{"id":33292,"date":"2025-07-23T16:33:39","date_gmt":"2025-07-23T08:33:39","guid":{"rendered":"https:\/\/www.kdjingpai.com\/?p=33292"},"modified":"2025-07-23T16:33:39","modified_gmt":"2025-07-23T08:33:39","slug":"dapobileimianbai","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/ja\/dapobileimianbai\/","title":{"rendered":"\u6253\u7834\u58c1\u5792\uff1a\u514d\u8d39\u5f00\u653e\u7684\u00a0OpenMed\u00a0\u6a21\u578b\u5c06\u5982\u4f55\u6539\u53d8\u533b\u7597AI\u683c\u5c40\uff1f"},"content":{"rendered":"

\u957f\u671f\u4ee5\u6765\uff0c\u5c16\u7aef\u7684\u533b\u7597\u4eba\u5de5\u667a\u80fd\u6280\u672f\u4e00\u76f4\u88ab\u7981\u9522\u5728\u9ad8\u6602\u7684\u5546\u4e1a\u8bb8\u53ef\u548c\u4e0d\u900f\u660e\u7684\u201c\u9ed1\u7bb1\u201d\u7cfb\u7edf\u4e4b\u540e\u3002\u8fd9\u4f7f\u5f97\u8bb8\u591a\u7814\u7a76\u673a\u6784\u3001\u4e2d\u5c0f\u578b\u5f00\u53d1\u56e2\u961f\u548c\u4e00\u7ebf\u533b\u751f\u671b\u800c\u5374\u6b65\uff0c\u51cf\u7f13\u4e86\u6280\u672f\u521b\u65b0\u548c\u516c\u5e73\u5e94\u7528\u7684\u8fdb\u7a0b\u3002\u5982\u4eca\uff0c\u4e00\u4e2a\u540d\u4e3a\u00a0OpenMed<\/a><\/code>\u00a0\u7684\u65b0\u9879\u76ee\u6b63\u8bd5\u56fe\u6253\u7834\u8fd9\u4e00\u50f5\u5c40\u3002\u8be5\u9879\u76ee\u5728\u00a0Hugging Face<\/code>\u00a0\u793e\u533a\u53d1\u5e03\u4e86\u8d85\u8fc7380\u4e2a\u5148\u8fdb\u7684\u533b\u7597\u4e0e\u4e34\u5e8a\u6587\u672c\u547d\u540d\u5b9e\u4f53\u8bc6\u522b\uff08NER\uff09\u6a21\u578b\uff0c\u5e76\u5ba3\u5e03\u57fa\u4e8e\u00a0Apache 2.0<\/code>\u00a0\u8bb8\u53ef\u8bc1\u6c38\u4e45\u514d\u8d39\u5f00\u653e\u3002<\/p>\n

\u6b64\u4e3e\u7684\u610f\u4e49\u4e0d\u4ec5\u5728\u4e8e\u63d0\u4f9b\u514d\u8d39\u7684\u66ff\u4ee3\u54c1\uff0c\u66f4\u5728\u4e8e\u5176\u53d1\u5e03\u7684\u6a21\u578b\u5728\u591a\u4e2a\u57fa\u51c6\u6d4b\u8bd5\u4e2d\uff0c\u6027\u80fd\u751a\u81f3\u8d85\u8d8a\u4e86\u4e3b\u6d41\u7684\u5546\u4e1a\u95ed\u6e90\u65b9\u6848\u3002\u8fd9\u9884\u793a\u7740\u533b\u7597AI\u9886\u57df\u7684\u6280\u672f\u58c1\u5792\u6b63\u5728\u88ab\u5f00\u6e90\u529b\u91cf\u74e6\u89e3\u3002<\/p>\n

\u884c\u4e1a\u56f0\u5883\u4e0e\u5f00\u6e90\u65b9\u6848<\/h2>\n

\u533b\u7597AI\u9886\u57df\u7684\u53d1\u5c55\u9762\u4e34\u7740\u51e0\u4e2a\u6838\u5fc3\u969c\u788d\uff1a<\/p>\n

