PARE: A framework for removal of confounding effects from any distance-based dimension reduction method.

Dimension reduction tools preserving similarity and graph structure such as t-SNE and UMAP can capture complex biological patterns in high-dimensional data. However, these tools typically are not designed to separate effects of interest from unwanted effects due to confounders. We introduce the partial embedding (PARE) framework, which enables removal of confounders from any distance-based dimension reduction method. We then develop partial t-SNE and partial UMAP and apply these methods to genomic and neuroimaging data. For lower-dimensional visualization, our results show that the PARE framework can remove batch effects in single-cell sequencing data as well as separate clinical and technical variability in neuroimaging measures. We demonstrate that the PARE framework extends dimension reduction methods to highlight biological patterns of interest while effectively removing confounding effects.

Federated learning for privacy-preserving depression detection with multilingual language models in social media posts.

The incidences of mental health illnesses, such as suicidal ideation and depression, are increasing, which highlights the urgent need for early detection methods. There is a growing interest in using natural language processing (NLP) models to analyze textual data from patients, but accessing patients' data for research purposes can be challenging due to privacy concerns. Federated learning (FL) is a promising approach that can balance the need for centralized learning with data ownership sensitivity. In this study, we examine the effectiveness of FL models in detecting depression by using a simulated multilingual dataset. We analyzed social media posts in five different languages with varying sample sizes. Our findings indicate that FL achieves strong performance in most cases while maintaining clients' privacy for both independent and non-independent client partitioning.

U-Rate-UE; Measuring Perceived Recovery of the Affected Upper Extremity in Adults Post-stroke.

OBJECTIVE: To establish initial validity of "U-Rate-UE", a single-question scale regarding perceived recovery of the stroke affected upper extremity (UE). DESIGN: A retrospective longitudinal study of data collected at rehabilitation admission, 6 weeks, and 6 months since stroke. SETTING: Stroke rehabilitation and community-based. PARTICIPANTS: A convenience sample of 87 individuals, median (interquartile range) age 71.5 (65-80) years, 15.0 (12-20) days post-stroke. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: The affected UE was assessed using the Fugl-Meyer Motor Assessment, grip strength, Action Research Arm Test, the Box and Block Test (BBT), and The Rating of Everyday Arm-Use in the Community and Home. Participants also rated how much they perceive that their affected UE recovered from the stroke using U-Rate-UE; 0-100 (no to full recovery). Longitudinal changes in U-Rate-UE ratings were assessed. In addition, at 6 weeks and 6 months post-stroke, the change in BBT was calculated and participants were grouped into achieved/did not achieve the minimal detectable change (MDC). Correlations between U-Rate-UE to the other UE assessments were assessed at all 3 timepoints. RESULTS: Significant changes in U-Rate-UE were seen over time (P<.05). At 6 weeks and 6 months, participants who achieved BBT-MDC rated their recovery significantly higher than participants who did not. U-Rate-UE was moderately-strongly significantly correlated to UE assessments (rho=.61-.85, P<.001). CONCLUSIONS: The U-Rate-UE is supported for use with UE assessments contributing to comprehensive clinical understanding of the recovery of the affected UE in adults post-stroke.