Developing a system for bringing medical knowledge to a consumer level is a current challenge in health informatics today. Existing systems such as expert systems are computerized decision support systems that implement artificial intelligence to help provide advice when treating patients.

Past systems have been shown to benefit people working in medicine; medical students, medical professionals, providers, etc. The group of people it has been shown to be most valuable are to the medical students; those with the least experience and previous clinical knowledge. These existing systems can then be adapted to consumers based on this study.

The reason why decision support systems are not being used as widely as they should be is because there is reluctance from medical professionals and low priority assigned to AI in healthcare systems. However, setbacks from using expert systems include consumers having a hard time understanding how to use the technology, and no research has been done to show how effective these systems on the consumer level.

These systems not only provide decision-based support, but can also be used to help triage patients. This isn’t meant to replace medical professionals, but to be of aid in the process. This also allows for patients to be aware of certain drug interactions or health risks, and help the patient’s decision on whether or not it is appropriate to see a doctor.  

HouseCall, for example, is a consumer health informatics system based off of an existing physician knowledge base, called Iliad. The call for patient information to be more easily accessible is being addressed with this application; studies have shown that patients like interacting with the application and enjoy participating in trying to solve their medical problems.

Dr. Lisa Vizer, CHiP core faculty members and Assistant Research Professor in the School of Medicine, is doing research related to adapting, designing, and developing technology so that  patients can monitor their own health. By bringing medical knowledge to a consumer level, technologies can quickly pick up when treatment is becoming ineffective. Dr. Vizer has worked on a research project that focused on the usability of a patient education and motivation tool using heuristic evaluation. She has developed applications that provide educational content about a patient’s condition in order to identify problems more efficiently.

Telemedicine, a branch of medicine that aims to connect patients and providers in alternative ways rather than face-to-face communication, is a rapidly expanding field that seeks to improve the quality and accessibility of patient health. With the establishment of this kind of technology in medicine, health informatics will be taking a more consequential role.

The benefit of telemedicine is that underserved and isolated populations are able to access the care they need. Due to the digitization of health records, now information can be easily passed on and patients can receive faster care of the same quality. A current need is for systems to be configured for the accessibility of patient information internationally. The result from telemedicine connecting hospitals and their constituents is that global systems will become more efficient in dealing with world-wide epidemics as well as providing care to remote areas.

The data gathered from remote patient monitoring, a discipline of telemedicine where providers use heart rate sensors, pressure gauges, and other medical devices to gather information on patient health, is used to identify high-risk patients and then consequently provide a personalized treatment plan.

Because it’s hard to share and analyze healthcare information due to HIPAA regulations, an NIH sponsored project allows datasets to be anonymous so that they can be more accessible to researchers. The project strips any identifying information and leaves the treatment plan and how the patient responded. This allows researchers to provide health analysis and improve patient care.

Dr. Saif Khairat, Assistant Professor in the School of Nursing and CHiP core faculty member, is the Principle Investigator of Carolina Applied Informatics Research. The focus for this research team is on EHR usability, teleconsent, and telemedicine. More specifically in telemedicine, Dr. Khairat’s research focuses on improving clinical management of chronic diseases. Based on existing data, this study aims to find a systematic approach to classifying potential patients who would benefit from telemedicine, whether it be due to geographic distance, lack of distance, or costs.

Artificial intelligence is aiding healthcare providers in data acquisition, management, and analysis. Previous topics discussed in relation to AI have been personalized medicine, drug repurposing, and data visualization. Predictive analysis is another way to improve efficiency and diagnoses with machine learning.

Two companies, Prognos and Google, are delving into machine-learning with a focus on healthcare. Machine-learning is a branch of artificial intelligence that allows systems to learn data, identify patterns, and make decisions, for the most part, autonomously. Data can be learned through supervised or unsupervised learning, as well as other methods, where the algorithm is given a training set of labeled examples (supervised learning) or the algorithm explores the data to find an existing pattern to provide a data structure (unsupervised learning).

