Assessments of Lung Nodules by an Artificial Intelligence Chatbot Using Longitudinal Computed Tomography Images

This open-access dataset supports the study "Assessments of Lung Nodules by an Artificial Intelligence Chatbot Using Longitudinal Computed Tomography Images," allowing readers to reproduce the findings. The dataset comprises follow-up CT scan videos from patients within the National Lung Screening Trial (NLST) dataset and the local dataset. Each patient's videos are labeled as "1.MP4," "2.MP4," and "3.MP4." The compressed file "NLST_100_patients_follow-up_CT_mp4.zip" contains these videos of patients from the NLST dataset formatted for analysis with the GPT-4o. The compressed file "Local_training_20_patients_follow-up_CT_mp4.zip" contains these videos of patients from the local dataset formatted for analysis with the GPT-4o. Patient information, including malignancy status, nodule size, and location, is provided in the "NLST_100_patients_follow-up_information.csv" and the "Local_training_20_patients_follow-up_information.xlsx" files. The NLST Excel includes the original NLST radiologist assessments for margin (labeled as "Feature_margin_CT_report") and predominant attenuation (labeled as "Feature_Predominant attenuation_CT_report"). GPT-4o predictions for these features are listed in the columns "Feature_margin_GPT_report" and "Feature_Predominant attenuation_GPT_report," respectively.