Long W J
MIT Laboratory for Computer Science, Cambridge, MA 02139.
Proc Annu Symp Comput Appl Med Care. 1991:619-23.
This paper addresses the nature of the prior probabilities of diseases for probabilistic diagnostic reasoning. Because diseases differ in their chronicity, occurrence, reoccurrence, and likelihood of becoming part of the patient population, reasoning in terms of the frequency of disease episodes is necessary to capture the important distinctions. Even with these complexities, it is possible to formulate a reasonably accurate, computationally tractable, frequency estimation method for combinations of diseases. This method also suggests ways in which the needed numbers can be estimated from patient data.
本文探讨了概率诊断推理中疾病先验概率的本质。由于疾病在慢性程度、发生率、复发率以及成为患者群体一部分的可能性方面存在差异，因此有必要依据疾病发作频率进行推理，以把握这些重要差异。即便存在这些复杂性，仍有可能为疾病组合制定一种合理准确且计算上易于处理的频率估计方法。该方法还提出了从患者数据中估计所需数值的途径。
