# Biostatistics and Informatics Seminars

​​​​​

## Biostatistics and Informatics Seminars

### 2017 Seminars

5/10/17 - Junior faculty mentoring sponsored topic.

5/3/17 - Meeting of the Biostatistics Student Association (BSA) of the American Statistical Association (ASA).

4/26/17 - Matthew Mulvahill, MS - Topic forthcoming.

4/19/17 - Meeting of the Colorado/Wyoming Chapter of the American Statistical Association (ASA).

4/12/17 - Larry Hunter, PhD - Pharmacology - Topic forthcoming.

4/5/17 - Chris Gignoux, MD - Biomedicine - Topic forthcoming.

3/29/17 - Damian Wandler - Oracle - Topic forthcoming.

3/22/17 - Spring Break

3/16/17 (Thursday, 12 pm) - Keith Baggerly, PhD - Department of Bioinformatics and Computational Biology, MD Anderson Cancer Center, University of Texas -"Vitamin D, Recommendations, Reproducibility, and Controversy."

The easiest way to treat disease is often to prevent it from developing in the first place. Many prevention strategies involve nutrition and vitamin intake; these are important enough that many governments “recommend” target levels. Determining what these target levels should be involves integrating results from many types of studies and applying statistical modeling to identify intakes which will ensure most of the population (97.5%) will have their biochemical requirements met. Because these recommendations can direct policy, and can wind up affecting the health of millions, their derivations should ideally be clear and checkable. This isn’t always the case.  In this talk, we’ll review how the current US Recommended Dietary Allowance (RDA) for vitamin D, 600 International Units (IU) per day, was determined in a 2011 report from the Institute of Medicine (IOM). We find the RDA is too low by at least a factor of 2, and more likely by a factor of 5, due to errors in mathematical modeling (e.g., dividing by the wrong number). We then shift the discussion from mathematics to policy, and talk about how we’re now trying to fix the errors and what this entails.

3/16/17 (Thursday, 10 am) - Sara Selitsky - UNC Chapel Hill - Faculty candidate talk - Topic forthcoming.

3/14/17 (Tuesday, 9 am) - David Astling - University of Colorado - Faculty candidate talk - "Bioinformatic Approaches for Exploring Molecular Immune Responses."

3/8/17 - Fyong Xing, MS - University of Florida - Informatics Faculty Candidate - "High-throughput Biomedical Image Computing for Digital Health."

In biomedical informatics, a large amount of image data has been collected to support clinical diagnosis, treatment decision and medical prognosis. The large volume and the diversity of informatics across different imaging modalities require advanced and high-throughput image computing technologies to provide more accurate disease detection, deeper understanding of the mechanisms of disease progression, and better healthcare in precision medicine. With the ever-increasing amount of biomedical image data, it is very important to design and develop efficient technologies for large-scale biomedical image analysis. This talk will describe high-throughput biomedical image computing methods for digital health, focusing on three significant topics: object detection, segmentation, and image understanding in medical diagnosis. Specifically, I will present several novel machine learning and imaging informatics technologies to process biomedical big image data and the applications in medical diagnosis.

3/1/17 - Ayush Goyal, PhD - Texas A&M University - Informatics Faculty Candidate - "Multi-scale Filtering and Fuzzy Clustering for Segmentation and 3-D Reconstruction from Medical Images."

Recent advances in biomedical image processing have been applied to three-dimensional volumetric medical image data for reconstructing 3-D computational models of organs and vasculature. Specifically, this talk will present automatic image processing algorithms based on multi-scale filtering and fuzzy clustering applied on optical fluorescence microtomography and brain and cardiac MRI / CT images for reconstruction of whole-organ or patient-specific computational models of brain gray and white matter, cardiac geometry, coronary or cerebral vasculature. Additionally, pixel-clustering methods, such as adapted fuzzy c-means or k-means modified and combined with connected region labeling, used to extract and reconstruct the brain gray and white matter regions and cardiac left and right ventricles will be discussed for both calculation of brain gray and white matter volume and assessment of a patient’s cardiac left ventricle and myocardial wall with computation of clinical parameters such as ejection fraction, stroke volume, and end-systolic and end-diastolic volume. A side application - the same image processing methods of pixel clustering and multi-scale line filtering can also be applied for segmentation of other objects of interest such as cells or bacteria in microscopic and histological images. We will discuss applying this approach to automatic disease screening by detection of pathological agents such as mycobacterium tuberculosis in stained microscopic or histological smears of patients (sputum, blood, or urine).

2/15/17 - Laura Wiley, PhD - Division of Biomedical Informatics and Personalized Medicine - "Precision Medicine and the Learning Healthcare System: Leveraging Informatics to Improve Care."

Precision medicine and the learning healthcare system are predicated on the assumption that by turning data (be they molecular, clinical, or social) into knowledge we can improve health. Informatics is a key enabler of this vision. Using warfarin pharmacogenomics as an example, we will examine the role of informatics for data discovery, translation, implementation and evaluation.

2/1/17 - Journal Club - "Calling BS in the Age of Big Data".

Discussion of the syllabus for a new course at University of Washington.  The aim of this course is to teach people how to think critically about the data and models that constitute evidence in the social and natural sciences.

2/8/17 - Emanuele Giorgi - University of Lancaster - "Disease mapping and visualization using data from spatio-temporally referenced prevalence surveys."

We set out general principles and develop statistical tools for the analysis of data from spatio-temporally referenced prevalence surveys.  Our objective is to provide a tutorial guide that can be used in order to identify parsimonious geostatistical models for prevalence mapping.  A general variogram-based Monte Carlo procedure is proposed to check the validity of the modelling assumptions.  We describe and contrast likelihood-based and Bayesian methods of inference, showing how to account for parameter uncertainty under each of the two paradigms.  We also describe extensions of the standard model for disease prevalence that can be used when stationarity of the spatio-temporal covariance function is not supported by the data.  We discuss how to define predictive targets and argue that exceedance probabilities provide one of the most effective ways to convey uncertainty in prevalence estimates.  We describe statistical software for the visualization of spatio-temporal predictive summaries of prevalence through interactive animations.  Finally, we illustrate an application to historical malaria prevalence data from 1334 surveys conducted in Senegal between 1905 and 2014.

2014-2015 Biostatistics and Informatics Seminars and Journal Club

#### Colorado School of Public Health

13001 E. 17th Place
Mail Stop B119
Aurora, CO 80045

303.724.4585