Despite significant advances in the diagnosis and treatment of acromegaly, diagnostic delay remains substantial, often exceeding 6-10 years, while a significant proportion of patients remain inadequately controlled despite multiple available therapeutic modalities. Artificial intelligence (AI) is a promising tool to address several challenges across the acromegaly care pathway, from early disease detection to individualized treatment selection and outcome prediction. AI-assisted facial and hand image analysis has demonstrated the potential to identify subtle phenotypic changes and reduce diagnostic delay, facilitating earlier recognition of the disease. In parallel, machine learning and deep learning approaches have been increasingly employed to predict surgical remission, response to medical therapy, and radiotherapy outcomes, often outperforming individual biomarkers. AI models further support the development of precision medicine strategies capable of improving patient stratification and therapeutic decision-making. This review summarizes current and emerging applications of AI in acromegaly, including facial recognition technologies, radiomics and machine learning predictive models integrating clinical, biochemical, radiological, and molecular data.

The incidental thyroid uptake of radiolabeled ligands used in positron emission tomography/computed tomography (PET/CT) targeting cell-surface proteins is observed in up to 6% of all scans. While diffuse thyroid uptake of ligands targeting somatostatin receptor type 2 (SSTR2), prostate-specific membrane antigen (PSMA), fibroblast activation protein (FAPI), or choline, is observed in normal thyroid or usually signifies a benign thyroid abnormality, a heterogenous or focal uptake is associated with malignant lesions in up to 19.7% of patients. The PET/CTs utilizing these ligands are characterized by decreased diagnostic sensitivity for detection of thyroid cancer metastases than cross-sectional imaging or 18-fluorodeoxyglucose (18FDG)-PET/CT but are highly specific for detection of a subset of thyroid cancers characterized by a high target protein expression. Increased uptake of radiolabeled ligands targeting SSTR2, PSMA, FAPI, integrin αvβ3 and minigastrin can be used for patients' selection for peptide receptor radionuclide therapy (PRRT). PRRT can be applied in clinical trials and/or off label for patients with advanced radioactive iodine-refractory thyroid cancer who have limited treatment options or for medullary thyroid carcinoma of neuroendocrine origin. The PRRT-induced disease control rate approaches 60% and PRRT is well tolerated with limited side effects. Efficacy could be improved with careful patients' selection based on the expression of molecular targets and their function, quantified by standard uptake values (SUVmax) in diagnostic PET/CT imaging. Adopting an individualized dosimetry-based approach and synthesizing ligands with better tumor retention and residence time may help optimize treatment response.