    \n
  • \u9ad8\u6602\u7684\u8bb8\u53ef\u8d39\u7528<\/strong>\uff1a\u9876\u7ea7\u7684\u5546\u4e1aAI\u5de5\u5177\u8bb8\u53ef\u8d39\u7528\u9ad8\u6602\uff0c\u5c06\u9884\u7b97\u6709\u9650\u7684\u5b66\u672f\u673a\u6784\u548c\u521d\u521b\u516c\u53f8\u6392\u9664\u5728\u5916\u3002<\/li>\n
  • \u6280\u672f\u4e0d\u900f\u660e<\/strong>\uff1a\u5546\u4e1a\u5de5\u5177\u901a\u5e38\u4e0d\u516c\u5f00\u5176\u6a21\u578b\u67b6\u6784\u3001\u8bad\u7ec3\u6570\u636e\u548c\u5de5\u4f5c\u539f\u7406\uff0c\u7528\u6237\u96be\u4ee5\u8bc4\u4f30\u5176\u53ef\u9760\u6027\u548c\u6f5c\u5728\u504f\u89c1\u3002<\/li>\n
  • \u6280\u672f\u8fed\u4ee3\u7f13\u6162<\/strong>\uff1a\u90e8\u5206\u4ed8\u8d39\u6a21\u578b\u672a\u80fd\u8ddf\u4e0a\u6700\u65b0\u7684AI\u6280\u672f\u8fdb\u5c55\uff0c\u6027\u80fd\u9010\u6e10\u843d\u540e\u4e8e\u5b66\u672f\u754c\u7684\u524d\u6cbf\u7814\u7a76\u3002<\/li>\n
  • \u5e94\u7528\u53d7\u9650<\/strong>\uff1a\u4f18\u8d28\u7684AI\u80fd\u529b\u96c6\u4e2d\u5728\u5c11\u6570\u5927\u578b\u4f01\u4e1a\u624b\u4e2d\uff0c\u9650\u5236\u4e86\u6280\u672f\u666e\u60e0\u3002<\/li>\n<\/ul>\n

    OpenMed<\/code>\u00a0\u9879\u76ee\u76f4\u63a5\u56de\u5e94\u4e86\u8fd9\u4e9b\u6311\u6218\u3002\u5b83\u63d0\u4f9b\u7684 380 \u591a\u4e2a\u514d\u8d39\u00a0NER<\/code>\u00a0\u6a21\u578b\uff0c\u4e13\u6ce8\u4e8e\u8bc6\u522b\u533b\u5b66\u6587\u672c\u4e2d\u7684\u5173\u952e\u5b9e\u4f53\uff0c\u5982\u836f\u7269\u540d\u79f0\u3001\u75be\u75c5\u3001\u57fa\u56e0\u3001\u89e3\u5256\u7ed3\u6784\u7b49\u3002\u8fd9\u4e9b\u6a21\u578b\u5177\u5907\u4ee5\u4e0b\u7a81\u51fa\u7279\u70b9\uff1a<\/p>\n

      \n
    • \u2705\u00a0\u5b8c\u5168\u514d\u8d39<\/strong>\uff1a\u57fa\u4e8e\u00a0Apache 2.0<\/code>\u00a0\u8bb8\u53ef\u8bc1\uff0c\u5141\u8bb8\u81ea\u7531\u4f7f\u7528\u3001\u4fee\u6539\u548c\u5206\u53d1\u3002<\/li>\n
    • \u2705\u00a0\u5373\u7528\u6027<\/strong>\uff1a\u4e13\u4e3a\u5b9e\u9645\u5e94\u7528\u573a\u666f\u8bbe\u8ba1\uff0c\u65e0\u9700\u5927\u91cf\u989d\u5916\u5de5\u4f5c\u5373\u53ef\u90e8\u7f72\u3002<\/li>\n
    • \u2705\u00a0\u5c3a\u5bf8\u7075\u6d3b<\/strong>\uff1a\u6a21\u578b\u53c2\u6570\u91cf\u4ece 109M \u5230 568M \u4e0d\u7b49\uff0c\u9002\u5e94\u4e0d\u540c\u786c\u4ef6\u9700\u6c42\u3002<\/li>\n
    • \u2705\u00a0\u7ecf\u8fc7\u9a8c\u8bc1<\/strong>\uff1a\u5728\u8d85\u8fc7 13 \u4e2a\u533b\u7597\u9886\u57df\u7684\u6807\u51c6\u6570\u636e\u96c6\u4e0a\u8fdb\u884c\u4e86\u4e25\u683c\u7684\u6027\u80fd\u6d4b\u8bd5\u3002<\/li>\n
    • \u2705\u00a0\u751f\u6001\u517c\u5bb9<\/strong>\uff1a\u4e0e\u00a0Hugging Face<\/code>\u00a0\u548c\u00a0PyTorch<\/code>\u00a0\u7b49\u4e3b\u6d41\u6846\u67b6\u65e0\u7f1d\u96c6\u6210\u3002<\/li>\n<\/ul>\n

