Researchers have introduced Breath-Dx, a new benchmark for evaluating time-series classification models used in disease screening with electronic nose (eNose) technology. The benchmark utilizes the Scent of Health (S-OH) dataset, which includes data from 1,234 patients across nine diagnostic groups, and accounts for sensor drift, a critical real-world failure mode. Evaluations showed that all tested model families experienced a significant drop in performance under realistic temporal validation, with recurrent architectures demonstrating the most consistent results. AI
IMPACT This benchmark aims to improve the reliability and comparability of AI models for disease screening using eNose technology, addressing real-world challenges like sensor drift.
RANK_REASON The cluster describes a new benchmark and dataset for evaluating machine learning models in a specific domain, which falls under research. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →