检索增强生成大语言模型在肺病学领域的应用

打开文本图片集

中图分类号：R563 文献标识码：A 文章编号：1006-1959（2026）04-0094-05

Abstract：OjectiveToproeeectivenessndracflgelangageodelsipumoolgeriealgmentedeneratiootet throughthedevelopentofspecializedtextembedingodelsandmulifaceedretrievalmechansmsMethodsTispaperonstructed4 dimensionencoertetedgodelfroahasedohRcitectureouhfourstaerigprocesytingtlatet clinicalguidelinsadexpertpinionmateralsAmultiscaletextchnkingotimizationndmultiathinfomationetrievalhansm incorporatingectoeachndseahetseipleed，ndhalityfalwaseauatedsingitatedrol rank.Resultsleal4difocetlllll performance，with a highest mean reciprocal rank reaching O.7453，a highest first hit rate was 63.71% ，which was significantly higher than other models.Byoraaioudatraafulatfoaadi rankingstragttllliiroalct hit rate was 75.27% and a highest top-5 hit rate reaching 93.65 % .Conclusion The retrieval augmented generation with large language models has achievedhighaccuracyinpulmonologyretrieval，whichhelpsaddress therapid evolutionandcomplexityofpulmonology.

KeyWords：Retrieval augmented generation; Large language models;Pulmonology

大型语言模型在医学教育和决策中发挥了重要作用，其通常通过预训练任务来捕获语言的复杂性。（剩余8070字）