About the Center for Healthcare AI
Personalized, holistic healthcare: A fundamental moral issue
A distinct aspect of the USF Center for Healthcare AI is an explicit focus on developing AI as a tool to understand the whole person, Cura Personalis, by giving explicit attention to physical, mental, and environmental health dimensions, as well as to inclusion of marginalized populations that are often not incorporated into training healthcare AI algorithms. Working together with our clinical partners in the integrated health lab (iHealth Lab) in the School of Nursing and Health Professions, the Center for Healthcare AI strives to create enabling AI technology that will be clinically useful in the real world and have lasting impact on personalized, holistic healthcare.
The goal of personalized or precision healthcare is to provide the right care, to the right person, at the right time. This vision for healthcare recognizes that health is complex and multidimensional and that the siloed, one-size-fits all approach to healthcare is not sufficient. A holistic view of health necessarily must include mental health. Personalized healthcare also incorporates health equity, a recognition of the diversity of people and their diverse needs. Personalized care requires a new multidimensional, integrated approach to healthcare. But integrated healthcare is not possible without integrated, multidimensional data about a person, and integrated, multidimensional models of health.
The concept of personalized healthcare aligns with the values expressed by the phrase Cura Personalis – care for the whole person. Cura Personalis recognizes that mental health is not just an independent dimension of health that can be treated, or not, in isolation, without affecting other dimensions of health. Recent reports estimate that “mental illness a distant first in global burden of disease in terms of YLDs [years lost to disability], and level with cardiovascular and circulatory diseases in terms of DALYs [disease adjusted life years]. The unacceptable apathy of governments and funders of global health must be overcome to mitigate the human, social, and economic costs of mental illness.” 1. Thus, the multidimensional nature of health must include the mental dimension in its analysis. This is also a moral issue:
The fundamental truth of global mental health is moral: individuals with mental illness exist under the worst of moral conditions… The widespread stigma of mental illness, which prevails in countries as disparate as China, India, Kenya, Romania, Egypt, and the USA, marks individuals with severe psychiatric disorders as virtually non-human. None of the world's major religions—no matter how strong is its message of support on behalf of the most marginal and vulnerable sufferers—has been able to break this cycle of misery.”
Arthur Kleinman, MD Professor of Medical Anthropology and Psychiatry at Harvard Medical School 2
If we are to achieve personalized, holistic healthcare, and if we are to live up to the values of Cura Personalis, and to strive toward true equity and inclusiveness within the diversity of human health, a multidimensional approach that includes mental health in a fundamentally integrated way is essential. There is no health without integrated mental health, and there is no integrated health without integrated information.
AI as a Core Enabling Technology for Personalized, holistic Healthcare
The enormous progress that has been made in the application of AI to many areas of life has created the hope that AI may enable early disease detection and create an era of truly personalized medicine 3. However, “the inconvenient truth” is that current AI algorithms that dominate the research literature are generally of little use in clinical practice. Two barriers are largely responsible for this. First, AI alone is not sufficient to improve a fragmented healthcare system where the data required for AI to work is siloed away by incompatible data formats or systems. Related to this, most healthcare organizations lack the infrastructure required to train AI algorithms adequately and equitably 4. Bias and inequity plague such systems because of a failure to train algorithms with adequately represented populations. In addition, algorithms and training data can only represent models for which humans explicitly program them. Mental health dimensions are rarely included in comprehensive databases or incorporated into AI algorithms.
Research in the CAIM strives to overcome the barriers that have prevented AI from improving healthcare, including especially mental healthcare. While pursuing rigorous AI and predictive analytics algorithms, we also give careful attention to the importance of understanding health data and the need for holistic, integrated, data from diverse populations.
The Brain Changes First
A unique characteristic of mental, neurological, and neurodevelopmental (MNN) disorders is that the overt behavioral and or cognitive symptoms that define these conditions are the end product of a long developmental process. The brain changes first. It is known, for example, that 80% of the dopaminergic neurons in the substantia nigra must be lost before the first symptoms of Parkinson’s Disease appear 5. Children who have suffered various traumas have a much greater risk of developing diabetes, heart disease, and cancer, not just serious adult mental disorders 6. The developmental trajectory of nearly all mental and neurological disorders opens a window of opportunity for early intervention and prevention. But early intervention is predicated on early detection.
