The Center for Innovation in Brain Science (CIBS) utilizes health care claims data, UK Biobank, All of Us and Alzheimer’s Disease Neuroimaging Initiative (ADNI) to study the capability of therapeutics to reduce risk of neurodegenerative diseases and evaluate risk factors that exist in patients that do not respond to the therapeutics. In addition, together with data sets generated within CIBS, these data were used to conduct multi-omic analysis to understand the underlying pathophysiology of neurodegenerative diseases. In this NIH-funded program, the students approach the dataset to answer questions of their own choosing. Further, a program is starting in which Native American graduate students will be trained in medical informatics.
The data science component at CIBS is managed by Drs. Kathleen Rodgers, Roberta Brinton and Francesca Vitali. Under this leadership, 2 postdoctoral fellows and 8 graduate students have conducted studies leading to numerous publications. For each medical bioinformatics project, a physician with expertise in the disease under evaluation is brought onto the team. Dr. Vitali, together with Dr. Raymond Shang, leads to multi-omic analysis of existing datasets as well as datasets unique to CIBS.
Dr. Rodgers leads the Native American undergraduate program using our analytic system. She is working together with Dr. Teshia Solomon and Dr. James Cunningham to initiate the Native American graduate program.
The initial goals of the medical informatics program were to optimize therapeutic treatment for the prevention of neurodegenerative disease, particularly dementias to support the stated goal of the NIA National Alzheimer’s Project Act (NAPA) to prevent and effectively treat Alzheimer’s disease by 2025. With the onset of SARS-CoV-2 pandemic, our training program with Native American students was significantly altered and required distance learning . The resulting impact of the remote learning model led to our first 4 Native American students being trained in medical informatics. One of the goals of this educational component was to enable research using large datasets relevant to Native American populations. This became formalized into a year-long training resulting in an undergraduate certificate in Data Science and Visualization.
The compelling aspect of our research at CIBS is the potential development of precision treatments for a prevention of an intractable health care issue: Alzheimer’s disease (AD) and related dementias (ADRD). By the time cognitive changes are noted, significant damage has been done to the nervous system. While many drugs have been studied clinically and a vast majority have failed, we believe there is an on-going real world experiment with clinical practices. Combining the use of medical informatics datasets with our unique approach to evaluating responders/non-responders baseline characteristics, we can better understand how drugs used to treat specific underlying risk factors for ADRD can be optimally effective in treating the diseases.
Independent of medical informatics, we are leveraging several biomedical sources of data and the fusion of multi-scale datasets, machine learning, and big data techniques to understand the mechanisms underlying the treatments identified in our medical informatics campaign.
Examples include a bioinformatic pipeline that ranks potential drug combinations to prevent Alzheimer’s Disease (AD) using Drug-Disease and Drug-Target associations from biomedical repositories, such as DrugBank and text-mining and natural language process algorithms. In addition, we are exploring mitochondrial function and how they were related to the progression of Alzheimer's Disease by mining multi-omics datasets, integrating all data types, including clinical data, genome information (genome and modifications), transcriptomics, proteomics, and metabolomics.
In most of our publications, important parts of our code, the dates on which the research was completed and the version of the dataset used are referenced in the paper. This allows anyone in the community to replicate within the dataset used. We have also verified/validated our results using alternate datasets and updated versions of existing datasets. With our multi-omics approach, we use public datasets as well as internal data sets to validate results. With regards to the training programs for Native Americans, an additional consideration must be made establishing a trust relationship with the appropriate groups.
By using existing data generated in real life settings, important insights can be gained to optimize health care and reduce the risk for age-related neurodegenerative diseases. We use a combination of datasets with large numbers of patients with top line data associated with their records (91 million health care claims data set) to reduce numbers of patients with lifelong health data along with longitudinal evaluations (500 thousand records, UK Biobank). Interactions with physicians who have reviewed our articles or attended our presentations, it is apparent that the strength of the observations has changed the way they approach healthcare delivery. Most recently, one of our graduate student trainees, who has obtained his PhD and is in his third year of medical school, used one of our recent publications to inform a patient of the benefits of the use of statins in the reduction of dementia risk. Multi-omics analyses have provided the basis for a number of publications and impact our understanding of therapeutic options and targets. Overviews of our data science training programs for Native American students have been presented to groups of tribal leaders and elders representing 22 Native American tribes in Arizona.
