MIDAS at a Glance
Affiliated Faculty
External grants enabled
MIDAS funded research projects
External Partners
MIDAS 2021 Highlights
MIDAS continued to lead and facilitate responsible, ethical, and reproducible Data Science and AI research.
Ethical Data Science and AI
MIDAS Leads Multi-University Collaboration on Data Equity Systems
A team led by MIDAS Director, H.V. Jagadish, is establishing an information system to minimize the misuse and misinterpretation of data. Dr. Jagadish and MIDAS affiliated faculty, Tayo Fabusuyi, Robert Hampshire and Margaret Levenstein, along with colleagues at other universities, have completed initial work on methods to identify data and software bias that negatively impact equity in critical domains, such as mobility, housing, education, economic indicators and government transparency. The ultimate goal is to develop a system that will take potentially biased data from both public and private sources, and make them bias-adjusted and analysis-ready. This project will establish a Framework for Integrative Data Equity Systems (FIDES) for the study of systems that enable research on sensitive data while preventing misuse and misinterpretation.
Fair Representation in Arts and Data
“Fair Representation in Arts and Data” is a year-long collaboration between data scientists, artists and museum curators that was funded by the President’s Arts Initiative. The project analyzed the collection at UMMA using two widely used face detection algorithms to reveal biases in the algorithms and in museum artwork acquisition practices. Museum visitors can now get a first glimpse at the initial research findings through “White Cube, Black Box”, a display at the “YOU ARE HERE” exhibit at UMMA.
Reproducible Research
MIDAS Promotes Reproducible Data Science
A significant challenge across scientific fields is the reproducibility of research results, in both the narrow sense of reproducing the results with the same data and analytical methods, and in the broader sense of making the production of scientific knowledge transparent, traceable, and trustworthy. Data science research with increasingly complex data and long project pipelines faces reproducibility challenges every step of the way. MIDAS organized the 2020 Reproducibility Challenge to highlight high-quality reproducible work through examples of best practices across diverse fields and built an online resource collection. We now focus more intensely on actionable solutions that can be shared with other researchers to improve reproducibility. Our goal is to facilitate the development, validation and dissemination of tools and training to help make data science research more reproducible across many application domains.
MIDAS will host a “Reproducibility Day” celebrating the 2021 challenge in early 2022. Learn more about the 2021 Reproducibility Challenge here. For tools and resources relating to reproducible data science and research please visit the MIDAS Reproducibility Hub.
MIDAS provided seed funding, enabled data access, and catalyzed groundbreaking ideas and teams that cut across disciplinary boundaries.
Support for High-Impact Research
Data science and AI methodologies are transforming many research fields, blurring disciplinary boundaries and driving scientific discoveries through new perspectives. MIDAS affiliate faculty researchers are using cutting-edge methodologies for a variety of “endpoints” – significant scientific and societal challenges. MIDAS helps them develop new ideas and interdisciplinary teams through research incubation activities, pilot funding and dataset access. The examples below exhibit our faculty members’ creative and impactful research toward three “endpoints”
- Dr. Nancy Fleischer (Epidemiology), Dr. Karandeep Singh (Learning Health Sciences) and other MIDAS faculty members use public health survey data and medical record data to uncover inequity in COVID patients’ experiences, the care they receive, and long-term impacts.
- Dr. Rahul Ladhania (Health Management and Policy) and colleagues are improving Machine Learning methods to better predict persistent opioid use for all race and gender groups
- Dr. David Jurgens (School of Information) and colleagues use advanced Natural Language Processing methods to improve racial equity in doctor-patient interactions.
- Dr. Jon Zelner (Epidemiology) uses geospatial data to provide guidance to the State of Michigan on the unequal infection risks of different racial groups.
- Dr. Neil Carter (Environment and Sustainability) and colleagues combine many types of data and use probabilistic modeling to figure out how illicit wildlife trade networks operate
- Dr. Paramveer Dhillon (School of Information) and colleagues use Graph Neural Networks to analyze environmental sensor data to predict air quality during natural disasters.
