung cancer subtyping and 16.0% for IDH1 mutation prediction in gliomas. We found that richer feature representations obtained from self-supervised vision foundation models reduce performance variations between groups. These representations provide improvements upon weaker models even when those weaker models are combined with state-of-the-art bias mitigation strategies and modeling choices. Nevertheless, self-supervised vision foundation models do not fully eliminate these discrepancies, highlighting the continuing need for bias mitigation efforts in computational pathology. Finally, we demonstrate that our results extend to other demographic factors beyond patient race. Given these findings, we encourage regulatory and policy agencies to integrate demographic-stratified evaluation into their assessment guidelines.
Asunto(s)
Glioma , Neoplasias Pulmonares , Humanos , Sesgo , Negro o Afroamericano , Población Negra , Demografía , Errores Diagnósticos , Glioma/diagnóstico , Glioma/genética , BlancoRESUMEN
Silicon (Si) has emerged as a potent anode material for lithium-ion batteries (LIBs), but faces challenges like low electrical conductivity and significant volume changes during lithiation/delithiation, leading to material pulverization and capacity degradation. Recent research on nanostructured Si aims to mitigate volume expansion and enhance electrochemical performance, yet still grapples with issues like pulverization, unstable solid electrolyte interface (SEI) growth, and interparticle resistance. This review delves into innovative strategies for optimizing Si anodes' electrochemical performance via structural engineering, focusing on the synthesis of Si/C composites, engineering multidimensional nanostructures, and applying non-carbonaceous coatings. Forming a stable SEI is vital to prevent electrolyte decomposition and enhance Li+ transport, thereby stabilizing the Si anode interface and boosting cycling Coulombic efficiency. We also examine groundbreaking advancements such as self-healing polymers and advanced prelithiation methods to improve initial Coulombic efficiency and combat capacity loss. Our review uniquely provides a detailed examination of these strategies in real-world applications, moving beyond theoretical discussions. It offers a critical analysis of these approaches in terms of performance enhancement, scalability, and commercial feasibility. In conclusion, this review presents a comprehensive view and a forward-looking perspective on designing robust, high-performance Si-based anodes the next generation of LIBs.
RESUMEN
This study examines the challenges Pakistani farmers face in adopting global good agricultural practices (GGAP) and highlights the limitations in infrastructure and cost-based clauses. A questionnaire based on GGAP's fruit and vegetable module version 5.0 was developed and validated by the Department of Environmental Sciences, Government College University, Faisalabad. This was a survey-based study of 15 farmers divided into 5 groups according to their annual farm turnover. The findings of the study indicated that, although the basic paperwork requirements of GGAP were implementable, clauses related to capital investment and technical record-keeping were not. Results showed that 90-100% of farmers considered risk assessments, training, and documentation on their farms. However, 42-56% of clauses related to record-keeping, installation, visual presentation, and infrastructure development, and 24-37% of clauses related to external testing, health, safety, and hygiene were declared not implementable. The study revealed a need for adapting GGAP standards to Pakistan's unique agricultural conditions, suggesting the development of localized standards for more practical implementation. The study's findings highlight crucial insights for policymakers and stakeholders in the agriculture sector and suggest the need for target strategies to overcome implementation barriers and optimize the adaptation of Global GAP in Pakistan that would help to increase exports of agricultural commodities.
RESUMEN
The accelerated adoption of digital pathology and advances in deep learning have enabled the development of robust models for various pathology tasks across a diverse array of diseases and patient cohorts. However, model training is often difficult due to label scarcity in the medical domain, and a model's usage is limited by the specific task and disease for which it is trained. Additionally, most models in histopathology leverage only image data, a stark contrast to how humans teach each other and reason about histopathologic entities. We introduce CONtrastive learning from Captions for Histopathology (CONCH), a visual-language foundation model developed using diverse sources of histopathology images, biomedical text and, notably, over 1.17 million image-caption pairs through task-agnostic pretraining. Evaluated on a suite of 14 diverse benchmarks, CONCH can be transferred to a wide range of downstream tasks involving histopathology images and/or text, achieving state-of-the-art performance on histology image classification, segmentation, captioning, and text-to-image and image-to-text retrieval. CONCH represents a substantial leap over concurrent visual-language pretrained systems for histopathology, with the potential to directly facilitate a wide array of machine learning-based workflows requiring minimal or no further supervised fine-tuning.
