3rd Department of Obstetrics and Gynaecology, University of Athens Medical School, Athens, Greece.
Med Sci Monit. 2010 Oct;16(10):RA231-6.
Current (and probably future) practice of medicine is mostly associated with prediction and accurate diagnosis. Especially in clinical practice, there is an increasing interest in constructing and using valid models of diagnosis and prediction. Artificial neural networks (ANNs) are mathematical systems being used as a prospective tool for reliable, flexible and quick assessment. They demonstrate high power in evaluating multifactorial data, assimilating information from multiple sources and detecting subtle and complex patterns. Their capability and difference from other statistical techniques lies in performing nonlinear statistical modelling. They represent a new alternative to logistic regression, which is the most commonly used method for developing predictive models for outcomes resulting from partitioning in medicine. In combination with the other non-algorithmic artificial intelligence techniques, they provide useful software engineering tools for the development of systems in quantitative medicine. Our paper first presents a brief introduction to ANNs, then, using what we consider the best available evidence through paradigms, we evaluate the ability of these networks to serve as first-line detection and prediction techniques in some of the most crucial fields in gynaecology. Finally, through the analysis of their current application, we explore their dynamics for future use.
目前（可能也是未来）的医学实践主要与预测和准确诊断相关。特别是在临床实践中，人们越来越有兴趣构建和使用有效的诊断和预测模型。人工神经网络（ANNs）是一种被用作可靠、灵活和快速评估的有前景工具的数学系统。它们在评估多因素数据、整合来自多个来源的信息以及检测细微和复杂模式方面具有强大的能力。它们的能力和与其他统计技术的区别在于进行非线性统计建模。它们为逻辑回归提供了一种新的替代方法，逻辑回归是医学中用于开发因分割而产生的结果的预测模型的最常用方法。与其他非算法人工智能技术相结合，它们为定量医学系统的开发提供了有用的软件工程工具。我们的论文首先简要介绍了人工神经网络，然后，通过范例使用我们认为是最佳的现有证据，评估了这些网络在妇科领域一些最关键领域作为一线检测和预测技术的能力。最后，通过对其当前应用的分析，探索了它们未来的应用动态。