      OpenMed<\/code>\u00a0\u5de5\u5177\u7bb1\u8be6\u89e3<\/h2>\n

      OpenMed<\/code>\u00a0\u7684\u6a21\u578b\u5e93\u7ecf\u8fc7\u7cbe\u5fc3\u5fae\u8c03\u548c\u6d4b\u8bd5\uff0c\u5728\u00a0Gellus<\/code>\u00a0\u7b49\u6570\u636e\u96c6\u4e0a\u7684 F1 \u5206\u6570\u9ad8\u8fbe\u00a00.998<\/code>\uff0c\u5c55\u793a\u4e86\u5176\u5353\u8d8a\u7684\u6027\u80fd\u3002<\/p>\n

      \ud83d\udd2c \u6027\u80fd\u5bf9\u6bd4\uff1a\u5f00\u6e90\u00a0OpenMed<\/code>\u00a0\u4e0e\u95ed\u6e90\u5546\u4e1a\u6a21\u578b<\/h3>\n

      \u4e3a\u4e86\u76f4\u89c2\u5c55\u793a\u5176\u7ade\u4e89\u529b\uff0cOpenMed<\/code>\u00a0\u516c\u5e03\u4e86\u4e0e\u5f53\u524d\u6700\u5148\u8fdb\u7684\u95ed\u6e90\u5546\u4e1a\u6a21\u578b\u7684\u6027\u80fd\u5bf9\u6bd4\u3002\u6570\u636e\u663e\u793a\uff0cOpenMed<\/code>\u00a0\u4e0d\u4ec5\u5728\u591a\u4e2a\u6570\u636e\u96c6\u4e0a\u4e0e\u5546\u4e1a\u6a21\u578b\u76f8\u5f53\uff0c\u66f4\u5728\u67d0\u4e9b\u573a\u666f\u4e0b\u5b9e\u73b0\u4e86\u663e\u8457\u8d85\u8d8a\u3002<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
      \u6570\u636e\u96c6<\/th>\nOpenMed \u6700\u4f73 F1 \u5206\u6570 (%)<\/th>\n\u95ed\u6e90 SOTA F1 \u5206\u6570 (%)\u2020<\/th>\n\u5dee\u8ddd (OpenMed \u2013 SOTA)<\/th>\n\u5f53\u524d\u95ed\u6e90\u9886\u5148\u8005<\/th>\n<\/tr>\n<\/thead>\n
      BC4CHEMD<\/strong><\/td>\n95.40<\/strong><\/td>\n94.39<\/td>\n+1.01<\/strong><\/td>\nSpark NLP BertForTokenClassification<\/td>\n<\/tr>\n
      BC5CDR-Chem<\/strong><\/td>\n96.10<\/strong><\/td>\n94.88<\/td>\n+1.22<\/strong><\/td>\nSpark NLP BertForTokenClassification<\/td>\n<\/tr>\n
      BC5CDR-Disease<\/strong><\/td>\n91.20<\/strong><\/td>\n88.5<\/td>\n+2.70<\/strong><\/td>\nBioMegatron<\/td>\n<\/tr>\n
      NCBI-Disease<\/strong><\/td>\n91.10<\/strong><\/td>\n89.71<\/td>\n+1.39<\/strong><\/td>\nBioBERT<\/td>\n<\/tr>\n
      JNLPBA<\/strong><\/td>\n81.90<\/td>\n82.00<\/strong><\/td>\n\u20130.10<\/td>\nKeBioLM (knowledge-enhanced LM)<\/td>\n<\/tr>\n
      Linnaeus<\/strong><\/td>\n96.50<\/strong><\/td>\n92.70<\/td>\n+3.80<\/strong><\/td>\nBERN2 toolkit<\/td>\n<\/tr>\n
      Species-800<\/strong><\/td>\n86.40<\/strong><\/td>\n82.59<\/td>\n+3.81<\/strong><\/td>\nSpark NLP BertForTokenClassification<\/td>\n<\/tr>\n
      BC2GM<\/strong><\/td>\n90.10<\/strong><\/td>\n88.75<\/td>\n+1.35<\/strong><\/td>\nSpark NLP Bi-LSTM-CNN-Char<\/td>\n<\/tr>\n
      AnatEM<\/strong><\/td>\n90.60<\/td>\n91.65<\/strong><\/td>\n\u20131.05<\/td>\nSpark NLP BertForTokenClassification<\/td>\n<\/tr>\n
      BioNLP 2013 CG<\/strong><\/td>\n89.90<\/strong><\/td>\n87.83<\/td>\n+2.07<\/strong><\/td>\nSpark NLP BertForTokenClassification<\/td>\n<\/tr>\n
      Gellus<\/strong><\/td>\n99.80<\/strong><\/td>\n63.40<\/td>\n+36.40<\/strong><\/td>\nConNER<\/td>\n<\/tr>\n
      CLL<\/strong><\/td>\n95.70<\/strong><\/td>\n85.98<\/td>\n\u2014<\/td>\n(\u65e0\u5df2\u53d1\u5e03\u7684 SOTA)<\/em><\/td>\n<\/tr>\n
      FSU<\/strong><\/td>\n96.10<\/strong><\/td>\n\u2014<\/td>\n\u2014<\/td>\n(\u65e0\u5df2\u53d1\u5e03\u7684 SOTA)<\/em><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