Asunto(s)
Lenguaje , Aprendizaje Automático , Humanos , Flujo de TrabajoRESUMEN
Quantitative evaluation of tissue images is crucial for computational pathology (CPath) tasks, requiring the objective characterization of histopathological entities from whole-slide images (WSIs). The high resolution of WSIs and the variability of morphological features present significant challenges, complicating the large-scale annotation of data for high-performance applications. To address this challenge, current efforts have proposed the use of pretrained image encoders through transfer learning from natural image datasets or self-supervised learning on publicly available histopathology datasets, but have not been extensively developed and evaluated across diverse tissue types at scale. We introduce UNI, a general-purpose self-supervised model for pathology, pretrained using more than 100 million images from over 100,000 diagnostic H&E-stained WSIs (>77 TB of data) across 20 major tissue types. The model was evaluated on 34 representative CPath tasks of varying diagnostic difficulty. In addition to outperforming previous state-of-the-art models, we demonstrate new modeling capabilities in CPath such as resolution-agnostic tissue classification, slide classification using few-shot class prototypes, and disease subtyping generalization in classifying up to 108 cancer types in the OncoTree classification system. UNI advances unsupervised representation learning at scale in CPath in terms of both pretraining data and downstream evaluation, enabling data-efficient artificial intelligence models that can generalize and transfer to a wide range of diagnostically challenging tasks and clinical workflows in anatomic pathology.
Asunto(s)
Inteligencia Artificial , Flujo de TrabajoRESUMEN
Classic Hodgkin Lymphoma (cHL) is a tumor composed of rare malignant Hodgkin and Reed-Sternberg (HRS) cells nested within a T-cell rich inflammatory immune infiltrate. cHL is associated with Epstein-Barr Virus (EBV) in 25% of cases. The specific contributions of EBV to the pathogenesis of cHL remain largely unknown, in part due to technical barriers in dissecting the tumor microenvironment (TME) in high detail. Herein, we applied multiplexed ion beam imaging (MIBI) spatial pro-teomics on 6 EBV-positive and 14 EBV-negative cHL samples. We identify key TME features that distinguish between EBV-positive and EBV-negative cHL, including the relative predominance of memory CD8 T cells and increased T-cell dysfunction as a function of spatial proximity to HRS cells. Building upon a larger multi-institutional cohort of 22 EBV-positive and 24 EBV-negative cHL samples, we orthogonally validated our findings through a spatial multi-omics approach, coupling whole transcriptome capture with antibody-defined cell types for tu-mor and T-cell populations within the cHL TME. We delineate contrasting transcriptomic immunological signatures between EBV-positive and EBV-negative cases that differently impact HRS cell proliferation, tumor-immune interactions, and mecha-nisms of T-cell dysregulation and dysfunction. Our multi-modal framework enabled a comprehensive dissection of EBV-linked reorganization and immune evasion within the cHL TME, and highlighted the need to elucidate the cellular and molecular fac-tors of virus-associated tumors, with potential for targeted therapeutic strategies.
RESUMEN
BACKGROUND: Central line-associated bloodstream infections (CLABSIs) pose a significant risk to critically ill patients, particularly in intensive care units (ICU), and are a significant cause of hospital-acquired infections. We investigated whether implementation of a multifaceted intervention was associated with reduced incidence of CLABSIs. METHODS: This was a prospective cohort study over nine years. We implemented a bundled intervention approach to prevent CLABSIs, consisting of a comprehensive unit-based safety program (CUSP). The program was implemented in the Neonatal ICU, Medical ICU, and Surgical ICU departments at the Aga Khan University Hospital in Pakistan. RESULTS: The three intervention ICUs combined were associated with an overall 36% reduction in CLABSI rates and a sustained reduction in CLABSI rates for > a year (5 quarters). The Neonatal ICU experienced a decrease of 77% in CLABSI rates lasting â¼1 year (4 quarters). An attendance rate above 88% across all stakeholder groups in each CUSP meeting correlated with a better and more sustained infection reduction. CONCLUSIONS: Our multifaceted approach using the CUSP model was associated with reduced CLABSI-associated morbidity and mortality in resource-limited settings. Our findings suggest that a higher attendance rate (>85%) at meetings may be necessary to achieve sustained effects post-intervention.
