Abstract
Indeterminate pulmonary nodules are a frequent and challenging finding in both screening and incidental imaging. Existing clinical prediction models provide structured estimates of malignancy risk but remain limited in precision, particularly for patients with intermediate pre-test probability. This technical report proposes the Pilli Kai Score, a digital twin framework that integrates clinical variables, radiomic features, blood-based biomarkers, and positron emission tomography (PET) data into a unified probability estimate for malignancy. The framework outlines a multi-modal modeling strategy incorporating validated clinical predictors, standardized radiomics, biomarkers evaluated in pulmonary nodule populations, and PET categories when available. Prespecified validation targets include strong calibration across risk strata, an area under the receiver operating characteristic (ROC) curve exceeding 0.85, and a high negative predictive value to safely defer invasive procedures in benign disease, with comparative evaluation against established clinical models. No patient-level data are analyzed; instead, illustrative figures present the proposed workflow and anticipated performance benchmarks. If validated in multi-center studies, this framework could improve diagnostic accuracy, reduce avoidable interventions, and enable more personalized lung cancer care pathways.