HI 5311 Foundations of Health Information Sciences II

Foundations II covers the several types of models and modeling issues that arise in health information science and biomedicine, such as model building, fitting, and validation, and then covers a number of modeling approaches that are of general applicability to a wide range of health information science and biomedicine domains. These include computational, qualitative, quantitative, and logical models. The course is meant to introduce students to these topics in preparation for selected in-depth study in more advanced courses. A special feature will be the group projects.  The students will be divided in groups and will work, throughout the semester, to gain hands-on experience on developing, running, and testing models against patient and experimental data.  Problem Based Learning will be employed.  The problem selected is concerned with the development of patient-accurate simulation models of tumor growth.  Each group will be responsible for performing a parameter sensitivity simulation study of tumor growth using sophisticated cell-agent and continuum models developed in Professor Cristini's group.  The purpose of the investigation will be to understand how to calibrate model parameters to reproduce, in simulations, the morphologies observed from pathologic analysis of patient tumors.  This will expose the students to multidisciplinary research in collaboration with breast cancer pathologists at MD Anderson, and will provide a formidable in-depth training on model calibration, testing and validation.

Course Learning Objectives:

At the end of the course, students will be able to:
• Describe the role of models and modeling in health informatics and biomedicine
• Describe several different types of models (e,g,, qualitative, quantitative, statistical, causal, etc.) and when and where they might be used.
• Describe general modeling issues, including model selection, construction, validation and fitting.
• Describe the relationship between syntax, semantics, and computation.
• Describe and explain why some problems cannot be solved computationally in a reasonable time-frame
• Use the web ontology language (OWL) to define a small ontology.
• Use the Unified Modeling Language to build a model of data structures and simple workflow.
• Use a problem space to model a well-structured problem solving task.
• Write a rule-based system to implement the problem solving model
• Write a simple Medical Logic Model and explain the role of MLMs in clinical decision support
• Model simple scenarios involving uncertainty using Bayesian Networks, Markov Models, and influence diagrams.
• Describe the use of discrete, multi-agent and continuum simulation in health informatics and biomedicine and modify simple models
• Create a model of a health information science or biomedicine process and describe that model scientifically, including how it can be applied, it’s strengths and limitations.
• Fit a model to data and validate the model.

Topical Outline by week:

(See also weekly outline in Moodle)

1. Introduction and course overview (Lecturers: V. Cristini and T. Johnson)
2. Introduction to the theory of computation and grammars  (Lecturers: H. Wang and T. Johnson)
3. Agent-based modeling:  cell systems in 4-D (Lecturers: V. Cristini and P. Macklin)
4. Model calibration, testing and validation.  Case study:  using patient pathology data to calibrate a cell-agent model (Lecturers: V. Cristini, M. Edgerton and P. Macklin)
5. Modeling problem solving and decision making using problem spaces (Lecturer: T. Johnson)
6. Modeling problem solving and decision making using rule-based systems  (Lecturer: T. Johnson)
7. Developing predictive models:  multiscale modeling connecting the cell information to the disease (tissue) scale (Lecturers: V. Cristini and P. Macklin)
8. Midterm (no lecture)
9. Modeling decision making under uncertainty using Bayesian networks and influence diagrams  (Lecturer: T. Johnson)
10. spring break (no lecture)
11. Multiscale modeling continued (Lecturers: V. Cristini and P. Macklin)
12. Modeling decision making under uncertainty using Markov models  (Lecturer: I. Willcockson)
13. Continuum modeling of the tissue scale using Partial Differential Conservation Laws (Lecturer: V. Cristini)
14. Parameter-sensitivity studies:  predicting tumor size and growth in patients (Lecturer:  V. Cristini)
15. Knowledge representation:  from semantic networks, frames and scripts to formal ontologies (Lecturer: T. Johnson)
16. Final (no lecture)

Prerequisites

• Applied Mathematics course, or equivalent. (Appropriate mathematics background up to the college level including at least calculus and algebra. Knowledge of basic statistics.)
• Prefoundations course on Data Structures and Algorithms or equivalent.
• OR Approval of coordinator.