Monitoring the Brain through the Lifespan
The University of San Francisco, in partnership with the AI in Medicine initiative in the USF Data Institute is embarking on an ambitious effort to explore and implement integrated, personalized approaches to healthcare with the goal of changing mental healthcare from a reactive practice to a preventive one, while integrating mental health dimensions into routine care through the lifespan. The key to our approach is leveraging advanced information technology, novel discoveries in neurodiagnostics, while working closely with clinicians from our nursing and clinical psychology programs to ensure that our new approaches are sustainable and feasible in real world settings. The goal is to enable continuously updated risk profiles for a variety of mental and neurological diseases, delivered to clinicians and patients in time for early interventions.
1 Vigo D, Thornicroft G, Atun R. Estimating the true global burden of mental illness. Lancet Psychiatry 2016;3:171–8.
2 Kleinman A. Global mental health: a failure of humanity. Lancet 2009; 374:603–4.
3 Fogel AL, Kvedar JC. Artificial intelligence powers digital medicine. Npj Digit Med 2018;1:1–4.
4 Panch T, Mattie H, Celi LA. The “inconvenient truth” about AI in healthcare. Npj Digit Med 2019;2:1–3.
5 Insel TR. Mental disorders in childhood: shifting the focus from behavioral symptoms to neurodevelopmental trajectories. JAMA 2014;311:1727–8. doi:10.1001/jama.2014.1193
6 Berens AE, Jensen SKG, Nelson CA. Biological embedding of childhood adversity: from physiological mechanisms to clinical implications. BMC Med 2017;15:135.
Current Research Projects
Much of our current work related to personalized mental health has emerged from innovations in electroencephalogram (EEG) analysis, combined with clinical data and novel AI methods developed by our faculty. A brief list of projects and collaborators is given below. We are actively seeking funding to expand these efforts. A particularly important need is funding, lab space, equipment, and partners for developing new clinical protocols for integrating EEG measurements and data into the routine checkup clinical workflow. Also needed is funding for research time and computing resources for faculty, postdocs, and students.
This research is done in collaboration with Harvard Medical School, Boston Children’s Hospital, the University of Texas at San Antonio, and the NY Institute for Basic Research on Developmental Disorders. A private startup, mHealthcare, Inc, is seeking funding to partner with us for clinical testing of this technology. USF Nursing students are organizing laboratory testing to develop clinical protocols for our novel approach to monitoring infant brain development.
Led by child psychologist Michelle Bosquet Enlow at Harvard Medical School, Dr. Bosl is the principal neuroscientist on the project to find brain-based biomarkers for these highly prevalent disorders.
Detection of childhood epilepsies and personalized therapeutic plans based on personalized EEG analysis is being sought with the Epilepsy Center at Boston Children’s Hospital.
With the Sleep Clinic and Epilepsy Center at Boston Children’s Hospital, we are seeking biomarkers for nocturnal epileptiform discharges associated with autism, for atypical sleep architectures associated with various disorders, and cognitive impairment due to sleep disorders. Professor Bosl holds a patent for epilepsy and epileptogenicity biomarkers with Boston Children’s Hospital that is available for licensing.
Funding is being sought for a partnership with Trident.ai and the iHealth Lab to develop clinical biomarkers for PTSD and for monitoring therapeutic effectiveness, based on Professor Bosl’s algorithms for analyzing EEG signals.
This research was initiated after our faculty gave a presentation to the Indonesian Alzheimer’s Association on new technologies for Alzheimer’s. A new startup, Trident.ai, is working with us to seek private funding to support our faculty to pursue clinical research to demonstrate that early biomarkers that detect aberrations in brain function, are possible years before symptoms of mild cognitive impairment appear.
We are working closely with partners with the Global Organization of Health Education and Purple Point Neurodiagnostics to take our innovations to low income and low resource regions of the world. In addition, we are developing curricula to increase research capacity in the countries where we work.
Our goals for impacting healthcare include training a new generation of nurses, psychologists, health data scientists, and public health experts who can design integrated healthcare systems and are comfortable with new digital technologies that will power the revolution in personalized healthcare. Graduate programs that integrate biomedical and mental health informatics, AI methods, wearables and mobile devices into an information rich health monitoring ecosystem will be developed and taught.
Industrial Partnerships Sought
While we continue to apply for federal and state research grants to support our work, we actively seek philanthropists, private investors, and individuals with a deep interest in personalized, integrated healthcare to join our iHealth Lab. Some technologies may be ready for adoption and commercial scaleup to the bedside, while others require additional proof-of-concept research. All are aiming at high impact clinical implementation.
For more information and to get involved, please contact Elisabeth Merkel at email@example.com.