← Back to all news
arXiv Machine Learning By Padmini Krishnadas, Urs Hackstein, Alen Bosnjakovic, Philip J. Aston

Fuzzy Accuracy Compensates for Label Subjectivity in Classification of Skin Tone Using Wearable Photoplethysmography Signals

Read the original on arXiv Machine Learning →

The study investigates classifying Fitzpatrick skin tones from wearable photoplethysmography (PPG) signals, noting that traditional accuracy is low (40‑55 %) due to subjective labeling. By introducing a fuzzy accuracy metric—treating predictions within one class of the label as correct—accuracy rises dramatically, reaching up to 96 % on tree‑based models and 85‑87 % on deep learning and feature‑based approaches. The results suggest that PPG signals contain discernible skin‑tone information, especially when evaluated with the fuzzy metric.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Machine Learning
Jul 23

Good Practice Guide for quantifying uncertainties for machine learning models applied to photoplethysmography signals

arXiv:2607. 19999v1 Announce Type: new Abstract: This Good Practice Guide presents work done in the QUMPHY project (Uncertainty quantification for machine learning models applied to photoplethysmography signals) that considered both machine learning and uncertainty quantification for problems which used photoplethysmography (PPG) signals from wearable devices as input.

By P. Harris, C. Bench, M. Rinkevi\v{c}ius, V. Marozas, L. Coquelin, A. Thompson, M. Nandi, U. Hackstein, P. J. Aston
benchmarks
More like this →
arXiv AI
Sep 3

Disease Burden over Skin Tone: Decomposing the Dermatology-AI Generalization Gap

The study examines why dermatology AI models, largely trained on light‑skinned, cancer‑focused images, perform poorly when applied to diverse patient populations. By comparing a cancer‑trained baseline, two dermatology foundation models, and a general‑purpose vision model on tone‑stratified and disease‑shifted datasets, the authors find that disease‑distribution shift, rather than skin‑tone underrepresentation, is the primary cause of generalization failure. Representation analysis shows that cancer‑specialized features lack transferable structure, while dermatology‑pretrained features maintain stronger clustering, and lightweight adaptation with about ten labeled examples per category can recover most performance.

By Nirajan Kunwor, Sanjaya Poudel, Quoc-Huy Trinh, Jahidul Arafat, Sunil Kumar Gaire
fine-tuning
More like this →
arXiv AI
Sep 25

The Entropy Triangle Method (ETM): A novel framework for the prevention of cardiac arrhythmia with a review of more than 10,000 patients

The Entropy Triangle Method (ETM) is a new framework designed to predict cardiac arrhythmias. It combines feature engineering, entropy triangle oversampling, and disease prediction, and was tested on a 12‑lead ECG database of 10,646 patients with 11 heart rhythm categories. The study reports that the method achieves over 85% accuracy for non‑sinus rhythms, with support vector classifiers and a novel oversampling technique termed "shark scent" yielding the best performance.

By Arman daliri
More like this →