      \u2020 \u95ed\u6e90\u5206\u6570\u6765\u6e90\u4e8e\u6587\u732e\u4e2d\u5df2\u53d1\u8868\u7684\u6700\u9ad8\u540c\u884c\u8bc4\u5ba1\u6216\u6392\u884c\u699c\u7ed3\u679c\uff08\u901a\u5e38\u662f Spark NLP\u3001NEEDLE\u3001BERN2 \u7b49\u5546\u4e1a\u6a21\u578b\uff09\u3002<\/p>\n

      \"\u6253\u7834\u58c1\u5792\uff1a\u514d\u8d39\u5f00\u653e\u7684\u00a0OpenMed\u00a0\u6a21\u578b\u5c06\u5982\u4f55\u6539\u53d8\u533b\u7597AI\u683c\u5c40\uff1f-1\"<\/p>\n

      \u5c24\u5176\u503c\u5f97\u5173\u6ce8\u7684\u662f\uff0c\u5728\u00a0Gellus<\/code>\u00a0\u6570\u636e\u96c6\u4e0a\uff0cOpenMed<\/code>\u00a0\u7684 F1 \u5206\u6570\u9ad8\u51fa\u5148\u524d\u6700\u4f73\u6a21\u578b 36.4%\uff0c\u8fd9\u8868\u660e\u5728\u7279\u5b9a\u4efb\u52a1\u4e0a\uff0c\u4e13\u6ce8\u4f18\u5316\u7684\u5f00\u6e90\u6a21\u578b\u62e5\u6709\u5de8\u5927\u7684\u6f5c\u529b\u3002<\/p>\n

