GithubHelp home page GithubHelp logo

nlab's Introduction

NLab

Labeling specification

The dataset should be extensible and able to adapt to the application scenario of multiple tasks. The dataset is marked with fine granularity. For a round of dialogue, the intention, slot-value and entity scope of each sentence of patients and doctors are labeled. Intents aim to determine the speaker's motivation, which helps analyze the semantics of generated statements. Patient intents and doctor intents are specified by role, and common intents include inquire and chitchat. Slots are designed to give each sentence a domain, equivalent to an entity meaning, and specify several slots for conversation generation tasks. Slot-value and entity ranges are specific tagging information for each sentence and can be used for tasks such as named entity recognition, disease, and symptom prediction. Patient and doctor intention are shown in Tables 1 and 2.

  • Table 1 Definition of patient intent
Intention Role Explain Example
Inform Patient The patient informs the doctor da bian zheng chang,ye bu e xin 大便正常,也不恶心
Inquire Patient The patient asks the doctor wo de le ji xing fu xie,gai ru he zhi liao? 我得了急性腹泻,该如何治疗?
Want confirm Patient Want to get the doctor's confirmation shi chang dao xi jun bu ping heng ma? 是肠道细菌不平衡吗?
Confirm Patient Reply to the doctor for confirmation you 有
Deny Patient Deny the doctor's confirmation fa shao mei you 发烧没有
Chitchat Patient Chitchat between patient and doctor hao de, xie xie 好的,谢谢
  • Table 2 Definition of doctor intent
Intention Role Explain Example
Diagnosis Doctor Doctor diagnose about patient zhe shi shi guan fan liu 这是食管反流
Recommendation Doctor Advice given by a doctor to a patient jian yi chi dian ao mei la zuo 建议吃点奥美拉唑
Not recommendation Doctor Things that doctors don't advise patients to do bi mian shi yong ci ji xing shi wu 避免使用刺激性食物
Want confirm Doctor Want confirmation bian mi you me 便秘有吗
Confirm Doctor Reply to the patient for confirmation shi de 是的
Deny Doctor Deny the patient's confirmation bu ke yi 不可以
Inquire Doctor The doctor questioned the patient hai you qi ta zheng Zhuang ma? 还有其他症状吗?
Chitchat Doctor Chitchat between doctor and patient hao de, xie xie 好的,谢谢

Slots can be regarded as a category, such as diseases, symptoms, and drugs. Slots are the specific manifestations of slots. Each slot may have multiple slots. The specific slot definition is shown in Table 3, and the slot value can only be obtained by specific analysis of a sentence.

  • Table 3 Slot definition
Slot Slot value Explain Example
Symptom Multiple values can be identified from statements or referenced to medical standard terms Statement for symptoms that may occur du qi zhou wei 【yin yin zuo tong】 肚脐周围【隐隐作痛】
Disease Multiple values can be identified from statements or referenced to medical standard terms Statements that refer to diseases that may not appear explicitly shu yu 【gao zhi xing yi xian yan】 属于【高脂性胰腺炎】。
Drug Identified from the statement, there can be multiple values The name of the drug mentioned in the statement ci shu duo jiu chi 【si mi da】 次数多就吃【思密达】
Check Identified from the statement, there can be multiple values The statement mentions the check that was done You cha 【xue chang gui】ma 有查【血常规】吗
Symptomatic site Identified from the statement, there can be multiple values Statement where symptoms occur 【zuo ce lei gu bu wei】hen teng 【左侧肋骨部位】很疼
Food name Identified from the statement, there can be multiple values The food mentioned in the statement man xing chang wei yan hui fu qi ke yi chi【zhi shi】ma 慢性肠胃炎恢复期可以吃【芝士】吗
Treatment site Identify from the statement The treatment site mentioned in the statement 【san jia yi yuan】wu tong wei jing yi ban qi bai duo 【三甲医院】无痛胃镜一般七百多
Department Identify from the statement The department mentioned in the statement dao【pi fu ke】kan yi xia 到【皮肤科】看一下
Body part Identify from the statement The part of the body mentioned in the statement bu hui qian che dao 【zuo bian jian bu】 不会牵扯到【左边肩部】
Dosage of drug Identify from the statement The dosage recommended by the doctor pi wei xiu an pian【yi ci yi pian】 匹维溴铵片【一次一片】
Frequency of medication Identify from the statement The frequency of medication recommended by your doctor pei fei kang jiao nang,【yi tian liang ci】 培菲康胶囊,【一天两次】
Preventive measure Identify from the statement The advice provided in the statement to alleviate the condition 【shao chi dian rou zhi pin, he dou zhi pin】, 【duo chi shu cai shui guo】 【少吃点肉制品,和豆制品】,【多吃蔬菜水果】
Drug class Identify from the statement Class of drug mentioned in the statement 【a mo xi lin】shi xiao yan yao 【阿莫西林】是消炎药
State Positive, negative, other Whether the proposed disease or symptom is real -
Time Identify from the statement The time and duration of symptoms described by the patient 【liang san tian】 le. 【两三天】了。
Behavior Identify from the statement An expression that refers to an action in a statement yin yin zuo tong 隐隐作痛 shi liang shao chi 适量少吃

Medical dialogue system

The Medical Dialogue system is in the DS directory. Use wechat mini program to build the client, and use python to complete the dialogue process at the back end. The medical dialogue system is mainly composed of three modules: natural language understanding module: obtaining the intention expressed in the patient text and extracting the corresponding medical entity; Knowledge fusion module: fusion of the current questions, predicted entities, dialogue history text and entity-relationship, role information; Dialogue generation module: converts the fused information into understandable text information.

Environment setting

QT4.3.1

python3.7

Wechat developer tools

nlab's People

Contributors

codegodliu avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.