Data Analytics

Eagle’s data scientists use advanced analytics to help mission-driven organizations make smarter, faster decisions. We have a deep understanding of the complexities of government data systems and the privacy requirements of working within heavily regulated sectors.
By combining our analytics approach with broad subject matter expertise, we help our clients answer their most complex questions cost-effectively and thoroughly. For health and social programs and policies, we help our government clients forecast the major population trends that impact health outcomes—helping them become prepared and proactive.
We facilitate deep understanding of the analysis by packaging our findings in easily comprehensible formats that tell the stories contained in our clients’ data.

Capabilities

Eagle’s data scientists expertly employ a range of computational strategies to answer clients’ most urgent questions.

Data Strategy and Governance

Our DMBOK-informed strategy and governance approaches allow us to identify the data needed to better serve individuals, families, and communities—and how to acquire and manage it while safeguarding beneficiaries’ privacy and minimizing costs. We effectively manage the availability, usability, integrity, and security of the data in our clients’ enterprise systems.

Data Management

By following rigorous end-to-end data modeling, transformation, and quality assurance processes, we ensure clients’ data is accurate, reliable, and organized.

Data Engineering

We integrate complex legacy data systems and aggregate them to provide powerful new data repositories for optimizing our client’s programs for performance, maintainability, or cost. We design data repositories of any size, technology (e.g., SQL, NoSQL), complexity, and deployment strategy (e.g., on-premise, cloud, hybrid).

Advanced Analytics

We maximize the value of our client’s enterprise data by providing deep insights into the effectiveness, trends and costs of their programs using AI, ML, statistical and predictive analytics including supervised and unsupervised modeling techniques.

Dashboards, Visualizations, & Data Stories

We employ leading industry visualization platforms and support our clients’ decision-making through identification of important patterns and trends hidden in large, aggregated data sets.

Experience

Department of Health and Human Services (HHS), Substance Abuse and Mental Health Services Administration (SAMHSA)

  • Eagle leveraged spatial data science techniques using Python and Tableau to determine the influence of socio-demographic characteristics on admission to and discharge from substance use and mental health treatment services, which has provided valuable insight to federal program officers.
  • Eagle determined the facility and patient and characteristics most predictive of psychiatric inpatient readmissions, revealing facility policies that inadvertently lead to higher readmission rates.
  • Eagle developed user-friendly dashboards, dynamic reports, and visualizations for data reporting entities, grantees, and federal government users that allow monitoring performance measurements, data quality, and data results. We successfully collaborated with its federal clients to deconstruct the program requirements and use data analytics techniques across various processes catering to different stakeholders.
  • We regularly analyzed large national healthcare data sets and produced quarterly and annual reports on the findings, documenting national, regional, and state-level trends in mental health and substance use treatment.
  • We conducted a feasibility study to determine how a set of new data elements would perform when collected by states. Subsequently, we incorporated these new data elements into the data sets we collected, maintained, and analyzed for SAMHSA and state mental health and substance use agencies.
  • To maintain data quality, we applied data mining techniques such as frequent pattern mining that facilitated the identification of relationally dependent data trends and data anomalies that would otherwise be challenging to observe.
  • Eagle developed SAMHSA’s Behavioral Health Treatment Services Locator, a confidential and anonymous source of information for persons seeking treatment facilities in the United States or U.S. Territories for substance use/addiction and/or mental health problems.

Other Engagements

  • Eagle collaborated with the National Institutes of Health (NIH) to determine whether probabilistic linking in combination with deterministic linking matches patient records with greater accuracy than deterministic linking alone.
  • We incorporated the state-of-the-art technique of Precision Analytics into our ECIDS program evaluation design for state-level Early Childhood education programs.

Success Stories

Proving the Worth of Precision Behavioral Health

The Need: Machine learning has been underused in behavioral health relative to other areas of medicine. We sought to test for a federal client, SAMHSA, whether machine learning models could robustly predict which patient subgroups would respond best to different modes of substance use therapies—as with precision medicine more generally.

Our Response: We chose a SAMHSA data set with a rich set of patient and treatment variables, the Treatment Episode Data Set, incorporating data on most substance use patients in the U.S. We then trained Naïve Bayes, logistic regression, random forest, and CatBoost models to predict whether treatment was completed successfully. The two ensemble learning models—random forests and CatBoost—were the most successful at predicting treatment outcomes, with accuracy levels far surpassing the chance level of 50 percent.

The Benefits: We demonstrated that machine learning can robustly predict which individual patients will succeed in different treatment programs, in a proof of concept for precision behavioral health. Moreover, we uncovered the features—including length of stay in treatment and other factors—that were most crucial for successful prediction of treatment outcomes. This illustrated the utility of machine learning and other advanced analytics to develop recommendations for providers.

A data quality dashboard highlighting state partners’ progress in data collection

The Need: SAMHSA sought to analyze data quality profiles between years and states for the Treatment Episode Data Set (TEDS). TEDS is a new substance use and mental health data set developed previously for SAMHSA by Eagle, based on data contributed by states. Through this analysis, SAMHSA sought to encourage states and state representatives to improve their data quality.

Our Response: Eagle developed a data quality dashboard to easily analyze and highlight states’ progress in data collection. We built the dashboard with open-source technologies including Python, JavaScript, Plotly.js, and HTML/CSS. Each state’s data quality profile was customizable by state, year, data set, and admission type. A SAS program Eagle developed categorized the TEDS data as valid (no anomalies), unknown (missing), not collected (by a given state), or non-valid (anomalous with respect to other data in that patient’s record). This categorization was reflected in the dashboard, which allowed state partners to easily understand the level of their state’s data quality.

The Benefits: The TEDS Operations team utilized the dashboard quarterly to observe trends in data collection. The team was able identify the quantity of non-valid responses, and the type of non-validity. These tangible metrics allowed SAMHSA and the Eagle team to advise and guide state representatives to improve the quality of their data, a particularly important ongoing mission for the client.