      \ud83d\udd2c \u6309\u5e94\u7528\u9886\u57df\u5212\u5206<\/h3>\n

      \u4e0b\u8868\u5c06\u6570\u636e\u96c6\u4e0e\u5176\u5bf9\u5e94\u7684\u533b\u7597\u9886\u57df\u8fdb\u884c\u6620\u5c04\uff0c\u5e76\u6839\u636e\u5404\u9886\u57df\u6570\u636e\u96c6\u7684\u7efc\u5408\u8868\u73b0\u63a8\u8350\u4e86\u76f8\u5e94\u7684\u6a21\u578b\u3002<\/p>\n\n\n\n\n\n\n\n\n\n\n
      \u9886\u57df<\/th>\n\u5305\u542b\u7684\u6570\u636e\u96c6<\/th>\n\u53ef\u7528\u6a21\u578b\u6570\u91cf<\/th>\n\u53c2\u6570\u91cf\u8303\u56f4<\/th>\n\u63a8\u8350\u6a21\u578b<\/th>\n<\/tr>\n<\/thead>\n
      \u836f\u7406\u5b66<\/strong><\/td>\nbc5cdr_chem<\/code>,\u00a0bc4chemd<\/code>,\u00a0fsu<\/code><\/td>\n90 \u4e2a<\/td>\n109M – 568M<\/td>\nOpenMed-NER-PharmaDetect-SuperClinical-434M<\/code><\/td>\n<\/tr>\n
      \u75be\u75c5\/\u75c5\u7406\u5b66<\/strong><\/td>\nbc5cdr_disease<\/code>,\u00a0ncbi_disease<\/code><\/td>\n60 \u4e2a<\/td>\n109M – 434M<\/td>\nOpenMed-NER-PathologyDetect-PubMed-v2-109M<\/code><\/td>\n<\/tr>\n
      \u57fa\u56e0\u7ec4\u5b66<\/strong><\/td>\njnlpba<\/code>,\u00a0bc2gm<\/code>,\u00a0species800<\/code>,\u00a0linnaeus<\/code>,\u00a0gellus<\/code><\/td>\n150 \u4e2a<\/td>\n335M – 568M<\/td>\nOpenMed-NER-GenomicDetect-SnowMed-568M<\/code><\/td>\n<\/tr>\n
      \u89e3\u5256\u5b66<\/strong><\/td>\nanatomy<\/code><\/td>\n30 \u4e2a<\/td>\n560M<\/td>\nOpenMed-NER-AnatomyDetect-ElectraMed-560M<\/code><\/td>\n<\/tr>\n
      \u80bf\u7624\u5b66<\/strong><\/td>\nbionlp2013_cg<\/code><\/td>\n30 \u4e2a<\/td>\n355M<\/td>\nOpenMed-NER-OncologyDetect-SuperMedical-355M<\/code><\/td>\n<\/tr>\n
      \u4e34\u5e8a\u8bb0\u5f55<\/strong><\/td>\ncll<\/code><\/td>\n30 \u4e2a<\/td>\n560M<\/td>\nOpenMed-NER-BloodCancerDetect-ElectraMed-560M<\/code><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

      \u26a1\ufe0f \u6309\u6a21\u578b\u5927\u5c0f\u9009\u62e9<\/h3>\n\n\n\n\n\n\n\n\n
      \u5927\u5c0f<\/th>\n\u53c2\u6570\u91cf<\/th>\n\u6700\u4f73\u9002\u7528\u573a\u666f<\/th>\n<\/tr>\n<\/thead>\n
      \u7d27\u51d1\u578b<\/strong><\/td>\n109M<\/td>\n\u5feb\u901f\u539f\u578b\u548c\u4f4e\u8d44\u6e90\u73af\u5883<\/td>\n<\/tr>\n
      \u5927\u578b<\/strong><\/td>\n335M – 355M<\/td>\n\u7cbe\u5ea6\u4e0e\u6027\u80fd\u7684\u5747\u8861\u9009\u62e9<\/td>\n<\/tr>\n
      \u8d85\u5927\u578b<\/strong><\/td>\n434M<\/td>\n\u5168\u80fd\u578b\uff0c\u6027\u80fd\u4f18\u5f02<\/td>\n<\/tr>\n
      \u5de8\u5927\u578b<\/strong><\/td>\n560M – 568M<\/td>\n\u8ffd\u6c42\u6781\u81f4\u7cbe\u5ea6\u7684\u4efb\u52a1<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

      \"\u6253\u7834\u58c1\u5792\uff1a\u514d\u8d39\u5f00\u653e\u7684\u00a0OpenMed\u00a0\u6a21\u578b\u5c06\u5982\u4f55\u6539\u53d8\u533b\u7597AI\u683c\u5c40\uff1f-2\"<\/p>\n

      \ud83d\udcca \u5404\u6570\u636e\u96c6\u4e0a\u7684\u6700\u4f73\u6a21\u578b<\/h3>\n

