COURSEWORK

In collaboration with mentors, each trainee will engage in a customized curriculum that emphasizes cross-disciplinary research training and integration. The coursework on AI and big data analytics will help trainees establish a roadmap, build core competency, and develop advanced skills in systems modeling for precision nutrition. 

1. Precision Nutrition and Artificial Intelligence Introductory Boot Camp: All trainees will engage in a hybrid-delivered week-long training intensive in nutrition and data science upon entry into the program to build an introductory-level proficiency and establish a foundation for interdisciplinary training. In this “boot camp” experience, students will be introduced to critical concepts in integrative nutritional biochemistry and physiology, metabolic disease progression, systems science, data science, machine learning, control systems, and biostatistics. Trainees will build coding and critical thinking skills through lectures, practice assignments, and team-based hands-on learning cases. All trainees will be provided with basic programming knowledge resources prior to the start of the boot camp to ensure that all trainees have basic coding skills in Python. Daily mentoring and coaching sessions will be implemented to promote trainee-faculty interaction and model interdisciplinary collaboration.

2. Leveling courses: During the first or second semester of the graduate program, trainees will take one or both of the two leveling courses. Students with limited technical background, but with a background in nutritional sciences, will take the Analytics leveling course. Students with limited background in nutritional sciences, but with an analytical background, will take the Nutrition leveling course. In the event that a trainee has metabolism expertise in a field like exercise physiology, they will take both leveling courses to ensure baseline nutrition and analytics knowledge. These courses already exist in Biomedical Informatics (BMI) and Exercise and Nutritional Sciences (ENS) programs (described below).

   • 1. • The Analytics leveling course (adapted from BMI undergraduate courses) will introduce trainees to the basic concepts, tools, and skills for analyzing health data following the Cross-Industry Standard Process for Data Mining and the extended Analytics Solutions Unified Method for Data Mining. The topics will cover data structure, data type (streaming, structured, unstructured, etc.), data modalities commonly encountered in nutrition research (EHRs, behavioral information, omics data, imaging, mHealth, etc.), data preprocessing, objectives of the analysis (descriptive, predictive, prescriptive, causal, etc.), and introduction to statistical and computational modeling. Basic matrix notation will be introduced so follow-up courses can use this notation to represent models and solutions, but the calculations will be performed using Python packages. Additional topics include data ETL (extract, transform, load) and a brief introduction to database and management tools (SQL and noSQL). An important approach will be biweekly programming assignments to provide hands-on training. Furthermore, a teaching assistant knowledgeable of the software tools will be available every semester of the INTERACT Program.

        • The Nutrition leveling course (NTR-510) will provide a thorough overview of nutritional needs and nutrition-related conditions in different stages of life, including pregnancy, lactation, infancy, childhood, adolescence, adulthood, and the elder years. This course emphasizes how developmental changes at each life stage contribute to specific dietary behaviors and nutritional needs. The course covers a range of diseases that can result from nutritional deficiencies or excess and how these issues can be addressed. The course also covers how cultural, environmental, behavioral, psychosocial, physical, and socioeconomic factors affect nutritional status through the lifespan.

3. Systems-Based Modeling for Precision Nutrition: Systems modeling is a powerful tool to address dynamic complexities in human health by incorporating elements at multiple levels ranging from genetic risks to environmental exposures, from individual-specific factors to population characteristics, and from resource allocations to delivery systems. This course will teach advanced computational methods of systems modeling to extract biomedical insights from heterogeneous big data sets. Given the broad coverage of this course, Drs. Whisner and Liu (Co-Program Directors) will lead the development and delivery of this new course and invite contributions from other mentors. A detailed syllabus example can be seen here.

4. Foundation Courses: All students will be required to take 9 credits of Foundational Courses (3 classes), which include a Research Methods course during the first semester, and the Machine Learning for Health Applications and Grant Writing courses during the second semester. These courses will help trainees build a solid foundation to conduct AI-assisted precision nutrition research.

   • 1. • Research Methods: There is an array of Research Methods classes at ASU and trainees will be required to take a Research Methods class deemed most relevant to their research area. Trainees in the ENS and BMI PhD programs will take one or a combination of EXW-700 (Research Methods), BMI-502 (Foundations of Biomedical Informatics Methods I), and BMI-505 (Foundations of Biomedical Informatics Methods II).

        • Machine Learning for Health Applications: BMI-555/IEE-520: Statistical Learning for Data Mining, taught by Dr. Runger (mentor), has been adapted over the years for advances in AI and technology tools. It surveys key algorithms from a computer science perspective allowing students to understand the assumptions, advantages, disadvantages, and relationships of methods including supervised, semi-supervised, unsupervised, and reinforcement learning. Applications to health data sets are used as examples to provide insight into the role of various tools in a solution. The class will culminate with a group project that requires students to analyze real-world clinical, behavioral, and nutritional data using the methods learned in class.

        • Grant Writing: The product of this course is a R21-formatted grant application responding to an RFA or PA from NIH on a topic that is, or could be, a research study for a thesis. The 16-week course addresses each of the application sections (Specific Aims, Significance, Innovation, Investigators, Approach, and Environment) and includes the development of an NIH Biosketch, NCBI bibliography, Human Subjects sections; multiple PD/PI leadership plan (if applicable); and resource sharing plan. Students complete assignments, review peer submissions, and engage in in-class peer activities and presentations. The final week is a mock study section, attended by the students, senior PhD students, and students’ mentors, each of whom review one to two grant application. The final product is a revised application, incorporating the reviews, and an introduction page responding to the comments from the peer review.

5. Concentration Courses: These courses provide flexibility for trainees to match courses to their specific research interests and study topics in greater depth. The data science courses cover a range of topics including applied statistics, machine learning, control systems, computer vision, deep learning, signal processing, and biomedical informatics. The nutrition science courses include topics such as nutritional biochemistry and physiology, nutrigenomics, SDoH, and health behavior change theory to policy and environment. These courses were carefully selected such that they satisfy the main degree requirements in addition to the requirements of the INTERACT Program. Students will select appropriate courses in collaboration with their mentors. All students will be required to take 12 credits in an area of concentration, with six credits from two of the AI/data science courses and six credits from two of the nutrition/metabolism courses. Collectively, the courses will inform trainees on different aspects of nutrition/metabolism (e.g., molecular, individual, community) and how to integrate different data types within these dynamic systems of health. It is important to note that the list of courses is non-exhaustive. Trainees may identify other courses suited to their research focus and the mentors and training committee will review suggestions. Below are tables that include a list of potential courses.

• Informatics Concentration Courses

• Nutrition/Metabolism Concentration Courses

6. Special Topics courses will provide hands-on experiences to students through group-based projects that involve reviewing and discussing relevant papers and working with real-world data on real-world problems. The topics covered in these courses will vary depending on the pair of faculty mentors who collaborate as instructors. In all instances, the courses will expose trainees to a variety of of research problems, analytical methods, and a team-based approach for problem-solving using heterogeneous biomedical data (e.g. NIH Common Fund datasets). Students will gain an appreciation of the opportunities and challenges of working in multi-disciplinary teams on emerging nutrition/metabolism and AI/systems modeling problems.