AI in Medical Applications

The database compiled for this study is a comprehensive and meticulously curated repository designed to evaluate the efficacy of anti-VEGF therapy in patients with Diabetic Macular Edema (DME). It includes clinical and imaging data from 193 diabetic patients, aged 18-70 years, who participated in a single-center, randomized, parallelgroup, double-masked clinical trial. The database encompasses detailed demographic and clinical information, such as age, gender, medical history, duration of diabetes, and baseline measurements like blood pressure and intraocular pressure.

We present the SynSUM benchmark, a synthetic dataset linking unstructured clinical notes to structured background variables. The dataset consists of 10,000 artificial patient records containing tabular variables (like symptoms, diagnoses and underlying conditions) and associated clinical notes describing the fictional patient encounter in the domain of respiratory diseases.

This dataset comprises a comprehensive analysis of state-of-the-art techniques and systems for seizure detection and classification, based on various papers and studies. It integrates detailed metadata on publications, including their year, methodologies, seizure types (both ILAE-2017 and paper-specific), datasets, and biomarker utilization. The dataset also provides performance metrics such as accuracy, sensitivity, specificity, false-positive rates, and AUC-ROC values, alongside additional technical details about machine learning models, feature extraction techniques, and biomarkers.