      \u4ee5\u4e0b\u662f\u6bcf\u4e2a\u6570\u636e\u96c6\u4e0a\u8868\u73b0\u6700\u4f73\u7684\u6a21\u578b\u53ca\u5176 F1 \u5206\u6570\u548c\u5927\u5c0f\u7684\u6458\u8981\u3002<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
      \u6570\u636e\u96c6<\/th>\n\u6700\u4f73\u6a21\u578b<\/th>\nF1 \u5206\u6570<\/th>\n\u6a21\u578b\u5927\u5c0f (\u53c2\u6570)<\/th>\n<\/tr>\n<\/thead>\n
      bc5cdr_chem<\/code><\/td>\nOpenMed-NER-PharmaDetect-SuperClinical-434M<\/code><\/a><\/td>\n0.961<\/td>\n434M<\/td>\n<\/tr>\n
      bionlp2013_cg<\/code><\/td>\nOpenMed-NER-OncologyDetect-SuperMedical-355M<\/code><\/a><\/td>\n0.899<\/td>\n355M<\/td>\n<\/tr>\n
      bc4chemd<\/code><\/td>\nOpenMed-NER-ChemicalDetect-PubMed-335M<\/code><\/a><\/td>\n0.954<\/td>\n335M<\/td>\n<\/tr>\n
      linnaeus<\/code><\/td>\nOpenMed-NER-SpeciesDetect-PubMed-335M<\/code><\/a><\/td>\n0.965<\/td>\n335M<\/td>\n<\/tr>\n
      jnlpba<\/code><\/td>\nOpenMed-NER-DNADetect-SuperClinical-434M<\/code><\/a><\/td>\n0.819<\/td>\n434M<\/td>\n<\/tr>\n
      bc5cdr_disease<\/code><\/td>\nOpenMed-NER-DiseaseDetect-SuperClinical-434M<\/code><\/a><\/td>\n0.912<\/td>\n434M<\/td>\n<\/tr>\n
      fsu<\/code><\/td>\nOpenMed-NER-ProteinDetect-SnowMed-568M<\/code><\/a><\/td>\n0.961<\/td>\n568M<\/td>\n<\/tr>\n
      ncbi_disease<\/code><\/td>\nOpenMed-NER-PathologyDetect-PubMed-v2-109M<\/code><\/a><\/td>\n0.911<\/td>\n109M<\/td>\n<\/tr>\n
      bc2gm<\/code><\/td>\nOpenMed-NER-GenomeDetect-SuperClinical-434M<\/code><\/a><\/td>\n0.901<\/td>\n434M<\/td>\n<\/tr>\n
      cll<\/code><\/td>\nOpenMed-NER-BloodCancerDetect-ElectraMed-560M<\/code><\/a><\/td>\n0.957<\/td>\n560M<\/td>\n<\/tr>\n
      gellus<\/code><\/td>\nOpenMed-NER-GenomicDetect-SnowMed-568M<\/code><\/a><\/td>\n0.998<\/td>\n568M<\/td>\n<\/tr>\n
      anatomy<\/code><\/td>\nOpenMed-NER-AnatomyDetect-ElectraMed-560M<\/code><\/a><\/td>\n0.906<\/td>\n560M<\/td>\n<\/tr>\n
      species800<\/code><\/td>\nOpenMed-NER-OrganismDetect-BioMed-335M<\/code><\/a><\/td>\n0.864<\/td>\n335M<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

      \u5feb\u901f\u4e0a\u624b\u4e0e\u89c4\u6a21\u5316\u5e94\u7528<\/h2>\n

      \u501f\u52a9\u00a0Hugging Face Transformers<\/code>\u00a0\u5e93\uff0c\u96c6\u6210\u00a0OpenMed<\/code>\u00a0\u6a21\u578b\u7684\u8fc7\u7a0b\u975e\u5e38\u7b80\u5355\uff0c\u53ea\u9700\u4e09\u884c\u4ee3\u7801\u5373\u53ef\u8c03\u7528\u3002<\/p>\n

