Closing the loop between environment, brain and mental health: how far we might go in real-life assessments?

PURPOSE OF REVIEW: Environmental factors such as climate, urbanicity, and exposure to nature are becoming increasingly important influencers of mental health. Incorporating data gathered from real-life contexts holds promise to substantially enhance laboratory experiments by providing a more comprehensive understanding of everyday behaviors in natural environments. We provide an up-to-date review of current technological and methodological developments in mental health assessments, neuroimaging and environmental sensing. RECENT FINDINGS: Mental health research progressed in recent years towards integrating tools, such as smartphone based mental health assessments or mobile neuroimaging, allowing just-in-time daily assessments. Moreover, they are increasingly enriched by dynamic measurements of the environment, which are already being integrated with mental health assessments. To ensure ecological validity and accuracy it is crucial to capture environmental data with a high spatio-temporal granularity. Simultaneously, as a supplement to experimentally controlled conditions, there is a need for a better understanding of cognition in daily life, particularly regarding our brain's responses in natural settings. SUMMARY: The presented overview on the developments and feasibility of "real-life" approaches for mental health and brain research and their potential to identify relationships along the mental health-environment-brain axis informs strategies for real-life individual and dynamic assessments.

Deep transfer learning compared to subject-specific models for sEMG decoders.

Objective. Accurate decoding of surface electromyography (sEMG) is pivotal for muscle-to-machine-interfaces and their application e.g. rehabilitation therapy. sEMG signals have high inter-subject variability, due to various factors, including skin thickness, body fat percentage, and electrode placement. Deep learning algorithms require long training time and tend to overfit if only few samples are available. In this study, we aim to investigate methods to calibrate deep learning models to a new user when only a limited amount of training data is available.Approach. Two methods are commonly used in the literature, subject-specific modeling and transfer learning. In this study, we investigate the effectiveness of transfer learning using weight initialization for recalibration of two different pretrained deep learning models on new subjects data and compare their performance to subject-specific models. We evaluate two models on three publicly available databases (non invasive adaptive prosthetics database 2-4) and compare the performance of both calibration schemes in terms of accuracy, required training data, and calibration time.Main results. On average over all settings, our transfer learning approach improves 5%-points on the pretrained models without fine-tuning, and 12%-points on the subject-specific models, while being trained for 22% fewer epochs on average. Our results indicate that transfer learning enables faster learning on fewer training samples than user-specific models.Significance. To the best of our knowledge, this is the first comparison of subject-specific modeling and transfer learning. These approaches are ubiquitously used in the field of sEMG decoding. But the lack of comparative studies until now made it difficult for scientists to assess appropriate calibration schemes. Our results guide engineers evaluating similar use cases.