Forcura Expedites Patient Care Using Generative AI on Amazon Bedrock

Founded in 2012, Forcura provides a HITRUST-certified solution for healthcare workflow management to simplify patient care transitions and support clients’ ongoing care delivery. The company serves over 900 healthcare providers, who represent about 1 million patients in the United States, each year.

Home-based-care organizations need to make referral decisions quickly to provide seamless patient care. Quick action also helps healthcare providers maintain strong relationships or fulfill their contractual obligations with referring organizations and health insurance companies. “Our clients sometimes need to review referrals for hospice patients,” says Laura Montesantos, vice president of product management at Forcura. “There is no time to waste for the kind of treatment being delivered.” Seeing an opportunity, Forcura’s engineering team set out to use generative AI to pull key information, such as patient demographics, clinical history, and requested services, into a concise summary.

Initially, Forcura built a prototype for referral summaries using OpenAI but then pivoted to using Anthropic's Claude models in Amazon Bedrock and chose Amazon Bedrock because of its scalability, advanced security features, and compatibility with other AWS services. Thus, Forcura can use the data it stores in Amazon Simple Storage Service (Amazon S3), an object storage service built to retrieve virtually any amount of data from anywhere, for its referral summary feature.

Also, by using Amazon Bedrock, Forcura has access to a broad choice of models from multiple AI companies. Rather than contracting with multiple third-party vendors for different use cases, Forcura can evaluate model capabilities and pick the perfect fit for every need. “Amazon Bedrock is like a five-star buffet of generative AI models,” says David Kerr, chief technology officer at Forcura. “We get immediate access whenever AWS adds a new model or feature to the buffet, which is really powerful for supporting our use cases.”

Forcura chose to use Anthropic’s Claude 3 models for its referral summary feature. Thus, the company can process thousands of referrals daily and generate summaries nearly instantly. As a HITRUST-certified healthcare organization that manages confidential patient information, Forcura also appreciates Anthropic’s commitment to model safety and security. Further, by using Amazon Bedrock, Forcura can keep all its data secure in its cloud environment under the AWS Shared Responsibility Model.

Forcura worked closely with representatives from AWS throughout the development process. “Forcura prides itself on being invested in its clients and their success,” says Kerr. “AWS lives up to those ideals as much as we do with flawless support, timely responsiveness, and a willingness to go beyond satisfying the initial request. Collaborating with an outstanding cloud service provider on this feature is a competitive differentiator in the market.”

In April 2024, Forcura started piloting the referral summary feature with the intention of testing it for 90 days with three clients. Because of the pilot’s immediate success, within 1 month of launch, the company made the feature available for early adoption to another seven clients. Then, Forcura opened the feature for general release in under 90 days from the pilot start date.

To maintain high accuracy of referral summaries, Forcura created a human-feedback loop and works directly with clients if any issues arise. “Before the implementation, almost every client asked about accuracy,” says Montesantos. “Some of the pilot clients indicated that their users were not going to trust the feature. After the implementation, it became a nonissue.” Some clients have reported that referral summaries have become an essential tool for supporting their operations.

Forcura’s clients expect innovation and appreciate that the company proactively developed the referral summary feature without waiting for a request from users. Taking advantage of the flexibility of Amazon Bedrock, Forcura can switch models smoothly and focus on new product ideas instead of devoting time to solving implementation challenges. “We’ve quickly identified and successfully met our clients’ needs with referral summaries,” says Montesantos. “We will continue using Amazon Bedrock to deploy new generative AI solutions to solve our clients’ major challenges and earn the right to be their vendor of choice.”