      from transformers import pipeline\r\nner_pipeline = pipeline(\"token-classification\", model=\"OpenMed\/OpenMed-NER-PharmaDetect-SuperClinical-434M\", aggregation_strategy=\"simple\")\r\ntext = \"Patient prescribed 10mg aspirin for hypertension.\"\r\nentities = ner_pipeline(text)\r\nprint(entities)\r\n# \u8f93\u51fa: [{'entity_group': 'CHEMICAL', 'score': 0.99..., 'word': 'aspirin', 'start': 28, 'end': 35}]\r\n<\/code><\/pre>\n
      \u5bf9\u4e8e\u9700\u8981\u5904\u7406\u5927\u89c4\u6a21\u6570\u636e\u96c6\u7684\u573a\u666f\uff0c\u8be5\u9879\u76ee\u4e5f\u63d0\u4f9b\u4e86\u9ad8\u6548\u7684\u6279\u5904\u7406\u65b9\u6848\u3002<\/p>\n
      from transformers.pipelines.pt_utils import KeyDataset\r\nfrom datasets import Dataset, load_dataset\r\nimport pandas as pd\r\n# \u52a0\u8f7d\u516c\u5f00\u7684\u533b\u7597\u6570\u636e\u96c6\uff08\u4f7f\u7528\u4e00\u4e2a\u5b50\u96c6\u8fdb\u884c\u6d4b\u8bd5\uff09\r\nmedical_dataset = load_dataset(\"BI55\/MedText\", split=\"train[:100]\")\r\ndata = pd.DataFrame({\"text\": medical_dataset[\"Completion\"]})\r\ndataset = Dataset.from_pandas(data)\r\n# \u4f7f\u7528\u9002\u5408\u60a8\u786c\u4ef6\u7684\u6279\u5904\u7406\u5927\u5c0f\r\nbatch_size = 16  # \u6839\u636e\u60a8\u7684 GPU \u663e\u5b58\u8fdb\u884c\u8c03\u6574\r\nresults = []\r\nner_pipeline = pipeline(\"token-classification\", model=\"OpenMed\/OpenMed-NER-PharmaDetect-SuperClinical-434M\", device=0) # \u4f7f\u7528GPU\r\nfor out in ner_pipeline(KeyDataset(dataset, \"text\"), batch_size=batch_size):\r\nresults.extend(out)\r\nprint(f\"\u6279\u5904\u7406\u5b8c\u6210 {len(results)} \u6761\u6587\u672c\")\r\n<\/code><\/pre>\n
      NER<\/code>\u00a0\u6280\u672f\u89e3\u9501\u7684\u771f\u5b9e\u4e16\u754c\u4ef7\u503c<\/h2>\n
      \u547d\u540d\u5b9e\u4f53\u8bc6\u522b\uff08NER\uff09\u6280\u672f\u80fd\u591f\u4ece\u975e\u7ed3\u6784\u5316\u7684\u6587\u672c\u4e2d\u81ea\u52a8\u63d0\u53d6\u548c\u5206\u7c7b\u5173\u952e\u4fe1\u606f\u3002\u5728\u533b\u7597\u9886\u57df\uff0c\u8fd9\u9879\u6280\u672f\u662f\u6fc0\u6d3b\u4e34\u5e8a\u7b14\u8bb0\u3001\u60a3\u8005\u8bb0\u5f55\u548c\u79d1\u7814\u6587\u732e\u4e2d\u6d77\u91cf\u6570\u636e\u4ef7\u503c\u7684\u50ac\u5316\u5242\u3002<\/p>\n
        \n
      • \ud83d\udd12 \u60a3\u8005\u9690\u79c1\u4fdd\u62a4\uff08\u6570\u636e\u53bb\u6807\u8bc6\u5316\uff09<\/strong>\uff1aNER<\/code>\u00a0\u53ef\u4ee5\u81ea\u52a8\u8bc6\u522b\u5e76\u79fb\u9664\u533b\u7597\u8bb0\u5f55\u4e2d\u7684\u4e2a\u4eba\u5065\u5eb7\u4fe1\u606f\uff08PHI\uff09\uff0c\u5982\u59d3\u540d\u3001\u5730\u5740\u7b49\u3002\u8fd9\u5728\u4fdd\u62a4\u60a3\u8005\u9690\u79c1\u3001\u9075\u5b88\u00a0HIPAA<\/code>\u00a0\u7b49\u6cd5\u5f8b\u6cd5\u89c4\u7684\u540c\u65f6\uff0c\u4e3a\u533b\u5b66\u7814\u7a76\u63d0\u4f9b\u4e86\u5408\u89c4\u3001\u5b89\u5168\u7684\u6570\u636e\u6765\u6e90\uff0c\u5176\u6548\u7387\u548c\u51c6\u786e\u6027\u8fdc\u8d85\u624b\u52a8\u5904\u7406\u3002<\/li>\n
