Deep learning is emerging as a transformative tool for chronic pain research. In this systematic review, we evaluated deep learning applications across diverse data modalities (neuroimaging, electrophysiological signals, motion capture, and wearable sensors) for chronic pain diagnosis, classification, and prognostication. A comprehensive search of seven electronic databases identified 22 eligible studies using convolutional neural networks, recurrent neural networks, and hybrid models. Study quality was appraised using PROBAST and PRISMA guidelines. While these studies demonstrate the potential to overcome limitations of subjective pain assessments, heterogeneity in methodologies, sample sizes, and evaluation metrics precluded meta-analysis. Key challenges include data imbalance, limited external validation, and variability in preprocessing approaches. Despite these limitations, deep learning generates non-invasive biomarkers and supports precision medicine strategies in chronic pain management, potentially improving clinical decision-making and reducing the societal burden of chronic pain.
The COVID-19 pandemic has underscored the necessity for advanced diagnostic tools in global health systems. Infrared Thermography (IRT) has proven to be a crucial non-contact method for measuring body temperature, vital for identifying febrile conditions associated with infectious diseases like COVID-19. Traditional non-contact infrared thermometers (NCITs) often exhibit significant variability in readings. To address this, we integrated machine learning algorithms with IRT to enhance the accuracy and reliability of temperature measurements. Our study systematically evaluated various regression models using heuristic feature engineering techniques, focusing on features' physiological relevance and statistical significance. The Convolutional Neural Network (CNN) model, utilizing these techniques, achieved the lowest RMSE of 0.2223, demonstrating superior performance compared to results reported in previous literature. Among non-neural network models, the Binning method achieved the best performance with an RMSE of 0.2296. Our findings highlight the potential of combining advanced feature engineering with machine learning to improve diagnostic tools' effectiveness, with implications extending to other non-contact or remote sensing biomedical applications. This paper offers a comprehensive analysis of these methodologies, providing a foundation for future research in the field of non-invasive medical diagnostics.