Multi-task learning (MTL) aims to improve generalization performance by learning multiple related tasks simultaneously. While sometimes the underlying task relationship structure is known, often the structure needs to be estimated from data at hand. In this paper, we present a novel family of models for MTL, applicable to regression and classification problems, capable of learning the structure of task relationships. In particular, we consider a joint estimation problem of the task relationship structure and the individual task parameters, which is solved using alternating minimization. The task relationship structure learning component builds on recent advances in structure learning of Gaussian graphical models based on sparse estimators of the precision (inverse covariance) matrix. We illustrate the effectiveness of the proposed model on a variety of synthetic and benchmark datasets for regression and classification. We also consider the problem of combining climate model outputs for better projections of future climate, with focus on temperature in South America, and show that the proposed model outperforms several existing methods for the problem.

No abstract

No abstract

Family home visiting is a widely accepted strategy used with disadvantaged families to mitigate the effects of poverty. However, gaps persist in knowledge of effective intervention approaches for home visiting relative to specific client risks such as parenting and psychosocial problems. The purpose of this study was to inductively create clusters from electronic health records of 484 public health nursing clients, using client characteristics and intervention data. Four clinically relevant client clusters were generated using Mixed Membership Naïve Bayes methods. Fourteen distinct intervention clusters were generated using KMETIS, a graph partitioning method. The content of the intervention clusters illustrates the complexity of public health nursing practice. This study leverages current nursing documentation technology capacity to advance nursing knowledge. Future research is needed to explore relationships between client and intervention clusters and their associations with client outcomes, with the end goals of improving home visiting practice and client outcomes.

The ability of brain tissue to regenerate is limited; therefore, brain diseases (i.e., trauma, stroke, tumors) often lead to irreversible motor and cognitive impairments. Therapeutic interventions using various types of injectable biomaterials have been investigated to promote endogenous neural differentiation. Despite promising results in pre-clinical studies, the translation of regenerative medicine to the clinic has many challenges due to the lack of reliable imaging systems to achieve accurate evaluation of the treatment efficacy. Methods : In this study, we developed a dual-channel fluorescence imaging technique to simultaneously monitor tissue ingrowth and scaffold disintegration. Enzymatically crosslinked gelatin-hyaluronic acid hydrogel was labeled with 800 nm fluorophore, ZW800-3a, while the regenerated tissue was highlighted with 700 nm brain-specific contrast agent, Ox1. Results : Using the multichannel fluorescence imaging system, tissue growth and degradation of the NIR hydrogel were simultaneously imaged in the brain of mice. Images were further analyzed and reconstructed to show both visual and quantitative information of each stage of a therapeutic period. Conclusion : Dual-channel in vivo imaging systems can provide highly accurate visual and quantitative information of the brain tissue ingrowth for the evaluation of the therapeutic effect of NIR hydrogel through a simple and fast operating procedure.