Artificial intelligence has been extensively applied in healthcare services recently, and clinical decision support systems driven by artificial intelligence are one of the applications. Early-stage clinical evaluation of artificial intelligence (AI)-based clinical decision support systems lies between preclinical development (in silico), offline validation, and large-scale trials, but few AI-related clinical studies have addressed human factors evaluations and reported the implementation environment, user characteristics, selection process and algorithm identification of AI systems. In order to bridge the development-to-implementation gap in clinical artificial intelligence and to promote the transparent and standardized reporting of early-stage clinical studies of AI-based decision support systems. A reporting guideline for the developmental and exploratory clinical investigations of decision support systems driven by artificial intelligence (DECIDE-AI) was published in 2022. This paper aimed to interpret the background, development process and key items of the DECIDE-AI guideline and promote its understanding as well as dissemination in China.
Objective To evaluate the survival outcomes of segmentectomy versus lobectomy for T1c non-small cell lung cancer (NSCLC). Methods We searched PubMed, EMbase, Cochrane Central Register of Controlled Trials (CENTRAL), CNKI (China National Knowledge Infrastructure), and Wanfang Data, with the search time limit set from the inception of the databases to February 2024. Three researchers independently screened the literature, extracted relevant information, and evaluated the risk of bias of the included literature according to the Newcastle-Ottawa Scale (NOS). Meta-analysis was conducted using STATA 15.1. Results A total of 8 retrospective cohort studies were included, involving 7 433 patients. The NOS scores of the included studies were all ≥7 points. Patients who underwent lobectomy had significantly higher five-year overall survival (OS) rates compared to those who underwent segmentectomy (adjusted HR=1.11, 95%CI 0.99-1.24, P=0.042). Compared with lobectomy, segmentectomy showed no significant difference in adjusted three-year OS rate (adjusted HR=0.88, 95%CI 0.62-1.24) and adjusted five-year lung cancer-specific survival (adjusted HR=1.10, 95%CI 0.80-1.51, P=0.556) of patients with T1c NSCLC. Moreover, there were no differences in the five-year adjusted relapse-free survival (adjusted HR=1.23, 95%CI 0.82-1.85, P=0.319), and adverse events (OR=0.57, 95%CI 0.37-0.90, P=0.015) in the segmentectomy group were significantly less than those in the lobectomy group. Subgroup analysis based on whether patients received neoadjuvant therapy showed that among studies that excluded patients who received neoadjuvant therapy, no significant difference in 5-year adjusted OS rate was observed between the segmentectomy group and lobectomy group (adjusted HR=1.02, 95%CI 0.81-1.28, P=0.870). Conclusion Segmentectomy and lobectomy show no significant difference in long-term survival in stage T1c NSCLC patients, with segmentectomy associated with fewer postoperative complications. Further high-quality research is needed to confirm the comparative efficacy and safety of lobectomy and segmentectomy for T1c NSCLC patients.