- Dr. Ayumi Fujisaki-Manome (Climate and Space Sciences and Engineering) and colleagues develop machine learning models for ice forecasting to support the Great Lakes shipping community.
- Dr. Joshua Newell (Environment and Sustainability) and colleagues use geospatial data and social media data and Deep Learning to identify communities that are vulnerable to flooding and heat, and plot climate change vulnerability maps for the State of Michigan.
- Dr. Libby Hemphill (School of Information) and colleagues analyze social media data to identify extremist groups.
- Dr. Rada Mihalcea (Computer Science and Engineering) and colleagues develop AI algorithms based on linguistic analysis of media and online data to automatically detect fake news.
- Dr. Sarita Schoenebeck (School of Information) and colleagues analyze large volumes of social media data and design large-scale public experiments to reduce harassment and the spread of misinformation online.
- Dr. Michael Traugott (Center for Political Studies) and colleagues analyze media reports and social media data to determine the political sentiments of the public during major political campaigns.
Helping Researchers Secure Major External Grants Through Pilot Funding
The 2021 round of Propelling Original Data Science pilot grant program supports 17 research projects, involving 36 (co-) Principal Investigators from 7 schools / colleges at Ann Arbor and Flint. Since 2015, MIDAS research funding has enabled a total of 52 research projects, and the project teams have gone on to secure more than $100,000,000 in funding from external grants.
Developing Thematic Research Areas
Machine Learning in Drug Discovery
Through the NSF’s Industry–University Cooperative Research Centers (IUCRC) program, MIDAS Associate Director Dr. Kayvan Najarian is leading the effort to develop the Center for Data-Driven Drug Development and Treatment Assessment (DATA). It brings together data scientists, mathematicians, biomedical researchers, and healthcare providers to produce reproducible methodologies that will make a broad impact on drug discovery and biomedical applications of data science. By forming collaborations with industry, government, and community partners, the project will enable the dissemination and translation of research into impactful products and services for the betterment of society. DATA will also build a world-class data science community that is inclusive and promotes diversity at all stages of the academic pipeline.
Training
Michigan Data Science Fellows
This postdoc program at MIDAS provides outstanding young researchers with intensive data science and AI experience as they ready themselves for independent research and faculty positions. Upon completing the program, our first cohort of Fellows have gone on to faculty positions at Penn State, Carnegie Melon, the University of Texas-Austin, and right here at Michigan, as well as careers at Ford Motor Company and General Motors.
Teaching Data Science to Faculty Researchers
The Introduction to Data Science for Biomedical Researchers Bootcamp is the first bootcamp that MIDAS offers to help researchers adopt data science methods. The intensive training started the participants on the journey to learn machine learning and its clinical applications, incorporate data science methods in their research programs and grant proposals. Following the success of this event, MIDAS is now developing similar bootcamps for researchers in other domains.
MIDAS collaborated with academia, industry and public-sector partners to maximize the scientific and societal impact of data science and AI research.
Academic Partnerships
Future Leaders Summit
MIDAS organizes the annual Future Leaders Summit (previously known as the Data Science Consortium) to promote collaboration among data science institutes and to foster the careers of the future research leaders. Participants are PhD students and postdocs from universities with mature data science programs as well as those beginning to build such programs, from major research universities to minority serving institutions. More than half of all participants are women and underrepresented minorities. The theme of the 2022 Summit is “Responsible Data Science and AI.
Data Science Coast-to-Coast
MIDAS partnered with 6 peer academic data science institutes in 2021 to host the Data Science Coast to Coast (DS C2C) seminar series meant to foster a broad-reaching data science community. In the first half of 2021 five seminars were held, each featuring one faculty member and one postdoctoral research fellow from the partnering universities. DS C2C is the launching point for continued research discussion and fruitful future collaborations.
Industry Collaboration
Rocket Companies Michigan Data Science Fellow
MIDAS collaborates with a large number of companies each year for joint research and talent recruitment. This year, we launched the inaugural Rocket Companies Data Science Fellows position. The Fellow will work in an intellectually vibrant environment, and build collaboration with other Fellows, MIDAS faculty members, and the Rocket Companies data science team. The research focus area, jointly decided by MIDAS and the Rocket Companies, is ethical data science and AI, broadly defined, including considerations of fairness, diversity, explanations, and reproducibility.
