Identification of pancreatic cancer risk factors from clinical notes using natural language processing

• Published in: Pancreatology : official journal of the International Association of Pancreatology (IAP) ... [et al.] 2024-06, Vol.24 (4), p.572-578
• Main Authors: Sarwal, Dhruv, Wang, Liwei, Gandhi, Sonal, Sagheb Hossein Pour, Elham, Janssens, Laurens P., Delgado, Adriana M., Doering, Karen A., Mishra, Anup Kumar, Greenwood, Jason D., Liu, Hongfang, Majumder, Shounak
• Format: Article
• Language: English
• Subjects: Adult; Aged; Algorithms; Artificial intelligence; Cancer screening; Carcinoma, Pancreatic Ductal - diagnosis; Carcinoma, Pancreatic Ductal - genetics; Cohort Studies; Electronic Health Records; Female; Humans; Male; Middle Aged; Natural Language Processing; Pancreatic cancer; Pancreatic ductal adenocarcinoma; Pancreatic Neoplasms - diagnosis; Pancreatic Neoplasms - genetics; Risk Factors
• ISSN: 1424-3903; 1424-3911
• DOI: 10.1016/j.pan.2024.03.016

Screening for pancreatic ductal adenocarcinoma (PDAC) is considered in high-risk individuals (HRIs) with established PDAC risk factors, such as family history and germline mutations in PDAC susceptibility genes. Accurate assessment of risk factor status is provider knowledge-dependent and requires extensive manual chart review by experts. Natural Language Processing (NLP) has shown promise in automated data extraction from the electronic health record (EHR). We aimed to use NLP for automated extraction of PDAC risk factors from unstructured clinical notes in the EHR. We first developed rule-based NLP algorithms to extract PDAC risk factors at the document-level, using an annotated corpus of 2091 clinical notes. Next, we further improved the NLP algorithms using a cohort of 1138 patients through patient-level training, validation, and testing, with comparison against a pre-specified reference standard. To minimize false-negative results we prioritized algorithm recall. In the test set (n = 807), the NLP algorithms achieved a recall of 0.933, precision of 0.790, and F1-score of 0.856 for family history of PDAC. For germline genetic mutations, the algorithm had a high recall of 0.851, while precision and F1-score were lower at 0.350 and 0.496 respectively. Most false positives for germline mutations resulted from erroneous recognition of tissue mutations. Rule-based NLP algorithms applied to unstructured clinical notes are highly sensitive for automated identification of PDAC risk factors. Further validation in a large primary-care patient population is warranted to assess real-world utility in identifying HRIs for pancreatic cancer screening.