Readers Write: Solving Healthcare’s $125 Billion Fax Problem

Solving Healthcare’s $125 Billion Fax Problem
By Thomas Thatapudi

Thomas Thatapudi, MBA is CIO of AGS Health.

In the six years since Centers for Medicare and Medicaid Services called for an end to faxing in healthcare, the industry remains stubbornly attached to fax communications. Fax-led communications solutions are prevalent today, as organizations continue exchanging more than 9 billion fax pages each year, representing about $125 billion in annual costs to the healthcare system.

It is a problematic practice in terms of data integrity, productivity, and efficiency. According to information compiled by DirectTrust, 30% of tests must be re-ordered due to lost faxes and 25% of faxes fail to arrive before a patient’s first visit. Faxes also often require manual indexing for integration into the EHR and other health information systems, a time-consuming process that requires human and financial resources that are hard to come by for many healthcare organizations.

Luckily, fax indexing lends itself to automation. In particular, generative AI (GenAI) and agentic AI excel at automating mundane and repetitive tasks. However, it is unrealistic to expect digital workers, such as AI agents that mimic human actions, to accurately index 100% of the faxes from the outset. Thus the best outcomes are realized when digital workers are paired with human counterparts who manage exceptions and handle specialized information and requests, at least until the digital workers have accrued enough on-the-job training via machine learning and deep learning models to take over higher complexity tasks.

Implementing a hybrid fax indexing model accelerates processing and eliminates the care delays that are caused by improperly managed faxes. It also alleviates the strain on increasingly scarce resources, reducing costs and freeing internal staff to focus on higher-value tasks.

However, achieving these outcomes requires careful orchestration of a workflow that seamlessly integrates digital workers (AI agents) and their human counterparts delivering on quality, timeliness, and accuracy.

Designing the Digital Workforce

The heart of a successful hybrid fax indexing strategy is a well-designed digital workflow model that helps orchestrate workflows between humans and digital workers. It starts with mapping the necessary technologies, a step that is best informed by shadowing human indexers to fully understand the process and map any unique needs. This information is also used to plan the implementation and conduct feasibility testing.

Like their human counterparts, digital workers are armed with an array of intelligence and automation tools, including optical character recognition (OCR), to analyze faxed documents and convert them into machine-readable text. They use natural language processing (NLP) models to interpret and manipulate the data contained within. GenAI is then leveraged to classify faxes based on the sender’s documentation format, determine its confidence threshold, and either index it into a documentation management system or EHR or divert it to the validation workflow for manual processing.

Machine learning allows digital workers to adapt to new document formats and categorize data according to providers’ templates and styles. Further, each processed fax enhances accuracy, efficiency, and capabilities while reducing exceptions.

Monitoring effectiveness is crucial to success. Establish clear KPIs, such as the volume of faxes indexed per day, indexing accuracy, turnaround times, and productivity levels to assess progress over time.

AI Grounded in Reality

While automated fax indexing is a relatively new entry into the burgeoning field of healthcare AI, it is quickly making an impact. One health system’s implementation of automated fax indexing has put it on track to save approximately $2 million in annual expenses. Automation has reduced the number of manual indexers that are required to process the health system’s fax volume, which allows key team members to focus on higher-value tasks while achieving a near-perfect accuracy rate and 24-hour turnaround time. As digital workers “learn” over time, the automation rate will increase, while the need for human intervention decreases, adding to the anticipated cost savings.

While it is unlikely that we will see a fax-free healthcare system in the near-term future, leveraging readily available automation and AI tools makes it possible to digitize the process and alleviate its associated cost, productivity, and patient safety burdens.

Automated fax indexing is yet another example of a thoughtful AI application that solves an age-old problem that, until now, has been stubbornly resistant to change.