新万博manbetx官网【首页】

学术运动

数据、模子和决议计划--生物医学工程的办法和使用
2019-10-15

来路:      供稿:                点击次数:      字号:【        

题      目:数据、模子和决议计划--生物医学工程的办法和使用
主  讲  人:Knut  M?ller
时      间:2019年10月15日(星期二)
地      点:新教一 214
主 办 单 位:化学系

主讲人引见:Knut  M?ller,德国Hochschule Furtwangen University医学与生命迷信学院传授、使用研讨所副主任,于1986年和1992年辨别取得德国波恩大学盘算机迷信(辅修:运筹学)硕士学位和人体医学三级国度测验(医学博士)学位,1991年取得波恩大学呆板学习和呆板人技能博士学位。1991年至1998年,担当德国波恩大学盘算机迷信系和神经内科的助理传授。自1998年以来,担当德国Furtwangen University的医学信息学传授。2006年,成为技能医学研讨所(ITeM)的开创长处,2017年(与奥克兰大学和坎特伯雷大学协作)成为两国知识工程医学研讨所(BIKEM)的开创长处。在过来的10年里,发布了300多篇论文,取得了万万欧元的大众项目基金。研讨范畴次要包罗大数据、生物信号处置(如电阻抗断层成像)、生理建模、条理辨认和决议计划支持,并使用于生物医学工程,如机器通气、伤口愈合和病愈。
内 容 介 绍:Decision making is crucial in every day’s life, but even more important e.g. in economics, politics or in medical care. “Best” decisions (according to some metric used for optimization) are always based on models (implicit or explicit), that are employed to estimate effects of actions on the target system. Models have to be predictive, e.g. may predict drug effects on a human body in relation to dose and time. Depending on prior knowledge (or degree of uncertainty) about the domain, models can reflect physical properties (or other scientific established relations) or may employ descriptive statistics or may be mixtures of both. In Biomedical applications special constraints coming from medical device regulations (legal requirements) must be considered. Models therefore should be practically (or at least) structurally identifiable. Currently, data driven approaches e.g. the popular machine learning (like Convolutional Neural Networks, Hidden Markov Models) are also evaluated for medical decision support. In the talk we will therefore introduce different type of models, their dependence on data (quality, completeness) and properties for medical decision making.
备       注:

 


Baidu
sogou