2292 - 利用人工智能变革急诊分诊：在 IBM Power 上构建和部署概念验证分诊机器人.pdf

《2292 - 利用人工智能变革急诊分诊：在 IBM Power 上构建和部署概念验证分诊机器人.pdf》由会员分享，可在线阅读，更多相关《2292 - 利用人工智能变革急诊分诊：在 IBM Power 上构建和部署概念验证分诊机器人.pdf（13页珍藏版）》请在三个皮匠报告上搜索。

1、Orlando,FLOctober 69IBM TechXchange 2025Session code 2292Dr.Eva-Angela LeiblAndreas LeiblTransforming ER Triage with AIBuilding and Deploying a Proof-of-Concept triage bot on IBM PowerChallenges of the Healthcare SectorIncreasing staff shortagesInability to recruit sufficient personnelIncreasing dem

2、andWe get older,but not healthierSolution:Automation-Traditional-Self service-AIIBM TechXchange|2025 IBM Corporation2Patients coming in-By ambulance-With doctors referralNot a problemPatients coming on their ownWe need to address pressure from these patientsEmergency DepartmentInfluxIBM TechXchange|

3、2025 IBM Corporation3Language barriersNot everyone understand and speaks English wellComputer illiteracy Need a voice interfaceSingle large model impracticalMissing data to train oneRisk of over-or undertriage Potential danger to lifeTask:Use AI to Determine Severity of Patient ConditionThe Challeng

4、es:4IBM TechXchange|2025 IBM CorporationImprecise nature of human conversationHow many ways to express pain?Multilingual is a problemLLMs only reason well in EnglishTo prepare for upcoming multilingual use cases,over 5%of the Llama 3 pretraining dataset consists of high-quality non-English data that

5、 covers over 30 languages.However,we do not expect the same level of performance in these languages as in English.”(Introducing Meta Llama 3,2024,https:/ works best if English onlyInternally using just one language allows for use of smaller LLMsNon-English languages are covered by the user interface

6、5IBM TechXchange|2025 IBM CorporationFrom a Masters Thesis in“Digital Transformation in Healthcare”Title:Digital pathways in the emergency department:AI as a pioneer of modern triage“Author:Dr.Eva-Angela LeiblTriage bot=Team of small botsMost sub-bots can be small LLMsSub-bots perform single defined