      • \ud83d\udd17 \u533b\u7597\u77e5\u8bc6\u56fe\u8c31\u6784\u5efa\uff08\u5b9e\u4f53\u5173\u7cfb\u63d0\u53d6\uff09<\/strong>\uff1a\u5728\u8bc6\u522b\u51fa\u836f\u7269\u3001\u75be\u75c5\u7b49\u5b9e\u4f53\u540e\uff0c\u8fdb\u4e00\u6b65\u7684\u6280\u672f\u53ef\u4ee5\u5206\u6790\u5b83\u4eec\u4e4b\u95f4\u7684\u5173\u7cfb\uff08\u5982\u201c\u836f\u7269 A \u5bfc\u81f4\u526f\u4f5c\u7528 B\u201d\uff09\u3002\u8fd9\u6709\u52a9\u4e8e\u6784\u5efa\u533b\u7597\u77e5\u8bc6\u56fe\u8c31\uff0c\u4e3a\u4e34\u5e8a\u51b3\u7b56\u63d0\u4f9b\u652f\u6301\uff0c\u52a0\u901f\u836f\u7269\u7814\u53d1\uff0c\u5e76\u6700\u7ec8\u5b9e\u73b0\u4e2a\u6027\u5316\u6cbb\u7597\u3002<\/li>\n
      • \ud83d\udca1 \u4f18\u5316\u533b\u7597\u6210\u672c\u4e0e\u7ba1\u7406\uff08HCC \u7f16\u7801\uff09<\/strong>\uff1a\u5206\u7ea7\u75c5\u60c5\u7c7b\u522b\uff08HCC\uff09\u7f16\u7801\u662f\u533b\u7597\u652f\u4ed8\u65b9\uff08\u5982 Medicare\uff09\u7528\u4e8e\u9884\u6d4b\u6210\u672c\u548c\u8bbe\u5b9a\u62a5\u9500\u7387\u7684\u5173\u952e\u6d41\u7a0b\u3002NER<\/code>\u00a0\u53ef\u4ece\u75c5\u5386\u4e2d\u81ea\u52a8\u63d0\u53d6\u8bca\u65ad\u4fe1\u606f\u4ee5\u8f85\u52a9\u7f16\u7801\uff0c\u786e\u4fdd\u533b\u7597\u673a\u6784\u56e0\u6cbb\u7597\u590d\u6742\u75c5\u4f8b\u83b7\u5f97\u516c\u5e73\u62a5\u916c\uff0c\u540c\u65f6\u5e2e\u52a9\u8bc6\u522b\u9ad8\u98ce\u9669\u60a3\u8005\u4ee5\u8fdb\u884c\u4e3b\u52a8\u5e72\u9884\u3002<\/li>\n<\/ul>\n
        \u901a\u8fc7\u63a8\u52a8\u8fd9\u4e9b\u5173\u952e\u4efb\u52a1\u7684\u81ea\u52a8\u5316\uff0cNER<\/code>\u00a0\u6280\u672f\u6b63\u5728\u5c06\u6c89\u7761\u7684\u533b\u7597\u6587\u672c\u8f6c\u5316\u4e3a\u53ef\u64cd\u4f5c\u7684\u89e3\u51b3\u65b9\u6848\uff0c\u4ece\u800c\u63d0\u5347\u6570\u636e\u5b89\u5168\u6027\u3001\u52a0\u901f\u7814\u7a76\u8fdb\u7a0b\u3001\u6539\u5584\u60a3\u8005\u9884\u540e\u5e76\u964d\u4f4e\u8fd0\u8425\u6210\u672c\u3002OpenMed<\/code>\u00a0\u7684\u51fa\u73b0\uff0c\u65e0\u7591\u5c06\u6781\u5927\u5730\u52a0\u901f\u8fd9\u4e00\u8fdb\u7a0b\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
        \u957f\u671f\u4ee5\u6765\uff0c\u5c16\u7aef\u7684\u533b\u7597\u4eba\u5de5\u667a\u80fd\u6280\u672f\u4e00\u76f4\u88ab\u7981\u9522\u5728\u9ad8\u6602\u7684\u5546\u4e1a\u8bb8\u53ef\u548c\u4e0d\u900f\u660e\u7684\u201c\u9ed1\u7bb1\u201d\u7cfb\u7edf\u4e4b\u540e\u3002\u8fd9\u4f7f\u5f97\u8bb8\u591a\u7814\u7a76\u673a\u6784\u3001\u4e2d\u5c0f\u578b\u5f00\u53d1\u56e2\u961f\u548c\u4e00\u7ebf\u533b\u751f\u671b\u800c\u5374\u6b65\uff0c\u51cf\u7f13\u4e86\u6280\u672f\u521b\u65b0\u548c\u516c\u5e73\u5e94\u7528\u7684\u8fdb\u7a0b\u3002\u5982\u4eca\uff0c\u4e00\u4e2a\u540d\u4e3a\u00a0OpenMed\u00a0\u7684\u65b0\u9879\u76ee\u6b63\u8bd5\u56fe\u6253\u7834\u8fd9\u4e00\u50f5\u5c40\u3002\u8be5\u9879\u76ee\u5728…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[46],"tags":[],"class_list":["post-33292","post","type-post","status-publish","format-standard","hentry","category-news"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts\/33292","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/comments?post=33292"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts\/33292\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/media?parent=33292"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/categories?post=33292"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/tags?post=33292"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}