Artificial Intelligence: A Probability-
Based Algorithm for
Gastroenteropancreatic
Neuroendocrine Tumours (GEP-NETs)
Dale Srinivas*, Anu Singh
Twin Island University, Bushlot, Guyana
Email: krishna28us@[Link]
Abstract
Background: Gastroenteropancreatic neuroendocrine tumours (GEP-NETs) are
heterogeneous malignancies with variable presentation and prognosis. Probability-based
approaches using Bayesian updating principles can integrate clinical, biochemical, imaging,
and histological findings to guide timely diagnosis and management.
Objective: To present a Bayesian algorithm for the diagnostic evaluation and management
of GEP-NETs.
Methods: Prior probability is anchored using clinical risk factors and presentation.
Sequential likelihood updates are applied using biomarkers (chromogranin A, 5-HIAA),
imaging (CT, MRI, PET), and biopsy data. Decision thresholds inform further testing,
surgical planning, and treatment pathways.
Conclusion: Bayesian probability-based algorithms provide a structured and clinically
useful framework for diagnosing and managing GEP-NETs.
Introduction
Gastroenteropancreatic neuroendocrine tumours (GEP-NETs) represent a heterogeneous
group of neoplasms arising from the diffuse neuroendocrine system. Their incidence has
been steadily increasing due to improved imaging and heightened awareness. Clinical
presentations range from incidental radiological findings to classical hormone-secreting
syndromes. The diversity of manifestations poses diagnostic and therapeutic challenges.
Traditional pathways rely on clinical evaluation, biochemical markers, imaging, and
histology, but often lack integration. Bayesian updating provides a structured approach to
sequentially incorporate multiple data points, refining diagnostic certainty and guiding
management. This paper presents a probability-driven diagnostic and management
algorithm for GEP-NETs.
Methods
The algorithm was developed using a Bayesian framework, allowing clinicians to update
prior probability estimates with new diagnostic evidence. The steps include:
1. Define the diagnostic question (target event).
2. Anchor prior probability (symptoms, risk factors, prevalence).
3. Sequentially update using biomarkers, imaging, functional scans, and histology.
4. Determine posterior probability and map to decision thresholds.
5. Translate into treatment pathways (surgery, SSA, PRRT, targeted therapies,
chemotherapy).
Results
Worked Example: Bayesian Probability Update
Prior Probability Posterior after Positive PET Posterior after Negative
(LR+=10) PET (LR-=0.08)
5% 34.5% 0.4%
10% 52.6% 0.9%
20% 71.4% 1.6%
30% 81.1% 3.3%
Discussion
The Bayesian framework ensures that each diagnostic step meaningfully shifts probability
rather than being interpreted in isolation. For example, a positive 68Ga-DOTATATE PET/CT
markedly increases diagnostic certainty, justifying biopsy and staging, whereas a negative
scan in a low-prevalence setting reduces probability below action thresholds. This approach
aligns with ENETS/ESMO/NANETS guidelines and can be embedded into AI-driven clinical
decision support systems.
Conclusion
Bayesian probability-based algorithms provide a structured and clinically meaningful
approach for the diagnosis and management of GEP-NETs. By integrating sequential
diagnostic evidence, clinicians can enhance diagnostic accuracy, better stratify patients, and
optimise treatment planning. Future work should validate these frameworks in prospective
cohorts.
References
1. Pavel M, Öberg K, Falconi M, et al. Gastroenteropancreatic neuroendocrine neoplasms:
ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Annals of
Oncology. 2020;31(7):844–860.
2. Falconi M, Eriksson B, Kaltsas G, et al. ENETS Consensus Guidelines Update for the
Management of Patients with Functional and Non-Functional Pancreatic Neuroendocrine
Tumors. Neuroendocrinology. 2016;103(2):153–171.
3. Hope TA, Bergsland EK, Bozkurt MF, et al. Appropriate use criteria for somatostatin
receptor PET imaging in neuroendocrine tumors. Journal of Nuclear Medicine.
2018;59(1):66–74.
4. Sundin A, Arnold R, Baudin E, et al. ENETS Consensus Guidelines for Radiological, Nuclear
Medicine & Hybrid Imaging in Neuroendocrine Tumors. Neuroendocrinology.
2017;105(3):212–244.
5. Chan DL, Pavlakis N, Schembri GP, et al. Dual somatostatin receptor/FDG PET/CT imaging
in metastatic neuroendocrine tumors: prognostic utility. Journal of Nuclear Medicine.
2017;58(1):91–96.
6. Modlin IM, Moss SF, Chung DC, Jensen RT, Snyderwine E. Priorities for improving
management of GEP-NETs. JNCI. 2008;100(18):1282–1289.
7. Deppen SA, Liu E, Blume JD, et al. Safety and efficacy of 68Ga-DOTATATE PET/CT for
neuroendocrine tumors. Journal of Nuclear Medicine. 2016;57(5):708–714.
8. Spiegelhalter DJ, Myles JP, Jones DR, Abrams KR. An Introduction to Bayesian Methods in
Health Technology Assessment. BMJ. 1999;319(7208):508–512.