Suppr超能文献

[用于眼压疾病诊断的专家系统]

[An expert system for the diagnosis of intraocular pressure disorders].

作者信息

Paycha F, Paycha A

出版信息

J Fr Ophtalmol. 1995;18(11):656-66.

PMID:8745763
Abstract

PURPOSE

This expert system (ES) is designed to avoid the four types of diagnosis errors: by omission or by confusion both of which may occur when collecting signs and symptoms and when analyzed data. This ES uses a thesaurus containing 42 classical and objective signs and a comprehensive library. For a given case report, it provides the practitioner with a positive diagnosis, and if possible, one or several etiologies taking into account all the data known about intraocular disorders.

METHODS

This ES is run on a PC-compatible computer with 640 KB RAM and without hard drive. The 90 KB program is written in compiled Basic. After checking the case record to rule out nonsenses, the logic considers every sign of the case record as equivalent. Likewise, in the first stage all syndromes are considered of equal likelihood. The syndromic diagnosis is set by identification. Comparisons of prevalence interfere, when needed, in a second phase, within the framework of etiological diagnosis.

RESULTS

When tested on a set of 180 cases records combined with 115 cases extracted from the literature corresponding to more than 1,900 diaagnoses, this ES yielded 100% correct positive and differential diagnoses, thus validating the logial hypotheses. Probability of error by omission was less than a 3% and probability of error by confusion was well below 0.1%. Besides, the logical analysis shelds light on the intellectual mechanisms of diagnosis.

CONCLUSIONS

This ES is user-friendly, easy to update and can answer clinical problems the ophthalmologist has to cope with in diagnosis and therapy of intraocular pressure disorders.

摘要

目的

本专家系统旨在避免四种诊断错误:漏诊和误诊,这两种错误可能在收集体征和症状以及分析数据时出现。该专家系统使用一个包含42种经典客观体征的词库和一个综合数据库。对于给定的病例报告,它会为从业者提供肯定诊断,如有可能,还会给出一种或几种病因,同时考虑到所有已知的眼内疾病数据。

方法

本专家系统在一台具有640 KB随机存取存储器且无硬盘驱动器的兼容个人电脑上运行。90 KB的程序用编译Basic语言编写。在检查病例记录以排除无意义信息后,逻辑将病例记录中的每个体征视为同等重要。同样,在第一阶段,所有综合征被认为具有同等可能性。通过识别来确定综合征诊断。在病因诊断框架内,必要时在第二阶段进行患病率比较。

结果

在一组180例病例记录以及从文献中提取的115例病例(对应超过1900次诊断)上进行测试时,该专家系统得出了100%正确的肯定诊断和鉴别诊断,从而验证了逻辑假设。漏诊错误概率小于3%,误诊错误概率远低于0.1%。此外,逻辑分析揭示了诊断的智力机制。

结论

本专家系统用户友好,易于更新,能够回答眼科医生在眼内压疾病诊断和治疗中必须应对的临床问题。

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验