Machine learning model: Predicting prognosis in neuroendocrine tumors

#4076

Introduction: Neuroendocrine tumors (NETs) have a heterogeneous clinical course. Even with similar Ki-67 proliferative index/morphology, the course varies in patients with similar pathological characterization.

Aim(s): A novel machine learning model for prognostication in NETs.

Materials and methods: 30 patients with well-differentiated NET with 2 time point 68Ga- DOTATATE PET/CT scanning were analyzed using TRAQinform IQ (AIQ Solutions) technology. 18/30= female and 12/30=male. 8/30 had a pancreatic primary, 16/30 patients had a small bowel, and 6 patients had other primaries. 18/30 had G1 tumors ,12/30 had G2 tumors. The TRAQinform Profile was calculated to predict prognosis using a random survival forest previously trained on data from a study of 25 patients. The model performance was evaluated using the c-index.

Conference:

Presenting Author:

Authors: Varghese D, Naimian A, Yazdian P, Lokre O, Perk T,

Keywords: Neuroendocrine tumor, artificial intelligence, DOTATATE scan,