The development of digital instruments for mental health monitoring using biosensor data from wearable devices can enable remote, longitudinal, and objective quantitative benchmarks. To survey developments and trends in this field, we conducted a systematic review of artificial intelligence (AI) models using data from wearable biosensors to predict mental health conditions and symptoms. Following PRISMA guidelines, we identified 48 studies using a variety of wearable and smartphone biosensors including heart rate, heart rate variability (HRV), electrodermal activity/galvanic skin response (EDA/GSR), and digital proxies for biosignals such as accelerometry, location, audio, and usage metadata. We observed several technical and methodological challenges across studies in this field, including lack of ecological validity, data heterogeneity, small sample sizes, and battery drainage issues. We outline several corresponding opportunities for advancement in the field of AI-driven biosensing for mental health.
Dementia is a progressive neurological disorder that affects the daily lives of older adults, impacting their verbal communication and cognitive function. Early diagnosis is important to enhance the lifespan and quality of life for affected individuals. Despite its importance, diagnosing dementia is a complex process. Automated machine learning solutions involving multiple types of data have the potential to improve the process of automated dementia screening. In this study, we build deep learning models to classify dementia cases from controls using the Pitt Cookie Theft dataset from DementiaBank, a database of short participant responses to the structured task of describing a picture of a cookie theft. We fine-tune Wav2vec and Word2vec baseline models to make binary predictions of dementia from audio recordings and text transcripts, respectively. We conduct experiments with four versions of the dataset: (1) the original data, (2) the data with short sentences removed, (3) text-based augmentation of the original data, and (4) text-based augmentation of the data with short sentences removed. Our results indicate that synonym-based text data augmentation generally enhances the performance of models that incorporate the text modality. Without data augmentation, models using the text modality achieve around 60% accuracy and 70% AUROC scores, and with data augmentation, the models achieve around 80% accuracy and 90% AUROC scores. We do not observe significant improvements in performance with the addition of audio or timestamp information into the model. We include a qualitative error analysis of the sentences that are misclassified under each study condition. This study provides preliminary insights into the effects of both text-based data augmentation and multimodal deep learning for automated dementia classification.
Dementia is a progressive neurological disorder that profoundly affects the daily lives of older adults, impairing abilities such as verbal communication and cognitive function. Early diagnosis is essential for enhancing both lifespan and quality of life for affected individuals. Despite its importance, diagnosing dementia is complex and often necessitates a multimodal approach incorporating diverse clinical data types. In this study, we fine-tune Wav2vec and Word2vec baseline models using two distinct data types: audio recordings and text transcripts. We experiment with four conditions: original datasets versus datasets purged of short sentences, each with and without data augmentation. Our results indicate that synonym-based text data augmentation generally enhances model performance, underscoring the importance of data volume for achieving generalizable performance. Additionally, models trained on text data frequently excel and can further improve the performance of other modalities when combined. Audio and timestamp data sometimes offer marginal improvements. We provide a qualitative error analysis of the sentence archetypes that tend to be misclassified under each condition, providing insights into the effects of altering data modality and augmentation decisions.
Dementia is a progressive neurological disorder that profoundly affects the daily lives of older adults, impairing abilities such as verbal communication and cognitive function. Early diagnosis is essential for enhancing both lifespan and quality of life for affected individuals. Despite its importance, diagnosing dementia is complex and often necessitates a multimodal approach incorporating diverse clinical data types. In this study, we fine-tune Wav2vec and Word2vec baseline models using two distinct data types: audio recordings and text transcripts. We experiment with four conditions: original datasets versus datasets purged of short sentences, each with and without data augmentation. Our results indicate that synonym-based text data augmentation generally enhances model performance, underscoring the importance of data volume for achieving generalizable performance. Additionally, models trained on text data frequently excel and can further improve the performance of other modalities when combined. Audio and timestamp data sometimes offer marginal improvements. We provide a qualitative error analysis of the sentence archetypes that tend to be misclassified under each condition, providing insights into the effects of altering data modality and augmentation decisions.