David Nelson

Dr. Nelson's current research interests involve reliable model selection in an environment where the ratio of the number of potential variables p to the sample size n is larger than what has been traditionally encountered. Such "small n, large p" situations have become especially prevalent in modern "-omics" data. However, the problem arises as well in many other fields, as well as in most epidemiologic studies. His major focus has been on adapting ensemble methods from the machine learning community, such as random forests and boosting, to non-traditional problems, such as building effective models for predicting disease outcomes from environmental exposure data, developing mass spectrometry signatures as prognostic indicators for health outcomes of interest, and other complex problems involving the interplay of genomic data, measurement processes, and epidemiology.

Dr. Nelson became Principal Biostatistician for the Cancer Prevention Institute of California in October 2006. Prior to that time, he was a Group Leader for Biostatistics and Bioinformatics within the Computations Department at Lawrence Livermore National Laboratory (LLNL), where he had been a computer scientist, statistician, and software engineer for over 25 years.

He was a senior member of the team that founded the Department of Energy's Joint Genome Institute (JGI) in Walnut Creek. He has been the director or co-director for two successful P01 statistics and data management cores, as well as founding member and statistical lead for the Livermore Microarray Center at LLNL. He has been providing statistical and informatics consulting expertise to scientists at the CPIC, LLNL, the JGI, and in the private industry for over two decades.

Dr. Nelson brings to CPIC an extensive background in statistics, mathematics, computer science, and software engineering, especially as they relate to applying modern statistical techniques to complex problems involving the interplay of genomic data, measurement processes, and epidemiology. He is involved in several large-scale ongoing studies at CPIC, where he is adapting and extending current research in machine learning to the unique needs and problems of CPIC investigators.