Background

The Flint Water Crisis highlighted challenges faced by communities engaging with the academy in response to a public health emergency. Differences in priorities may lead to community perspectives and needs being ignored. To address this issue the Community Engagement Program of the NIH funded Michigan Institute for Clinical & Health Research (MICHR) collaborated with Flint-based community partners to create/design the Research Readiness and Partnership Protocol (R2P2) to guide and facilitate how communities and academic institutions work together in future crises. A toolkit was conceived and developed, based on this protocol, as a first step in developing an instrument with relevance to communities throughout the state of Michigan.

Objectives

To create and evaluate a toolkit, using an iterative community-engaged process incorporating input from statewide community representatives, that facilitates community/academic collaboration during a public health crisis, building on the R2P2 content.

Methods

MICHR and the Community Based Organization Partners (CBOP), in Flint, MI, conducted a community-academic team-led literature review to identify common elements in disaster response protocols and key interviews (n=31) to glean perspectives from a cross section of Flint community leaders. Key findings from R2P2 were extracted to create guidelines and recommendations. From this R2P2 protocol the FAST-PACE (Promoting Academic & Community Engagement) toolkit was co-developed and assessed by the community. Listening sessions were conducted to gather input, and enhancements were made using an iterative process.

Results:

Facilitating transparency, engaging in clear bi-directional communication, and power-sharing were identified as the most critical issues necessary to support communities exercising their agency in acting to improve community health throughout the state of Michigan. The use of tailored features and technology (including AI) were identified by community to enhance accessibility. Iterations of the toolkit will be demonstrated highlighting enhancements incorporated based on statewide community input.

Conclusion(s):

The FAST-PACE Toolkit was created using CeNR principles, in the aftermath of the Flint Water Crisis, by Flint-based organizations and expanded to equip communities across the state to respond to public health crises and systemic injustice. It provides a concrete roadmap for building equitable research partnerships built on respect and trust.

The integration of artificial intelligence (AI) into healthcare systems presents a profound transformation in medical and public health practices, including enhanced diagnostics, disease tracking, medical treatment, and patient outcomes. However, the rapid advancement of AI technologies poses significant challenges in establishing coherent and adaptable policy frameworks. This presentation explores the critical nexus between transparency, trust, and policy management in navigating the complexities of AI adoption within healthcare.

Addressing the incongruity between the expedited advancements in technology and the often sluggish pace of policy adaptation is challenging. After conducting a multi-state review of similar policies and proposed legislation, our team advocated for the establishment of a state-level government task force inspired by models such as a healthcare scope of practice task force and COVID-19 task force. This taskforce would work to monitor the evolving landscape of AI applications and propose timely policy interventions to state legislators to safeguard ethical principles and patient welfare, in addition to offering additional public protections to increase transparency through public hearings and discourse. Central to the effectiveness of this taskforce is the inclusion of multidisciplinary expertise from diverse backgrounds, including technology, health, legal, and government sectors.

Furthermore, we will discuss the process of disseminating of a policy brief to relevant stakeholders, both in the private and public sectors. By demonstrating the methodologies of stakeholder analysis, drafting a policy synopsis, and engaging with lawmakers, we hope to empower individuals to navigate the complex intersection of research and policy in their own issues from AI and beyond.

Background. The potential of ChatGPT in various applications, including its ability to analyze documents and generate recommendations, prompted this study to evaluate its effectiveness in a public health context, particularly for pediatric button battery injuries, a significant and growing problem.

Objectives. The objective of this investigation was to use ChatGPT to synthesize primary prevention recommendations by analyzing the most relevant documents in this healthcare setting. Given the increasing prevalence of button battery injuries among children and their serious and potentially fatal complications, there is an urgent need for effective prevention strategies.

Methods. Our approach involved the inclusion of four papers in ChatGPT-4: two clinical-epidemiological studies illustrating case series of button battery injuries in children, one focusing on the symptomatology of these injuries, and another presenting data on emergency room visits in the United States. Using ChatGPT's sophisticated processing capabilities, we sought to distill the essential information and generate comprehensive prevention recommendations.

Results. The process revealed ChatGPT's ability to accurately interpret complex medical data and its usefulness in public health advocacy, proposing measures such as improving product safety, raising awareness among practitioners, and promoting regulatory changes. In addition, the unique symptoms associated with button battery injuries synthesized by ChatGPT underscored the importance of rapid diagnosis and treatment.

Conclusions. This exploration highlights the potential of ChatGPT as a valuable tool for public health research and policy development, offering an innovative approach to address important health problems such as pediatric button battery injuries through artificial intelligence-driven analysis and recommendation generation.

Background: The U.S. Center for Disease Control and Prevention Social Vulnerability Index (SVI) based on Flanagan et al. (2011) research, compilates a score based on sixteen U.S. Census social indicators that creates maps at the census tract level to identify communities at-risk during a manmade or natural disaster. These disasters can include extreme weather events, chemical spills, or disease outbreaks. The SVI is an integral component in emergency preparedness utilizing geographic information systems and geospatial science with the ability to examine the four phases of a disaster - preparedness, response, recovery, and mitigation.

Methods: With the emergence of artificial intelligence and machine learning, our research uncovered that our Public Health Emergency Vulnerability Index (PHEVI) which is based on a decision-tree algorithm utilizing SVI secondary data at the census block level and geospatial clustering provides greater accuracy than the SVI to identify the highest risk communities during a disaster that already face health inequities. Results: Pilot testing of the PHEVI shows potential in examining health equity, manmade and/or natural disasters to identify communities at the highest risk of negative outcomes before, during and after an event. Furthermore, we also discovered that the PHEVI is conducive to examining higher disease outcomes (communicable and chronic) in marginalized communities.

Conclusions: The PHEVI is a newly created tool that can measure health equity and disasters to uncover smaller enclaves of populations at the highest risk during an emergency. With the increase of extreme weather events due to climate change and the inequitable conditions and health outcomes during the COVID-19 pandemic, we believe the PHEVI can advance scientific research that investigates complex neighborhood constructs and intersectionality that perpetuate health inequities in our communities.

Public Health Implications: Incorporating artificial intelligence, machine learning, and geographic information systems, provides a powerful tool to predict communities that are the highest risk of a negative health outcome while providing the geographic boundaries at the census block level. This will provide greater insights of smaller pockets of marginalized populations often missed with larger geographic units.