Digital microfluidic biochips have emerged as a popular alternative for laboratory experiments. Pin-count reduction and cross-contamination avoidance are key design considerations for practical applications with different droplets being transported and manipulated on highly integrated biochips. This paper presents the first design automation flow that considers the cross-contamination problems on pin-constrained biochips. The factors that make the problems harder on pin-constrained biochips are explored. To cope with these cross contaminations, this paper proposes: 1) early crossing minimization algorithms during placement, and 2) systematic wash droplet scheduling and routing that require only one extra control pin and zero assay completion time overhead for practical bioassays. Experimental results show the effectiveness and scalability of our algorithms for practical bioassays.

Digital microfluidic biochips have emerged as a popular alternative for laboratory experiments. Pin-count reduction and cross-contamination avoidance are key design considerations for practical applications with different droplets being transported and manipulated on highly integrated biochips. We present in this paper the first design automation flow that considers the cross-contamination problems on pin-constrained biochips. We explore the factors that make the problems harder on pin-constrained biochips. To cope with these cross contaminations, we propose (1) early crossing minimization algorithms during placement, and (2) systematic wash droplet scheduling and routing that require only one extra control pin and zero assay completion time overhead for practical bioassays. Experimental results show the effectiveness and scalability of our algorithms for practical bioassays.

Stroke is the third leading cause of death and the principal cause of serious long-term disability in the United States. Accurate prediction of stroke is highly valuable for early intervention and treatment. In this study, we compare the Cox proportional hazards model with a machine learning approach for stroke prediction on the Cardiovascular Health Study (CHS) dataset. Specifically, we consider the common problems of data imputation, feature selection, and prediction in medical datasets. We propose a novel automatic feature selection algorithm that selects robust features based on our proposed heuristic: conservative mean. Combined with Support Vector Machines (SVMs), our proposed feature selection algorithm achieves a greater area under the ROC curve (AUC) as compared to the Cox proportional hazards model and L1 regularized Cox feature selection algorithm. Furthermore, we present a margin-based censored regression algorithm that combines the concept of margin-based classifiers with censored regression to achieve a better concordance index than the Cox model. Overall, our approach outperforms the current state-of-the-art in both metrics of AUC and concordance index. In addition, our work has also identified potential risk factors that have not been discovered by traditional approaches. Our method can be applied to clinical prediction of other diseases, where missing data are common and risk factors are not well understood.

Digital microfluidic biochips have emerged as a popular alternative for laboratory experiments. To make the biochip feasible for practical applications, pin-count reduction is a key problem to higher-level integration of reactions on a biochip. Most previous works approach the problem by post-processing the placement and routing solutions to share compatible control signals; however, the quality of such sharing algorithms is inevitably limited by the placement and routing solutions. We present in this paper a comprehensive pin-constrained biochip design flow that addresses the pin-count issue at all design stages. The proposed flow consists of three major stages: 1) pin-count aware stage assignment that partitions the reactions in the given bioassay into execution stages; 2) pin-count aware device assignment that determines a specific device used for each reaction; and 3) guided placement, routing, and pin assignment that utilize the pin-count saving properties from the stage and device assignments to optimize the assay time and pin-count. For both the stage and device assignments, basic integer linear programming formulations and effective solution-space reduction schemes are proposed to minimize the assay time and pin-count. Experimental results show the efficiency of our methods and a 55-57% pin-count reduction over the state-of-the-art algorithms/flow.

Digital microfluidic biochips have emerged as a popular alternative for laboratory experiments. To make the biochip feasible for practical applications, pin-count reduction is a key problem to higher-level integration of reactions on a biochip. Most previous works approach the problem by post-processing the placement and routing solutions to share compatible control signals; however, the quality of such sharing algorithms is inevitably limited by the placement and routing solutions. We present in this paper a comprehensive pin-constrained biochip design flow that addresses the pin-count issue at all design stages. The proposed flow consists of three major stages: (1) pin-count aware stage assignment that partitions the reactions in the given bioassay into execution stages, (2) pin-count aware device assignment that determines a specific device used for each reaction, and (3) guided placement, routing, and pin assignment that utilize the pin-count saving properties from the stage and device assignments to optimize the assay time and pin count. For both the stage and device assignments, exact ILP formulations and effective solution-space reduction schemes are proposed to minimize the assay time and pin count. Experimental results show the efficiency of our algorithms/flow and a 55--57% pin-count reduction over the state-of-the-art algorithms/flow.