Assuntos
Neoplasias Ósseas/diagnóstico , Neoplasias Ósseas/cirurgia , Condrossarcoma/diagnóstico , Septo Nasal , Neoplasias Nasais/diagnóstico , Adolescente , Biópsia por Agulha , Condrossarcoma/cirurgia , Feminino , Seguimentos , Humanos , Imageamento por Ressonância Magnética , Neoplasias Nasais/cirurgia , Tomografia Computadorizada por Raios X , Resultado do TratamentoRESUMO
Because psychological assessment typically lacks biological gold standards, it traditionally has relied on clinicians' expert knowledge. A more empirically based approach frequently has applied linear models to data to derive meaningful constructs and appropriate measures. Statistical inferences are then used to assess the generality of the findings. This article introduces artificial neural networks (ANNs), flexible nonlinear modeling techniques that test a model's generality by applying its estimates against "future" data. ANNs have potential for overcoming some shortcomings of linear models. The basics of ANNs and their applications to psychological assessment are reviewed. Two examples of clinical decision making are described in which an ANN is compared with linear models, and the complexity of the network performance is examined. Issues salient to psychological assessment are addressed.
Assuntos
Tomada de Decisões Assistida por Computador , Diagnóstico por Computador/estatística & dados numéricos , Redes Neurais de Computação , Determinação da Personalidade/estatística & dados numéricos , Adolescente , Adulto , Idoso , Transtorno da Personalidade Antissocial/diagnóstico , Criança , Humanos , Pessoa de Meia-Idade , Prognóstico , Psicometria , Reprodutibilidade dos Testes , Transtornos Relacionados ao Uso de Substâncias/diagnóstico , Transtornos Relacionados ao Uso de Substâncias/mortalidade , Veteranos/psicologia , Veteranos/estatística & dados numéricosRESUMO
We model functions that use genetic information as input and trait information as output to understand genetic linkage in complex diseases. Using simulated data from GAW11, we have applied categorical classification methods and neural network analysis. We use sharing at selected markers as input, and the classification of the sib pair (for example, affected-affected or affected-unaffected) as output. In addition, our methods include environmental risk factors as predictors of phenotype. Categorical and neural network methods each led to results consistent with findings from other methods such as the logistic regression method of Rice et al. [this issue]. Post-analysis comparison with the GAW11 answers showed that these methods are capable of detecting correct signals in a single replicate. One advantage of our methods is that they allow analysis of the entire genome at once, so that interactions among multiple trait-influencing loci may be detected. Furthermore, these methods can use a variety of sib pairs rather than affected pairs only.