Asunto(s)
Infecciones Relacionadas con Catéteres , Control de Infecciones , Unidades de Cuidados Intensivos , Humanos , Infecciones Relacionadas con Catéteres/prevención & control , Infecciones Relacionadas con Catéteres/epidemiología , Estudios Prospectivos , Pakistán/epidemiología , Control de Infecciones/métodos , Cateterismo Venoso Central/efectos adversos , Infección Hospitalaria/prevención & control , Infección Hospitalaria/epidemiología , Incidencia , Países en Desarrollo , Bacteriemia/prevención & control , Bacteriemia/epidemiología , Sepsis/prevención & control , Sepsis/epidemiologíaRESUMEN
OBJECTIVES: To evaluate the effect of the presence of a physician in the triage area on the number of patients who leave without being seen (LWBS) and some of the factors affecting emergency department (ED) crowding. METHODS: This was a pre-post study carried out at King Fahad Specialist Hospital, Dammam, Saudi Arabia. The 3-month study, consisting of 7826 patients, was split into pre-physician and post-physician periods. Variables compared across these periods were the number of LWBS patients, length of hospital stay, time to physician, and time to disposition decision. Statistical analysis was carried out using R version 4.3.0. RESULTS: Our results showed that the presence of a triage physician significantly decreased the number of LWBS patients (p<0.001) and the time taken to encounter an ED physician (p<0.001). However, it did not have any significant impact on the length of hospital stay (p=0.5) or time to disposition decision (p=0.9). CONCLUSION: The appointment of a triage physician has streamlined patient flow and decreased LWBS rates in the ED, demonstrating the need for more thorough research in this area.
Asunto(s)
Médicos , Mejoramiento de la Calidad , Humanos , Triaje/métodos , Factores de Tiempo , Servicio de Urgencia en Hospital , Hospitales Especializados , Tiempo de Internación , Aglomeración , Estudios RetrospectivosRESUMEN
Highly multiplexed protein imaging is emerging as a potent technique for analyzing protein distribution within cells and tissues in their native context. However, existing cell annotation methods utilizing high-plex spatial proteomics data are resource intensive and necessitate iterative expert input, thereby constraining their scalability and practicality for extensive datasets. We introduce MAPS (Machine learning for Analysis of Proteomics in Spatial biology), a machine learning approach facilitating rapid and precise cell type identification with human-level accuracy from spatial proteomics data. Validated on multiple in-house and publicly available MIBI and CODEX datasets, MAPS outperforms current annotation techniques in terms of speed and accuracy, achieving pathologist-level precision even for typically challenging cell types, including tumor cells of immune origin. By democratizing rapidly deployable and scalable machine learning annotation, MAPS holds significant potential to expedite advances in tissue biology and disease comprehension.
Asunto(s)
Aprendizaje Automático , Patólogos , Humanos , Diagnóstico por Imagen , Proteómica/métodosRESUMEN
Large-scale genomic data are well suited to analysis by deep learning algorithms. However, for many genomic datasets, labels are at the level of the sample rather than for individual genomic measures. Machine learning models leveraging these datasets generate predictions by using statically encoded measures that are then aggregated at the sample level. Here we show that a single weakly supervised end-to-end multiple-instance-learning model with multi-headed attention can be trained to encode and aggregate the local sequence context or genomic position of somatic mutations, hence allowing for the modelling of the importance of individual measures for sample-level classification and thus providing enhanced explainability. The model solves synthetic tasks that conventional models fail at, and achieves best-in-class performance for the classification of tumour type and for predicting microsatellite status. By improving the performance of tasks that require aggregate information from genomic datasets, multiple-instance deep learning may generate biological insight.