Advanced Analytics, Artificial Intelligence, and Machine Learning
IMPAQ uses advanced analytics, artificial intelligence, and machine learning to help clients better understand their markets, make informed decisions, and prepare to deliver on the demands of the future.
In developing and implementing new policies, optimizing operations, or achieving other goals involving complex systems, having a large volume of high quality data allows for better decision making.
Policymakers and program managers often have large amounts of data, but are not able to use it to its full potential. Unlocking valuable insights from large-scale data requires new approaches to preparation, management, and analysis, including harnessing the power of artificial intelligence (AI) and machine learning.
Making the Complex Clear and Actionable
IMPAQ's advanced analytics experts specialize in developing innovative solutions that combine the latest analytics methods and techniques, knowledge of advanced computational theories, and an intimate understanding of human-centered design.
Whether developing a highly accurate predictive model driven by AI, or discovering rich insights through Natural Language Processing, our approach is custom-tailored to our clients’ needs and grounded in real world evidence and expertise.
With our support, clients better understand their customers, learn how external forces can affect their operations, and are prepared to make informed, evidence-based decisions.
Data Science and Artificial Intelligence
With a team of dynamic mathematicians, statisticians, data scientists, and computational economists, IMPAQ’s Data Science and Artificial Intelligence experts help clients such as the Centers for Medicare & Medicaid Services, the U.S. Department of Labor, the U.S. Food and Drug Administration, and the AARP Foundation, benefit from machine learning prediction models, microsimulations, and text processing tools.
Our experts apply advanced analytic methods including machine learning, natural language processing, and topic modeling, to address public policy questions and solve complex challenges. Our efforts have helped clients:
- Predict physician participation in alternative payment programs.
- Simulate worker leave-taking behavior.
- Forecast labor demand shortages.
- Identify nursing facilities with a high likelihood of elder abuse.
Our team of creative thinkers uses state-of-the-art, cloud-based resources to meet our clients’ demands for computationally intensive methods, such as the analysis of millions of government documents, hundreds of thousands of public comments, point-of-sales transaction records, job postings, social media data, and other large depositories of administrative and survey data.
Statistical Consulting and Data Management
Our team of advanced statisticians and computational scientists support clients in the design of program pilots, demonstrations, and evaluations, as well as in the policy applications of modern statistical methods, such as Bayesian analysis, stochastic system dynamics, agent-based models, and network analysis.
We also support clients’ traditional statistical consulting needs, including survey design, sampling design, and all survey methodology requirements. Our experts help clients:
- Build and manage scalable data management services, such as Amazon Web Services (AWS).
- Generate synthetic data.
- Perform other data-intensive services that require secure and serverless, cloud-based, data storage and management environments.
Our advanced analytics capabilities include:
Create systems that learn with you.
IMPAQ's advanced analytics team uses cutting-edge and custom artificial intelligence (AI) and machine learning (ML) concepts to help our clients explore a wide range of social issues and policy challenges.
The team combines non-traditional data sources with advanced AI and ML techniques to reveal actionable insights. Examples of the non-traditional data sources we work with include unstructured text data and structured transactional, operational, and administrative data.
Notable examples of our AI/ML work include:
- Developing a ML-based model to help the Centers for Medicare and Medicaid Services (CMS) predict participation in the Merit-based Incentive Payment System (MIPS).
- Helping the U.S. Department of Labor predict leave taking behavior and simulate leave usage under alternative policy scenarios.
- Creating a classification model to help CMS determine when Affordable Care Act (ACA) applications will expire due to income issues.
Our experts have extensive experience performing AI/ML analyses with Python, Scikit-Learn, and PySpark, as well as with platforms such as Amazon Web Services EC2/Sagemaker and Azure Databricks.
Helping Detect Abuse in Nursing Homes using ML and NLP
IMPAQ's AI/ML experts applied ML, Natural Language Processing (NLP), and text analytics tools to CMS Form CMS-2567 data to identify nursing homes and skilled nursing facilities with a high likelihood of elder abuse. The team also developed an interactive user interface to query the text data for important information.
Generate insights from human expression.
Natural Language Processing (NLP) makes it possible for computers to understand the complexities of the human language, accounting for inconsistencies, errors, and semantics. An advancement from keyword-based approaches that tell us what text-based data says, NLP can tell us what it means.
IMPAQ's team of data scientists are well versed in NLP methods and applications to public policy. We have helped numerous clients better understand and analyze large volumes of unstructured text, and in turn, helped them glean actionable and succinct insights from these findings.
Helping the Department of Labor Respond to Public Comments on Proposed Regulations
Using IMPAQ's data processing and integration techniques, including NLP techniques, the U.S. Department of Labor (DOL) efficiently processed nearly 400,000 comments, letters, and signature campaigns submitted to the DOL by citizens as a way to provide feedback on proposed policies. IMPAQ developed automated processes to generate timely and actionable updates.
Process and maintain large-scale data with accuracy.
We offer large-scale data processing, integration, and maintenance services that focus on data as the primary output, including real-time filtering and the creation of synthetic data. Our experts bring a strict adherence to data validation and a deep understanding of how data attributes affect analytics and machine learning.
Our capabilities include data management within CMS systems; supporting clients with the CMS eXpedited Life Cycle (XLC) process; validation and verification; and data quality assurance and testing.
Working with AHRQ to Develop a Synthetic Health Care Database
IMPAQ is working with the Agency for Healthcare Research and Quality (AHRQ) to create a comprehensive synthetic health care database. Designed to promote price and quality transparency, the database will enable researchers and entrepreneurs to identify inefficiencies and opportunities for improvement in the health care system.
