Robust Treatment of Missing Data in Biomedicine
Principal Supervisor: Dr. Cassio P. de Campos
Second Supervisor: Dr. Amy Jayne McKnight
+ Project Description
Missing values are present in many types of data from a wide range of disciplines. Their treatment is a very common problem in statistical analysis of clinical and biological data. It has been shown that missing values in the data can severely affect subsequent experiments and results. When data are obtained from biological experiments, missingness might be due to the occurrence of imperfections during the experiment, insufficient machinery/reading resolution, dust or scratches on the slide, as well as other failures in the process.
This project regards the study of existing methods to deal with incomplete data. The goal is to build a novel robust method based on credal networks that will combine available data and results from existing state-of-the-art methods to yield more accurate estimates for the missing values. Algorithms will be experimented on multiple real-world data sets with clinical and genomic data of cancer patients, among others. Collaborations with biologists and medical doctors will bring the developments of this project to important real-world applications. Well-known as well as novel methods will be compared using the incomplete data sets to show the advantages and disadvantages of different approaches. Finally, sensitivity analysis can help to understand the results.
+ How to Apply
Applicants should apply electronically through the Queen’s online application portal at: https://dap.qub.ac.uk/portal/