Gallstone disease (GD) is a common gastrointestinal disease. Although traditional diagnostic techniques, such as ultrasonography, CT, and MRI, detect gallstones, they have some limitations, including high cost and potential inaccuracies in certain populations. This study proposes a machine learning-based prediction model for gallstone disease using bioimpedance and laboratory data. A dataset of 319 samples, comprising161 gallstone patients and 158 healthy controls, was curated. The dataset comprised 38 attributes of the participants, including age, weight, height, blood test results, and bioimpedance data, and it contributed to the literature on gallstones as a new dataset. State-of-the-art machine learning techniques were performed on the dataset to detect gallstones. The experimental results showed that vitamin D, C-reactive protein (CRP) level, total body water, and lean mass are crucial features, and the gradient boosting technique achieved the highest accuracy (85.42%) in predicting gallstones. The proposed technique offers a viable alternative to conventional imaging techniques for early prediction of gallstone disease.