PROJECTS | MICOM

MICHIGAN PUBLIC HEALTH INTEGRATED CENTER FOR OUTBREAK ANALYTICS AND MODELING

OUR PROJECTS

During emergencies and infectious outbreaks, MICOM plays a crucial role by modeling the spread of diseases and providing analytical tools that enable decision-makers to make swift, data-informed choices.

Our team is also involved in various initiatives, including wastewater monitoring, violence prevention, firearm-related injury reduction, respiratory illness surveillance and forecasting, and fostering collaborations to shape the next generation of public health professionals.

D.L.I.V.E. (Detroit Life is Valuable Everyday)

D.L.I.V.E is a hospital-based violence intervention initiative that works with youth and young adults who have sustained acute, intentional violent trauma. The MICOM team is working with D.L.I.V.E. as they transition their record-keeping to a database system. With this transition, the D.L.I.V.E. team is developing dashboards to visualize their work and impact over time, and to improve efficiency and identify potential strengths and weaknesses. The MICOM team is building these dashboards, working with the D.L.I.V.E. team to build their fluency in understanding the data and systems that exist, and once the infrastructure and ground-work is done, will be developing models and analytics to identify trends, anomalies, and forecast needs in the care population of D.L.I.V.E.

MiLighthouse Dashboard V2

MiLighthouse was a rapid response to the Covid-19 pandemic to assist private and public health officials. Originally created as a private dashboard that supported public health efforts by providing authorized local and state officials with vaccine data visualizations to track Covid-19 vaccine uptake, assist in outbreak investigation and resource allocation decisions. Dividing the map into the 82 Michigan counties, this tool helped officials respond rapidly to declining vaccination rates through tracking micro-level patterns of vaccine uptake, as well as track age, race, gender and location metrics to identify disparities in COVID-19 impact and vaccination rates. Based on popular demand, MiLighthouse was eventually opened out to community organizations who were also working towards educating their communities and advocating for vaccine uptake. Noting this demand for this tool as well as access to more information, MiLighthouse’s second iteration is focused on tracking additional vaccine uptake rates for diseases that occur seasonally such as influenza or diseases with long-term impacts such as MMR as well as opening out the dashboard to other key stakeholders such as school administrators.

Washtenaw County Firearms Dashboard & Analysis

Washtenaw County Health Department (WCHD) has begun looking to improve the data available around deaths and injuries related to firearms in the county. As part of this, they are also working alongside the Washtenaw County Sheriff’s Office under a grant from the Michigan Department of Health and Human Services (MDHHS). The MICOM team has been involved initially in the information gathering process, working with WCHD to understand the data streams available and how they might be used in modeling and analytics work to inform analyses of how well intervention efforts are going and to better understand where intervention efforts can be focused to have the largest impact.

Wastewater Testing Results Dashboard

The Wigginton-Eisenberg Laboratory (WEL) regularly tests for nine public health-relevant organisms in samples from five wastewater treatment plants across southeast Michigan. They retain the capacity to turn testing on for one additional organism, as requested by public health officials. In addition, the laboratory develops new wastewater based epidemiology tests and methodologies for anticipated up-and-coming public health-relevant organisms. The MICOM team works closely with the WEL in communications with public health officials and wastewater treatment plant stakeholders. They also maintain and continue to develop the WEL’s reporting dashboard, which not only reports the testing results but also provides analysis for determining the level system for each organism, in order to inform all dashboard users regarding their potential disease risk. Future work includes integrating the data that the WEL produces into forecasting & risk assessment models developed by the Modeling and AI Cores of MICOM, and then packaging the results of those models back into real-time reporting onto the dashboard.

Screen Shot 2024-08-29 at 9.11.45 AM.png

Local Health Department Interviews

Local Health Departments (LHDs) routinely engage in efforts to prevent potential outbreaks or respond to acute outbreaks but there is limited understanding of their workflow or how they use vaccine data to respond. The availability of large quantities of vaccination-related data through electronic health record (EHR) and other systems creates both opportunities and challenges for LHDs to use vaccine data in their prevention and response efforts. For this study, we use semi-structured interviews to examine LHD approaches to outbreak prevention and response and identify barriers and facilitators to vaccine data used in these efforts across the State of Michigan.

