Bobart Shane A, Hsu Enshuo, Potter Thomas, Truong Luan, Waterman Amy, Jones Stephen, Shafi Tariq
Division of Nephrology, Hypertension and Transplantation, Houston Methodist Hospital, Houston, TX.
Division of Nephrology and Hypertension, Mayo Clinic, Jacksonville, FL.
Kidney Med. 2025 Jun 14;7(8):101047. doi: 10.1016/j.xkme.2025.101047. eCollection 2025 Aug.
RATIONALE & OBJECTIVE: Kidney biopsy reports are in a nonindexed text format, and the diagnosis requires labor-intensive manual abstraction. Natural language processing (NLP) has not been rigorously tested for kidney biopsy diagnosis extraction. Our objective was to develop an accurate model to extract the biopsy diagnosis from free-text reports.
Text classification using NLP.
SETTING & PARTICIPANTS: 2,666 patients with 3,042 native kidney biopsy reports in the Portable Document Format, from June 2016 to December 2023.
Kidney biopsy diagnosis.
The performance of the NLP algorithm for all and the 20 most common diagnoses based on precision, recall, F1 score, and area under the receiver operating curve (AUROC).
A domain expert manually abstracted the diagnosis, and a renal pathologist validated a random subset (n = 200). Structured Query Language server and Python processed reports into machine-readable free text. We used PubMed Bidirectional Encoder Representations from Transformers to develop our NLP algorithm. We randomly split the reports into training (80%; n = 2,434) and testing (20%; n = 608) sets to train the NLP system. We further divided the testing set into 20% validation and 80% fine-tuning sets.
The median age was 57 years, with 50% female, 29% African Americans, and 23% Hispanic participants. The 5 most frequent glomerular diagnoses were diabetic kidney disease (23.7%), focal segmental glomerulosclerosis (15.5%), lupus nephritis (9.7%), immunoglobulin A nephropathy (8.9), and membranous nephropathy (7.2%). The Cohen kappa coefficient for interrater reliability was 0.76. PubMed Bidirectional Encoder Representations from Transformers fine-tuned with a training set showed the average AUROC for NLP performance in the testing set of 0.95 across all diagnoses with an F1 score of 0.57. For the 20 most common diagnoses, the AUROC was 0.97 with an F1 score of 0.72. Limitations: Single centered; sample size and use limited to research purposes.
We demonstrate an accurate and scalable NLP system to extract the primary diagnosis from free-text kidney biopsy reports, which can facilitate epidemiologic studies and identify patients for clinical trial recruitment.
肾活检报告为非索引文本格式，诊断需要耗费大量人力进行人工提取。自然语言处理（NLP）在肾活检诊断提取方面尚未经过严格测试。我们的目标是开发一种准确的模型，从自由文本报告中提取活检诊断。
使用NLP进行文本分类。
2016年6月至2023年12月期间，2666例患者有3042份原生肾活检报告，格式为便携式文档格式。
肾活检诊断。
NLP算法基于精度、召回率、F1分数和受试者工作特征曲线下面积（AUROC）对所有诊断以及20种最常见诊断的性能。
领域专家人工提取诊断，肾脏病理学家对随机子集（n = 200）进行验证。结构化查询语言服务器和Python将报告处理为机器可读的自由文本。我们使用来自Transformer的PubMed双向编码器表示来开发我们的NLP算法。我们将报告随机分为训练集（80%；n = 2434）和测试集（20%；n = 608）来训练NLP系统。我们进一步将测试集分为20%的验证集和80%的微调集。
中位年龄为57岁，50%为女性，29%为非裔美国人，23%为西班牙裔参与者。5种最常见的肾小球诊断为糖尿病肾病（23.7%）、局灶节段性肾小球硬化（15.5%）、狼疮性肾炎（9.7%）、免疫球蛋白A肾病（8.9%）和膜性肾病（7.2%）。评分者间可靠性的Cohen kappa系数为0.76。使用训练集进行微调的来自Transformer的PubMed双向编码器表示显示，在测试集中，NLP性能的平均AUROC在所有诊断中为0.95，F1分数为0.57。对于20种最常见的诊断，AUROC为0.97，F1分数为0.72。局限性：单中心；样本量和用途限于研究目的。
我们展示了一种准确且可扩展的NLP系统，可从自由文本肾活检报告中提取主要诊断，这有助于流行病学研究并识别适合临床试验招募的患者。
