Blackstraw helps AMN Healthcare make successful matches with Azure

July 15, 2024

In the healthcare staffing industry, recruiters face a constant, pressing challenge: to assess the qualifications of skilled candidates and match them with open positions quickly and accurately. For specialized roles like travel nurses and doctors, where rapid placement is critical to meet the urgency and demand of the field, overcoming that challenge is paramount. But it's exactly the kind of challenge that Blackstraw, a data and AI consulting firm, is prepared to solve.

As a Microsoft partner with a team of more than 300 skilled data scientists and engineers, Blackstraw helps organizations modernize their data infrastructure and operations with Azure. More than that, they can use the data to build and deploy AI solutions tailored to their customers' business needs.

"We have our feet in both areas: AI and data engineering. That's the foundation of how we enable our customers to capture, process, and store the data they need," said Atul Arya, Founder and CEO of Blackstraw. "But once you've processed and stored the data, what do you do to make sense of it? How do you solve business problems using that data? That's where we leverage AI."

An automated, accurate matching system

Given Blackstraw's experience as a Solutions Partner for Data & AI (Azure), Digital & App Innovation (Azure), and Infrastructure (Azure), AMN Healthcare turned to the Microsoft partner for an AI-powered solution—one that would help streamline their operations, reduce the burden on recruiters, and implement safeguards to reduce bias throughout the hiring process.

After Blackstraw worked with AMN Healthcare to assess their current data architecture, they used Azure Machine Learning studio to develop an automated matching system that could work with their existing data sources—including an on-premises SQL Server and Azure CosmosDB. With a portal and mobile application that both job candidates and recruiters can use, the system consists of four core components:

• An advanced multi-model matching engine, which uses classification and regression algorithms to consider factors such as a candidate's eligibility, probability of completing an assignment, and relevance of the order to their work history.
• A multi-phase machine learning pipeline that predicts critical outcomes like successful credentialing and interview clearances. After assigning specific weights to individual probability scores, the process produces a comprehensive match score with nuanced and precise prioritization.
• Sourcing tools that can help reduce bias by using natural language processing and machine learning to analyze resumes and cover letters without relying on keywords or phrases related to demographic characteristics. The tools can also screen candidates based on objective criteria (including skills, experience, and education) rather than subjective factors (such as name, gender, and ethnicity). 
• A Match Explainability Dashboard that provides a comprehensive analysis of the factors that influence the model's decisions, promoting both transparency and trust.

The application can automatically capture and process any changes to job orders in real time—and extract information from resumes to match it against job descriptions. "Think of it like a dating app: What is the probability of a match? This is a different environment, but it's the same concept. You want to improve the probability of a match against the job that is posted," said Arya.

Improving that probability requires higher accuracy and precision—which is why Blackstraw regularly retrains and performs strict maintenance on the models. "We're not only detecting changes over time, but we're also preventing any loss of accuracy by using Azure Machine Learning studio," Arya said.

Blackstraw has over 100 successful Azure implementations under their belt, but when it comes to AI, their pragmatic view is what keeps them grounded. "We focus not only on what's possible but also what's not possible, and we're very honest about it," said Arya. "There is a very critical question to answer: How do you manage the inaccuracies of AI? That's a question that does not get asked or answered enough, but we understand it—and we make sure we address it."

Matches made in minutes

Once AMN Healthcare deployed the AI-powered automated matching system, candidate matches were found in as little as one minute after new orders were entered into the system. Overall, average processing times fell to under six minutes, thanks to Azure Kubernetes Service and Kubernetes Event Driven Architecture—a marked improvement from the days recruiters spent parsing through resumes.

Beyond improving the matching process, Blackstraw's solution empowers AMN Healthcare to make data-driven decisions and optimize the entire staffing process. "The solution offers valuable insights into the anticipated order volume and bill rates for different disciplines and specialties in the upcoming months, enabling informed decision-making and strategic planning," said Hagan. "By providing visibility into future staffing demands, [the solution] enhances the management of healthcare staff supply, which fosters more efficient and effective processes—getting clinicians where they are needed faster to care for patients."

Looking ahead, Blackstraw and AMN Healthcare plan to work together to continue refining the solution so it evolves with the company. "As a large organization, AMN Healthcare has many different applications that this model works with or is connected to," explained Arya. "Making sure it integrates with all the applications they use is going to be a long-term strategy."

Equipped with their vast experience and deep Azure expertise, Blackstraw built a solution that drives meaningful impact for healthcare staffing providers, recruiters, and healthcare workers—but most importantly, for the patients who need quality care.