Overview

Explore the pragmatic application of big data, advanced analytics, and AI in population health and social determinants of health in this informative grand rounds presentation by Dr. Jonathan Weiner, co-director for the Center for Population Health IT at Johns Hopkins School of Public Health. Delve into the distinctions between population health informatics, public health informatics, and clinical informatics. Examine the digital ecosystem surrounding healthcare and population health, emphasizing the crucial role of social determinants of health in human well-being. Investigate various data sources, applications, and contexts for social determinants of health and population health analytics. Gain insights into overlapping data science fields, including AI, machine learning, big data, and predictive analytics. Explore the pillars, processes, and target outcomes of a value-based learning health system, highlighting the importance of analytics and big data. Address the challenges of big data in population health, including volume, variety, velocity, and veracity. Compare the effectiveness of machine learning techniques with standard regression in predicting chronic care outcomes and costs. Learn about a study linking medical, public health, and human service databases to better predict opioid overdoses. Consider the potential biases and harms associated with AI and analytics in healthcare. Conclude by examining key data science-related challenges and opportunities in the field.

Syllabus

Intro
Population Health Informatics vs. Public Health Informa and Clinical Informatics
Health Care and Population Health Is Surrounded by an "e-Health" Digital Ecosystem
Social Determinants of Health (SDOH) are more important than medical care when it comes to human wellbeing
Social Determinants of Health and Population Health Analytics: Dat Sources, Applications and Context
Working Definitions
The Overlapping Data Science Fields: Ai, ML, Big Data, Predictive Analytics and More there is not always consensus regarding terms
The Pillars, Processes and Target Outcomes of a Value-Based Learning Health System Analytics and Big Data are Key
The Big Data Phases Applying Predictive Modeling: From Raw Data to Knowledge
Big Data "Volume" Challenges in Population Health
Big Data "Variety" Challenges in Population Health
Big Data "Velocity" Challenges in Population Health
Big Data "Veracity" Challenges in Population Hea
Working Definition of Artificial Intelligence and Machine Learning
TYPES OF ARTIFICIAL INTELLIGENCE
"Deep Learning" (aka neural nets) uses brute force to look for layer of "hidden" relationships between inputs (aka features) and outputs
In addition to predictions, ML can be applied to other types of pop heal analytics: "feature" (ind. variable) discovery, classification and selection are key
Comparison of AUC of ML and standard regression: One stud predicting chronic care outcomes - Model Accuracy is similar
CPHIT Study comparing standard regression to ML technique for predicting costs
At CPHIT we linked a range of medical, public health and hur service databases in Maryland to better predict opioid overdoses
Let's not forget that Ai and other types of analytics can be a source of bias or harm, leading to a type of e-iatrogenesis
Some key data science related challenges and opportunities in