Assessing clinician performance using a multi-modality clinical decision-support system for lung cancer prognostication

Surgery is the primary treatment for early-stage lung cancer. Adjuvant therapy is offered to patients who are at a high risk of recurrence, however, determining the patients that would benefit from additional therapy is often difficult. In this study, we aimed to develop a clinical decision support system (CDSS) for post-surgery lung cancer prognostication integrating a multi-modality deep learning model (DLM). Pre-operative medical images and clinical, surgical, and pathological information were fed into an externally validated DLM. A CDSS was then developed to display the patient information and DLM results for potential clinical use. Four oncologists evaluated each patient's recurrence probability, their confidence level, and their post-surgery recommendations both with and without the DLM information. The CDSS DLM information demonstrated the potential to improve user prediction performance and confidence. This exploratory study is the first to integrate a multi-modality DLM for prognostication with a CDSS, as well as the first study of clinician attitudes towards CDSSs for lung cancer.