CISE Research Instrumentation: A Distributed Biomedical Computing Laboratory

9617349 Bic, Lubomir， F. Michael Dillencourt加州大学欧文分校CISE研究仪器：分布式生物医学计算实验室生物医学计算的许多重要问题领域都有能力生成大量相对粗粒度的计算，因此可以从基于局域网的工作站网络中受益匪浅。该项目将提供这样一个基础设施，基于Sun工作站集群，它将由以下四个项目共享：临床数据库中的知识发现-脑网络模型-毒理学模拟-分布式蛋白质结构预测。该项目的目标是在临床数据库中发现对诊断和治疗有用的新知识。它使用新颖的机器学习方法，特别适合医学数据库。第二个项目试图了解特定大脑区域的功能。它将开发基于神经网络算法的并行版本，以解决困难的信号处理任务。第三个项目开发了一个分布式模拟环境，支持模拟模型的构建和交互使用，用于预测人体内化学品（药物或毒素）的浓度，以评估健康风险。最后一个项目将并行化我们研究小组开发的一种新的蛋白质结构预测算法，随后将验证一种新的贝叶斯最优预测理论。

9617349 Bic, Lubomir, F. Michael Dillencourt University of California, Irvine CISE Research Instrumentation: A Distributed Biomedical Computing Laboratory Many important problem areas of biomedical computing have the ability to generate large volumes of relatively coarse-grain computations and thus can benefit greatly from a LAN-based network of workstations. This project will provide such an infrastructure, based on a cluster of Sun workstations, which will be shared by the following four projects: -Knowledge Discovery in Clinical Databases-Brain Network Models-Toxicology simulations-Distributed Protein Structure Prediction. The goal of this project is to discover new knowledge in clinical databases that will be useful for diagnosis and treatment. It uses novel machine learning methods that are particularly suited to medical databases. The second project attempts to understand the functions of specific brain regions. It will develop parallel versions of neural-net-based algorithms to solve difficult signal-processing tasks. The third project develops a distributed simulation environment that supports the construction and interactive use of simulation models for predicting concentrations of chemicals (drugs or toxins) in the human body to assess health risks. The last project will parallelize a novel protein structure prediction algorithm developed by our research group, and, subsequently, will validate a novel Bayes-optimal Prediction theory.