School Administrator Interviews

To support the MICOM tool-building teams, the qualitative team conducts regular interviews with stakeholders to develop tailored tools. This project is a part of that initiative and aims to understand vaccine uptake dynamics in Michigan, with a focus on how data is collected by school districts and used by local health departments. Increasing reluctance to maintain up-to-date vaccine schedules is making outbreaks more likely, highlighting the need for improved data collection and tool support. By examining the workflows of vaccination data entry and outbreak procedures, the project addresses the challenge of declining vaccination rates post-COVID-19.

Southeast Michigan Health Department Coding Workshops

Since August 2023, the MICOM team has developed content for and run five workshops on R coding topics for the epidemiologists at Washtenaw County Health Department. The focus of these has been to build the skills necessary to build dashboards and analytic tools in-house, as well as to shed light on the automation capabilities and quality assurance and quality control implementation that is possible using code. As these individuals have participated, they have also talked to their peers at other health departments, and as of July 2024, future workshops are being planned to include Livingston, Ingham, and Eaton County Health Departments

Michigan Respiratory Forecasting

Screen Shot 2024-08-08 at 12.15.52 AM.png

Complex location and time data are involved in predicting the spread of respiratory illnesses. Despite significant advancements in our understanding of how these diseases propagate among populations, several major challenges persist due to connected/related factors such as contact patterns, vaccine adoption, biological processes governing pathogen evolution, and changes in data collection procedures. Leveraging our experience in supporting national and state-level forecasts for COVID-19 and influenza through the CDC's COVID-19 Forecast Hub and FluSight initiatives, we are now adapting our models across counties in Michigan. Our methods range from mechanistic epidemiological models—based on ordinary differential equations (ODEs) or agent-based models (ABMs)—to AI-driven models using deep learning, and hybrid techniques that bridge both approaches. With these models, we aim to provide a variety of actionable epidemic forecasts to partners at the Michigan Department of Health & Human Services, helping them design effective public health policies and make better supply chain decisions.

Anomaly Detection

The goal of this project is to use anomaly detection methods (including spatial statistics, change point detection, time series analysis, and ML/AI methods) to develop an automated pipeline that will detect concerning signs in local surveillance data and flag them for local public health. Concerning signs will include indications that an outbreak might be starting (either locally or more distributed spatially) and detection of data quality issues (such as incorrect data entry, missing data, underreporting). Initially, we plan to develop these analytics as a standalone system, but eventually the goal is to have this system operate as a layer on top of the Michigan Disease Surveillance System (MDSS) to detect anomalies across a wide range of diseases/pathogens.

Respiratory Viruses in Decedents

Screen Shot 2024-08-08 at 12.16.09 AM.png

Current respiratory virus surveillance relies primarily on medically attended, symptomatic cases, which may distort the true patterns of respiratory virus infection. Estimating the underlying respiratory viral infection patterns, regardless of the symptomatic status, and how infection is associated with social vulnerability, may help us to better intervene and address issues of health equity, particularly racial & ethnic disparities. By testing the deceased individuals in medical examiners’ offices for respiratory viruses regardless of cause of death, we may be able to better understand these underlying patterns of infection. For this project, we partnered with the Wayne County Medical Examiner’s Office (covering Wayne & Monroe Countries of Michigan including Detroit), who tested nasopharyngeal swabs from more than 4,000 deceased (2020–2022) for a panel of respiratory viruses, including adenovirus, SARS-CoV-2, rhinovirus, and respiratory syncytial virus. We matched the addresses of the deceased to the Social Vulnerability Index (SVI) and Area Deprivation Index (ADI) and estimated associations between these metrics and respiratory virus infection. Our results suggest that the Minority Status & Language subtheme of SVI is generally the most informative for predicting respiratory virus infections and could be used to prioritize health-equity-focused distribution of public health resources.

MI-Support is a program offered through the Michigan Public Health Integrated Center for Outbreak Analytics and Modeling (MICOM) that aims to provide support and capacity building in response to urgent and/or under supported modeling and analytics needs for local and tribal health departments across the state of Michigan.

The MI-Support team will be deployed to assist with data analysis projects to enhance preparedness for public health emergencies and strengthen analytics and modeling capabilities statewide. (coming soon early 2025!)