This website requires JavaScript.

DxFormer: A Decoupled Automatic Diagnostic System Based on Decoder-Encoder Transformer with Dense Symptom Representations

Wei ChenCheng ZhongJiajie PengZhongyu Wei
May 2022
摘要
Diagnosis-oriented dialogue system queries the patient's health condition andmakes predictions about possible diseases through continuous interaction withthe patient. A few studies use reinforcement learning (RL) to learn the optimalpolicy from the joint action space of symptoms and diseases. However, existingRL (or Non-RL) methods cannot achieve sufficiently good prediction accuracy,still far from its upper limit. To address the problem, we propose a decoupledautomatic diagnostic framework DxFormer, which divides the diagnosis processinto two steps: symptom inquiry and disease diagnosis, where the transitionfrom symptom inquiry to disease diagnosis is explicitly determined by thestopping criteria. In DxFormer, we treat each symptom as a token, and formalizethe symptom inquiry and disease diagnosis to a language generation model and asequence classification model respectively. We use the inverted version ofTransformer, i.e., the decoder-encoder structure, to learn the representationof symptoms by jointly optimizing the reinforce reward and cross entropy loss.Extensive experiments on three public real-world datasets prove that ourproposed model can effectively learn doctors' clinical experience and achievethe state-of-the-art results in terms of symptom recall and diagnosticaccuracy.
展开全部
图表提取

暂无人提供速读十问回答

论文十问由沈向洋博士提出,鼓励大家带着这十个问题去阅读论文,用有用的信息构建认知模型。写出自己的十问回答,还有机会在当前页面展示哦。

Q1论文试图解决什么问题?
Q2这是否是一个新的问题?
Q3这篇文章要验证一个什么科学假设?
0
被引用
笔记
问答