基于智能辅助医疗保障系统构建肺孢子菌肺炎早期预警模型的研究OA

Objective To evaluate the clinical value of an early warning model for Pneumocystis jirovecii pneumonia(PJP)developed using an intelligent medical support system in immunocompromised hematology patients.Methods A retrospective study conducted involving 179 immunocompromised hematologic malignancy patients who presented with fever or respiratory symptoms and received treatment at the Department of Hematology,Shanghai 411 Hospital from January 2025 to February 2026.An early warning model was constructed using a smart medical support system that integrated multi-source data,including immune status,fungal serological markers,imaging features,and clinical text information.The model's efficacy was validated using metrics such as diagnostic accuracy.Results Among 36 patients confirmed with PJP by metagenomic next-generation sequencing,the prediction model identified 34 true positive cases out of 63 positive alerts.Notably,all true positive cases were diagnosed within 72 hours after symptom onset,which was significantly earlier than the clinical diagnosis group(30 true positives out of 44 clinically diagnosed cases,with only 30％(9/30)diagnosed within 72 hours).The prediction model demonstrated a sensitivity of 94.44％and a specificity of 79.72％for diagnosing PJP in immunocompromised patients with hematological diseases.Conclusion The intelligent medical support system significantly enhances early PJP diagnosis efficiency through multimodal data fusion and a dual-track knowledge base,establishing a new paradigm for fungal infection prevention and control in immunosuppressed patients.