RESUMO
Constructing agents with planning capabilities has long been one of the main challenges in the pursuit of artificial intelligence. Tree-based planning methods have enjoyed huge success in challenging domains, such as chess1 and Go2, where a perfect simulator is available. However, in real-world problems, the dynamics governing the environment are often complex and unknown. Here we present the MuZero algorithm, which, by combining a tree-based search with a learned model, achieves superhuman performance in a range of challenging and visually complex domains, without any knowledge of their underlying dynamics. The MuZero algorithm learns an iterable model that produces predictions relevant to planning: the action-selection policy, the value function and the reward. When evaluated on 57 different Atari games3-the canonical video game environment for testing artificial intelligence techniques, in which model-based planning approaches have historically struggled4-the MuZero algorithm achieved state-of-the-art performance. When evaluated on Go, chess and shogi-canonical environments for high-performance planning-the MuZero algorithm matched, without any knowledge of the game dynamics, the superhuman performance of the AlphaZero algorithm5 that was supplied with the rules of the game.
RESUMO
Although children and young adults are reportedly at lower risk for severe disease and death from infection with SARS-CoV-2, the virus that causes coronavirus disease 2019 (COVID-19), than are persons in other age groups (1), younger persons can experience infection and subsequently transmit infection to those at higher risk for severe illness (2-4). Although at lower risk for severe disease, some young adults experience serious illness, and asymptomatic or mild cases can result in sequelae such as myocardial inflammation (5). In the United States, approximately 45% of persons aged 18-22 years were enrolled in colleges and universities in 2019 (6). As these institutions reopen, opportunities for infection increase; therefore, mitigation efforts and monitoring reports of COVID-19 cases among young adults are important. During August 2-September 5, weekly incidence of COVID-19 among persons aged 18-22 years rose by 55.1% nationally; across U.S. Census regions,* increases were greatest in the Northeast, where incidence increased 144.0%, and Midwest, where incidence increased 123.4%. During the same period, changes in testing volume for SARS-CoV-2 in this age group ranged from a 6.2% decline in the West to a 170.6% increase in the Northeast. In addition, the proportion of cases in this age group among non-Hispanic White (White) persons increased from 33.8% to 77.3% during May 31-September 5. Mitigation and preventive measures targeted to young adults can likely reduce SARS-CoV-2 transmission among their contacts and communities. As colleges and universities resume operations, taking steps to prevent the spread of COVID-19 among young adults is critical (7).
Assuntos
Infecções por Coronavirus/diagnóstico , Infecções por Coronavirus/epidemiologia , Pneumonia Viral/diagnóstico , Pneumonia Viral/epidemiologia , Adolescente , Distribuição por Idade , COVID-19 , Teste para COVID-19 , Técnicas de Laboratório Clínico/estatística & dados numéricos , Humanos , Incidência , Pandemias , Estados Unidos/epidemiologia , Adulto JovemRESUMO
We introduce DeepNash, an autonomous agent that plays the imperfect information game Stratego at a human expert level. Stratego is one of the few iconic board games that artificial intelligence (AI) has not yet mastered. It is a game characterized by a twin challenge: It requires long-term strategic thinking as in chess, but it also requires dealing with imperfect information as in poker. The technique underpinning DeepNash uses a game-theoretic, model-free deep reinforcement learning method, without search, that learns to master Stratego through self-play from scratch. DeepNash beat existing state-of-the-art AI methods in Stratego and achieved a year-to-date (2022) and all-time top-three ranking on the Gravon games platform, competing with human expert players.
Assuntos
Inteligência Artificial , Reforço Psicológico , Jogos de Vídeo , HumanosRESUMO
BACKGROUND: An increasing number of diagnostic evaluations incorporate genetic testing to facilitate accurate and timely diagnoses. The increasing number and complexity of genetic tests continue to pose challenges in deciding when to test, selecting the correct test(s), and using results to inform medical diagnoses, especially for medical professionals lacking genetic expertise. Careful consideration of a diagnostic workflow can be helpful in understanding the appropriate uses of genetic testing within a broader diagnostic workup. CONTENT: The diagnosis of long QT syndrome (LQTS), a life-threatening cardiac arrhythmia, provides an example for this approach. Electrocardiography is the preferred means for diagnosing LQTS but can be uninformative for some patients due to the variable presentation of the condition. Family history and genetic testing can augment physiological testing to inform a diagnosis and subsequent therapy. Clinical and laboratory professionals informed by peer- reviewed literature and professional recommendations constructed a generalized LQTS diagnostic workflow. This workflow served to explore decisions regarding the use of genetic testing for diagnosing LQTS. SUMMARY AND OUTLOOK: Understanding the complexities and approaches to integrating genetic testing into a broader diagnostic evaluation is anticipated to support appropriate test utilization, optimize diagnostic evaluation, and facilitate a multidisciplinary approach essential for achieving accurate and timely diagnoses.
RESUMO
Modern genomic sequencing tests often interrogate large numbers of genes. Identification of appropriate reference materials for development, validation studies, and quality assurance of these tests poses a significant challenge for laboratories. It is difficult to develop and maintain expert knowledge to identify all variants that must be validated to ensure analytic and clinical validity. Additionally, it is usually not possible to procure appropriate and characterized genomic DNA reference materials containing the number and scope of variants required. To address these challenges, the Centers for Disease Control and Prevention's Genetic Testing Reference Material Program (GeT-RM) has partnered with the Clinical Genome Resource (ClinGen) to develop a publicly available list of expert curated, clinically important variants. ClinGen Variant Curation Expert Panels nominated 546 variants found in 84 disease-associated genes, including common pathogenic and difficult-to-detect variants. Variant types nominated included 346 single nucleotide variants, 104 deletions, 37 copy number variants, 25 duplications, 18 deletion-insertions, 5 inversions, 4 insertions, 2 complex rearrangements, 3 difficult-to-sequence regions, and 2 fusions. This expert-curated variant list is a resource that provides a foundation for designing comprehensive validation studies and for creating in silico reference materials for clinical genomic test development and validation.
Assuntos
Variações do Número de Cópias de DNA , Doença/genética , Rearranjo Gênico , Testes Genéticos/métodos , Genoma Humano , Sequenciamento de Nucleotídeos em Larga Escala/métodos , Mutação , Polimorfismo de Nucleotídeo Único , Simulação por Computador , DNA/genética , Bases de Dados Genéticas , Genômica/métodos , Humanos , Análise de Sequência de DNA/métodosRESUMO
A national workgroup convened by the Centers for Disease Control and Prevention identified principles and made recommendations for standardizing the description of sequence data contained within the variant file generated during the course of clinical next-generation sequence analysis for diagnosing human heritable conditions. The specifications for variant files were initially developed to be flexible with regard to content representation to support a variety of research applications. This flexibility permits variation with regard to how sequence findings are described and this depends, in part, on the conventions used. For clinical laboratory testing, this poses a problem because these differences can compromise the capability to compare sequence findings among laboratories to confirm results and to query databases to identify clinically relevant variants. To provide for a more consistent representation of sequence findings described within variant files, the workgroup made several recommendations that considered alignment to a common reference sequence, variant caller settings, use of genomic coordinates, and gene and variant naming conventions. These recommendations were considered with regard to the existing variant file specifications presently used in the clinical setting. Adoption of these recommendations is anticipated to reduce the potential for ambiguity in describing sequence findings and facilitate the sharing of genomic data among clinical laboratories and other entities.