What is the clinical evidence from this study?

Study design: Other. Population: Stress (n=15). Intervention: STFT-IG framework with Random Forest classifier vs. Other classifiers (DT, SVM, MLP) and feature selection methods. Primary outcome: Subject-independent stress prediction accuracy (LOSO cross-validation).

What question did this study set out to answer?

The aim is to improve stress prediction by selecting the most informative features from multimodal physiological signals using a supervised approach.

March 7, 2026Open Access

Supervised information gain-based feature selection for multimodal physiological signals in stress prediction

Q: What does this research mean for the field?

A supervised, Information Gain-based feature selection framework (STFT-IG) combined with a Random Forest classifier achieves a peak subject-independent stress prediction accuracy of 90.01% using multimodal physiological signals. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Key Result

The STFT-IG framework combined with a Random Forest classifier achieved a peak subject-independent stress prediction accuracy of 90.01% using multimodal physiological signals.

Key Points

The aim is to improve stress prediction by selecting the most informative features from multimodal physiological signals using a supervised approach.
Proposed a supervised Information Gain-based feature selection framework for physiological datasets.
Extracted time-frequency representations of features using Short-Time Fourier Transform from wrist-worn sensors.
Applied Random Forest classifier on selected features, evaluated using the Leave-One-Subject-Out protocol.
The STFT-IG framework achieved competitive performance in stress recognition.
Comparative analyses confirmed the effectiveness of the proposed approach over alternative methods.

Structured PICO

Population

15 subjects from the WESAD dataset (controlled laboratory study) with multimodal physiological signals recorded using wrist-worn and chest-worn wearables.

Intervention

Supervised Information Gain (IG)-based feature selection framework using Short-Time Fourier Transform (STFT) and Random Forest classifier

Comparator

Alternative feature selection methods (ANOVA F-test, Chi-square test, Pearson correlation, CCA) and other classifiers (Decision Trees, Support Vector Machines, Multilayer Perceptrons)

Outcome

Stress prediction performance (Precision, Recall, F1 score, and accuracy) under subject-independent evaluation using the Leave-One-Subject-Out (LOSO) protocol

A supervised Information Gain-based feature selection framework using STFT for multimodal physiological signals provides an efficient and interpretable pipeline for stress recognition with competitive subject-independent performance.

Limitations

Substantial performance drop on synthetically generated data (accuracy 49.67%) due to mismatches in feature statistics
Model learns dataset-specific features rather than generic stress features, limiting generalization to unseen, noisy real-world patterns

Abstract

Feature selection plays a critical role in identifying the most informative signals across multiple modalities in physiological datasets. While correlation-based approaches have traditionally been used to discover relationships among features, they are often unsupervised and do not incorporate class label information, limiting their discriminative power in supervised classification tasks. In this work, we propose a supervised, Information Gain (IG)-based feature selection framework for multimodal data acquired from wrist-worn sensors, including Blood Volume Pulse (BVP), Acceleration (ACC), Skin Temperature (TEMP), and Electrodermal Activity (EDA).Time–frequency feature representations are extracted from each modality using Short-Time Fourier Transform (STFT) and subsequently ranked according to their correlation with class labels using Information Gain(IG) criterion. The top-ranked features for each modality are then fused into a compact, high-quality representation for classification using Random Forest classifier. Experiments on the WESAD stress recognition dataset demonstrate that our proposed STFT-IG framework along with Random Forest classifier achieves competitive performance under subject-independent evaluation using the Leave-One-Subject-Out (LOSO) protocol. Furthermore, comparative analyses with alternative feature selection methods and feature representations are performed to contextualize the effectiveness of the proposed approach.

Bookmark

View Full Paper