Identifying patients with undiagnosed small intestinal neuroendocrine tumors using statistical and machine learning: Model development and validation study

#3981

Introduction: Diagnosis of small intestinal neuroendocrine tumors (SI-NETs) is often delayed due to non-specific symptoms and lack of awareness among primary and secondary care physicians. Earlier diagnosis may lead to improved clinical and patient outcomes. Clinical prediction models using machine learning could present novel opportunities for expedited diagnosis in primary care.

Aim(s): To develop and validate a model for identifying patients with undiagnosed SI-NETs in primary care using data from the Optimum Patient Care Research Database and the Clinical Practice Research Datalink.

Materials and methods: A cohort of adults contributing data to the Optimum Patient Care Research Database (01/01/2000-30/03/2023) was identified. This database collects de-identified data from general practices in the UK. Model development utilised statistical and machine learning methods: logistic regression, penalised regression, and XGBoost. Performance (discrimination and calibration) was assessed using internal-external cross-validation. Decision analysis curves compared clinical utility.

Conference:

Presenting Author:

Authors: Clift A, Mahon H, Khan G, Boardman-Pretty F, Worker A,

Keywords: machine learning, artificial intelligence, diagnosis, neuroendocrine tumor,