Prognos is a healthcare company that uses AI to predict diseases, resulting in earlier decisions in healthcare in collaboration with diagnostic, insurance, and life science companies.They have recently raised over $20 million toward their mission of predictive analytics. By combining healthcare and AI to predict disease earlier, they are able to identify patients in need of improved treatment decision-making, risk management, and quality improvement.  

DeepMind Health is another AI support system, brainchild of Google, that deals with predictive analytics. With deep machine-learning, DeepMind Health is able to understand images and provide an analysis that offers feedback and data segmentation for clinicians.

Dr. Carlton Moore, CHiP core faculty member and Associate Professor at UNC School of Medicine, has done research pertaining to artificial intelligence, more specifically natural language processing.

Natural language processing is the field of computer science that combines machine learning with computational linguistics. Dr. Moore’s research focuses on extraction of certain keywords in patient electronic health records. He is also interested in researching medical errors during transitions in patient care and assessing the adequacy of abnormal tests results that are followed up after ambulatory care.

Two datasets - Project Baseline and Alzheimer’s Disease Neuroimaging Initiative are affecting the way we look at healthcare. Data analysis is a key part of how we make decisions and changes in medicine; through digitization of this data, researchers are able to reduce costs and improve outcomes.

Project Baseline, a study created through the collaboration between Duke School of Medicine, Stanford, and Google, aims to take raw health data and create a baseline of this data that would be used to visualize connections between distinct populations. These connections can be utilized to make more informed decisions and to have a better understanding of health and disease. Participants are asked to take health tests, give bodily samples, and answer questionnaires about their health and lifestyles. Sensors and wearable devices are also provided to the participants to record heart rate, activity levels, and sleep habits.

Alzheimer’s Disease Neuroimaging Initiative (ADNI) is a highly curated data set that contains MRI and PET images, genetics, cognitive tests, CSF, and blood biomarkers to help predict the onset of Alzheimer’s Disease. The purpose of gathering this data is to research new treatments that will slow and stop the progression of Alzheimer’s Disease. Researchers participating in the study follow a uniform procedure and protocol to ensure that any inconsistencies are eliminated. The data is free and available to any authorized researchers who request the data.

Gathering and maintaining data is necessary in research, but data-sharing is also important; validating results and furthering research from prior results all stem from promoting data-sharing. The National Science Foundation has started a project to supply resources and bring attention to data management principles called Sustainable Digital Data Preservation and Access Network Partners (DataNet).

Similar to the Visible Human Project, having highly curated datasets means that research can be reproducible, results can be verified, and across disciplines, researchers can make more connections when datasets are well maintained and shared.  

What is the importance of studying genomics? Genomic medicine is a discipline that involves using an individual’s complete set of DNA to improve their clinical care. An exciting trend that has developed from studying human genomics is personalized or precision medicine. This is the practice of using genomics, epigenetics, environmental exposure, and other data to form a genetic profile in order to make an informed diagnosis or treatment plan for a patient.

For example, BRCA1 has been identified to be a gene that produces a tumor-suppressing protein. Mutations in this gene are commonly associated with breast cancer, one of the most common cancers found in women worldwide. When this protein malfunctions, breast cells start to divide uncontrollably, causing a mass to develop. Mutations such as the ones associated with BRCA1 can be passed down genetically; however, depending on environmental and lifestyle factors, these genetic mutations may not ever be expressed.

Getting genetic testing done can reveal if someone has mutations on the BRCA1 gene, and preventative measures can be taken to address the risk of breast cancer. Tools have also been developed to visualize the human genome such as BioViews, which is used to create a physical, annotated sequence map and DNA sequence display. Data can be analyzed, visualized, and annotated, making it easier to make connections and organize different datasets.

CHiP faculty member and Associate Professor, Dr. Di Wu, has worked with genomic data to explore drug repurposing for different cancer subtypes. Based on the genomic profiling of the tumours, normal cells, and cells in the cancer cell lineage, drug sensitivity is being researched to determine the most effective treatment.