The healthcare industry has long relied on faxing as a critical means of communication, but as healthcare data and systems have grown in scale, legacy fax-based workflows have become increasingly inefficient.

UTHealth Houston is home to two leading organizations at the forefront of digital healthcare innovation. The first is the McWilliams School of Biomedical Informatics, which is one of the largest biomedical and health informatics programs globally and has an entire department focusing on AI and data science. The second is the Center for Digital Healthcare Innovation, which is driving the development and implementation of clinical technologies to enhance patient care, education, and research across UTHealth Houston’s seven schools and two hospitals.

In this post, we will discuss how the McWilliams School of Biomedical Informatics and the Center for Digital Healthcare Innovation joined forces to address UTHealth Houston’s reliance on faxing. Although faxing is a critical means of communication for UTHealth Houston, the school’s legacy fax-based communication workflow is not able to efficiently handle the scale of today’s healthcare data and systems.

The teams—supported by UTHealth Houston Information Technology and HAQM Web Services (AWS)—built, deployed, and migrated a new solution called iDFax. This solution, which processes production data on AWS, harnesses the power of generative AI to revolutionize medical fax management. The team used a foundation model from HAQM Bedrock, which delivered a highly accurate, cost-effective, managed service experience on AWS. The solution is deployed in a secure landing zone environment, which provides security guardrails and centralized observability to meet compliance requirements.

iDFax overview

iDFax brings speed, accuracy, and efficiency to faxed data processing. Built using HIPAA eligible services, the solution is capable of scaling dynamically using AWS to meet the needs of various departments across UTHealth Houston.

iDFax offers a range of features and benefits, including:

• Time reduction: iDFax reduces fax processing times by more than 50%, from the legacy system’s 82-150 seconds per fax to just 28-68 seconds.
• Efficiency gains: The solutions automates key processes such as document categorization, splitting, and de-duplication. Its optical character recognition (OCR) capabilities—such as advanced image correction and handwriting recognition—provide over 95% accuracy.
• External system integration: iDFax enables a seamless connection of fax content to patient records through automatic extraction of physician and patient data for identity retrieval and record assignment with Epic, an electronic health record (EHR) software system.
• Real-time dashboards: A dynamic monitoring dashboard provides users with actionable insights into system usage, document types processed, and overall operational efficiency.
• Compliance and security: The system is aligned with HIPAA regulations, safeguarding sensitive healthcare communications.

Since its launch in June 2023, iDFax has successfully processed over 220,000 faxes for pilot sites at UTHealth Houston, as shown in the following diagram. It is now being deployed to the rest of UT Physicians clinics and will be able to process over 1 million faxes per year, resulting in cost savings of more than USD$2 million per year.

Figure 1: Faxed data processing metrics.

High-level architecture

As shown in the diagram below, the iDFax architecture includes the following steps:

• The inbound electronic fax data is programmatically uploaded to AWS at near real-time speed over the AWS Direct Connect links to an HAQM Simple Storage Service (HAQM S3) bucket in a secure AWS account.
• The data is processed by iDFax and packaged as docker containers running in the HAQM Elastic Compute Cloud (HAQM EC2) nodes from the messages in the HAQM Simple Queue Service (HAQM SQS). Data processing metadata is stored and tracked in HAQM DynamoDB tables.
• iDFax’s classification tasks and generative AI-driven analysis are handled using a foundation model on HAQM Bedrock.
• The results are either routed to integrated systems like Epic.com or stored in the AWS environment for data retention and retrieval.

Figure 2: A high-level architecture diagram of the solution.

Conclusion

The iDFax solution built and deployed by UTHealth Houston is an excellent example of how generative AI can be used to solve real-world problems at scale and drive cost savings. By leveraging efficient workflow designs and AWS cloud services, UTHealth Houston has transformed the medical fax management process, creating a powerful solution that can be used by healthcare providers to improve productivity, reduce costs, and enhance care delivery.