Plan for future opportunities and mitigate future risks.
Powered by machine learning, our data analytics services include:
Answer critical questions about your operations, such as the needs and pain points of your target audience and trends affecting user behavior.
We offer pattern recognition, anomaly detection, and real-time monitoring tools, using the latest techniques from the fields of information geometry and topological data analysis.
Prepare to meet the demands of the future.
By applying the latest methods in predictive analytics, including spatial-temporal forecasting, time-series forecasting, statistical process control theory, and more, IMPAQ helps clients anticipate future challenges and opportunities.
Helping CMS Predict Merit-based Incentive Payment System (MIPS) Participation
IMPAQ used advanced prediction models to help the Centers for Medicare and Medicaid Services (CMS) predict which physicians would participate in MIPS, a new option within CMS' Quality Payment Program. The results helped CMS make highly accurate predictions and informed improvements to program design.
Harness descriptive and predictive analytics to inform decision-making.
Our static and dynamic optimization methods allow clients to make more informed decisions by understanding both the potential short-term and long-term consequences of choices they make today.
Health Care Payment Model Analytics
Understand constantly evolving payment policies, make cost-saving decisions, and improve quality of care.
IMPAQ has used claims data to analyze the effects of proposed payment policies on government programs, such as Medicare, Medicaid, health insurance exchanges, and others. Through health care payment model analytics, we can help clients make informed policy decisions and better serve beneficiaries.
Consider human factors to improve systems, test solutions, and prepare for changes before they occur.
By understanding and experimenting with complex relationships within a system, policymakers and program managers can predict how changes, such as new policies or trends, may affect systems over time. Unlike models relying strictly on data and pattern recognition, simulations include human knowledge of cause and effect relationships in the real world.
Model the impact of policies and programs on a target population.
Conventional microsimulations use a combination of population data and pre-defined rules, based on survey or administrative data. These factors can guide a system to suggest the most likely decisions individuals will make on a given issue.
Agent-Based Modeling (ABM) Simulation
Understand complex, system-wide behaviors.
We can begin to understand how individual actions evolve to become group behaviors by analyzing how people interact with their environments and one another. We can also predict how new policies and other changes will influence group behavior in the future.
System Dynamics Simulations
Predict and plan for shocks to complex systems.
Our experts can simulate the long-term impacts of changes within a complex system by organizing individuals into groups based on their behaviors or other attributes and studying the flow into and out of these groups over time.
Simulating the Use of Paid Leave under Different Leave Policy Scenarios
IMPAQ helped the U.S. Department of Labor (DOL) use survey and administrative data on the use of paid leave to develop a Leave Simulation Model to test different scenarios of paid leave programs and estimate the implications and costs.
Create accurate, consistent, and streamlined processes.
In times of tightening budgets and heightened scrutiny on the use of public funds, today’s policymakers and program managers need to be more efficient. Businesses, too, want to replace redundant or outdated processes to make better use of time and money.
IMPAQ's experts in process improvement collaborate closely with our data automation teams to provide measureable improvements in efficiency and quality for private industry and government clients. We take a complex operation and break it down into simple, automated steps, improving accuracy, consistency, and timeliness of operations.
Streamlining Operations for the Newark Workforce Investment Board
IMPAQ worked with the Newark Workforce Investment Board to develop a data management system that automatically compiled and updated data from a variety of sources for use by the Board. We also designed an intuitive user interface in Java for decision makers to use the data to improve operational efficiency and maximize outcomes of the Board’s workforce development programs.
Unlock insights from large datasets that would have been impossible to use in the past.
Program managers face a dilemma: large datasets increase the accuracy of machine learning and simulation, yet they are often prohibitively expensive to store and process. IMPAQ's cloud-based multiprocessing and high performance computing capabilities bridge this gap and enable our clients to use precise analytic tools and achieve better predictive models.
Curbing Tobacco Use through Big Data
IMPAQ worked with the US Food and Drug Administration (FDA) Center for Tobacco Products to develop research protocols for the application of high-performance computing in their research and regulatory activities.
Tailor-made products and tools make it easy to learn from data.
We empower clients and end users to generate actionable insights from data. Clients improve their products, increase the quality of their services, and design efficient operations using IMPAQ's custom-built data systems and applications.
Helping New Yorkers Find the Right Insurance Plan
IMPAQ developed the NYPlanCosts Calculator—a first-in-the-nation price transparency platform that helps consumers with chronic conditions compare out-of-pocket costs when shopping for health insurance plans.
Driving insight through visualization.
IMPAQ helps clients discover patterns and relationships that can lead to new insights using effective data visualization, reports, and interactive web-based dashboards.
Relevant Solutions Rooted in Policy Expertise
Our data visualization experts work in close collaboration with our policy experts, economists, clinicians, and data scientists to ensure that our products are meaningful to the current policy environment and can be used to drive real world action.
Interactive Tools to Empower Decision Makers
Our data visualization experts bring a deep understanding of human-centered design to develop intuitive and actionable solutions. Data dashboards designed by IMPAQ are secure, as well as fully interactive and dynamic, allowing program managers and other decision makers to seamlessly aggregate, parse, explore, and analyze data.
Visualizing Social Determinants of Health
IMPAQ developed the SDoH Dashboard Action Tool to help community-level decision makers visualize data to improve health care access, quality, and equity. Recognized by the Agency for Healthcare Research and Quality (AHRQ) in its innovation challenge, the tool allows for nuanced geographical analyses across a broad array of variables that empower users to identify population-level challenges and interventions to address these challenges.
To discuss how we can help you collect, analyze, and interpret data, please contact Dr. Minh Huynh, Vice President, Advanced Analytics.