Data Science for Social Good
Enabling Digital Inclusion
MIDAS is the first academic partner in Microsoft’s digital inclusion initiative in metropolitan areas, including Detroit. The ultimate goal is to improve the quality of life and upward mobility through universal broadband internet access – particularly for underserved communities. Jing Liu, MIDAS managing director, leads a research team to support Detroit’s digital inclusion initiative, providing data analytics to pinpoint residents’ needs, track the initiative’s progress, and examine the long-term benefits of digital inclusion.
Supporting Student Learning for Native American Tribal Nations
A team of U-M undergraduate data science student volunteers are developing a database and a website, at the request of three Native American Tribal Nations in Michigan. The Tribes’ goal is to modernize their methods of data storage, query and reporting. Their student data was scattered on Excel spreadsheets, older databases and on paper, with no easy way to sort the data, let alone analyze them to track students’ performance and career outcomes, and to assess the need for support. The students, supervised by Tayo Fabusuyi (MIDAS affiliated faculty member), work with the Tribal educational directors to build the database and to tailor its functionality to the specific needs of each tribe.
Assessing Detroit Police Athletic League’s Youth Programs
The Detroit Police Athletic League (PAL) has been offering athletic programs since 1969 to school-age children from low-income, under-served families. MIDAS affiliated faculty member, Brady West, leads a team to analyze PAL survey data to assess the impact of its programs on participants and their families, and how these programs improve the relationship between the police force and the community. The project is funded through a generous donation from Mark and Elieen Petroff.
Celebrating the achievements of the researchers and students in the MIDAS community.
U-M, AKU Collaboration will Establish the First Data Science Hub in East Africa
Internal Medicine Professor and MIDAS Affiliate faculty member, Dr. Akbar Waljee, in partnership with The Aga Khan University, has been awarded a $6.5 million NIH grant to establish a data science hub in East Africa to harness AI, machine learning, and other cutting edge data science to improve healthcare delivery and outcomes in local communities.
U-M Lab Awarded $13.2M Grant to Help Thwart Substance Abuse Disorders
Drs. Inbal (Billie) Nahum-Shani and Daniel Almirall, co-directors of the Data Science for Dynamic Intervention Decision-Making Lab (d3lab) at the Survey Research Center and MIDAS affiliate faculty, received a $13.2 million NIDA P50 grant awarded to MIDAS affiliate faculty members to launch the Center for Methodologies for Adapting and Personalizing Prevention, Treatment, and Recovery Services for Substance Use Disorder and HIV (MAPS Center).
U-M Students Win the 2020 ProjectX Machine Learning Competition
ProjectX is the largest undergraduate Machine Learning competition in the world. The 2020 U-M team (Sanjeev Raja, Zhizhuo Zhou, Anh Tuan Tran, Amanda Yao, Ziwei Tian and Eric Chen), selected by MIDAS and with Dr. Sindhu Kutty (Computer Science and Engineering) as their mentor, competed against 22 other teams from the U.S., Canada and Europe. They won the Weather and Natural Disaster Prediction category, earning a $20,000 prize and presented their research, “The Devil is in the Details: Spatial and Temporal Super-Resolution of Global Climate Models using Deep Learning.” at the University of Toronto’s AI Conference.
The 2021 U-M team is working on a healthcare research project, with MIDAS Senior Scientist, Dr. Jonathan Gryak, as their mentor.
MIDAS Student Leadership Board
Representing multiple schools, colleges, majors, and all three degree levels (Undergraduate, Masters, and Doctoral), the MIDAS Student Leadership Board, with its 10 members, organize events for career building, diversity, equity and inclusion and serve as the bridge between data science student community and MIDAS leadership. The Board organized a Women+ Data Science Skill Building series in 2020 and 2021 and career meetings with industry representatives. In the Winter 2022 Semester the board will continue the Women+ Data Science series and organize data science hackathons.