The University of Michigan- Shanghai Jiao Tong University Joint Institute (UM-SJTU JI) stands at the forefront of technological innovation with its unique academic atmosphere and advanced scientific explore. JI boasts a strong research team which keeps innovating and strives to produce results that will lead the industry and ultimately benefit the human beings.
In order to help the public discover the mystery of scientific research, JI has launched a series of mini-workshop to introduce faculty’s research.
In his recent talk of “Predicting Neonatal Health based on Prenatal Data,” Professor Mian Li described how to analyze and predict the health status of a neonatal through machine learning.
Professor Mian Li (left) introducing the application of machine learning in predicting neonatal health status
Machine learning, as a multidisciplinary science incorporating probability theory, statistics, approximation theory, convex analysis and algorithm complexity, is the basis and core of the hottest field of artificial intelligence. Through analyzing large volume of empirical data, machine learning aims to develop computer programs that can extract models and trends in order to achieve intelligent prediction.
A newborn’s weight is an important indicator of a baby’s health; too light or too heavy birth weight will not only affect a child’s growth and development but will also increase its future risk of chronic diseases (hypertension, diabetes, obesity, cardiovascular and cerebrovascular diseases, etc.). Birth weight and maternal condition is closely related. Therefore, it is possible to predict a newborn’s birth weight and health based on the mother’s physical condition during pregnancy. With this foresight, pregnant women can timely correct their physical condition, such as adding nutrition, in order to improve the fetus’ health for the long term.
In the past, observations and predictions of the fetus were based primarily on ultrasound and the neonatal weight was estimated by the volume of fetus presented on the image, but this method required gestational age beyond 32 weeks and was unable to reflect the maternal effects. Applying the method of machine learning based on hundreds of thousands of prenatal physical examination data of pregnant women across the country, Li’s research team hopes to analyze the health status of newborns through the whole process of data cleaning, feature selection, modeling analysis and visualization of results. The research has reached the result stage, having completed the data cleaning and the characteristic analysis, and initially realized the classification modeling to achieve the overall 90% correct rate.
Mian Li is a special researcher, associate professor, and doctoral adviser at the JI. He was admitted to Tsinghua University’s Department of Automation in 1994 and graduated with a master degree in automation in 2001. He then went to study at the University of Maryland where he earned a doctoral degree in mechanical engineering with the award of the Best Doctoral Dissertation in 2007. After graduation, he worked as a postdoctoral associate at the University of Maryland. In September 2009 he returned to Shanghai and joined the JI. In July 2010 he was also appointed as Associate Professor at SJTU’s School of Mechanical Engineering. Since March 2012, he has been a special researcher, associate professor and doctoral supervisor at JI. Professor Li’s main research interests are in the design optimization of complex systems, including multidisciplinary optimization, multi-objective optimization, robust design and stability analysis, sensitivity analysis and uncertainty analysis, and simplified models and approximation algorithms. His innovative scientific research has achieved practical application results. At present, his research focuses on the application of advanced design methods, approximate models and optimization methods to interdisciplinary research fields, such as renewable energy systems, electric vehicles, vehicle energy system design optimization, complex intelligent information systems and so on.
