From fax machines to GenAI, are hospitals/health systems ready?

By Bill Fera, M.D., principal, Deloitte Consulting LLP

Generative artificial intelligence (AI) is the latest step along the journey toward the Future of Health that Deloitte first outlined in 2019. For hospitals and health systems, this emerging technology is poised to radically transform clinical workflows, drive new efficiencies, enhance the patient experience, and improve clinical outcomes.

For an industry that still sometimes relies on phone lines and fax machines to transmit information,¹ the adoption of generative AI might be seen as a technological leap. I see it as more of an evolutionary progression than a revolution. (See the Deloitte AI Institute’s Generative AI Dossier to explore use-cases across six industries, including life sciences and health care).

Over the past several years, many hospitals and health systems have embarked on a digital transformation. Health systems appear to be at the midway point in their adoption of various technologies (e.g., cloud, data analytics, AI, Internet of Things, natural language processing), according to a survey of health system executives conducted by the Deloitte Center for Health Solutions (see our report on Digital health integration). We are urging health care organizations to incorporate generative AI in parallel with other digital technologies as they retire legacy hardware and transition to a digital environment.

The first step in this journey is to establish a center of excellence with appropriate governance structures and trustworthy frameworks for the digital program. These frameworks should address guidelines and procedures for generative AI use-cases, prioritization of those use-cases, guidelines for technology infrastructure, and mandates for transparency that will inform future decisions. It should also ensure that the outputs are fair and impartial, robust and reliable, transparent and explainable, safe and secure, accountable and responsible, and respectful of privacy.

The life sciences sector appears to be a step or two ahead of health care in terms of generative AI adoption (see Can life sciences companies unlock the full value of GenAI?). About 25% of health care organizations have a generative AI governance model and a dedicated team to maintain and update models, according to preliminary results from our 2024 Health Care Outlook survey. About half of the health care executives we surveyed said they intend to establish a governance structure within the next 12 months. Without the proper guardrails—and a solid modernized data foundation—it could be difficult to scale generative AI use-cases. More than half of our survey respondents said they have begun to experiment with generative AI use-cases. Claims/payment, patient support and services, and prevention appear to be the top priority areas over the next year, according to respondents.

Consider this: In June, Mayo Clinic said it was using generative AI to automate some form-filling tasks to free up time for staff.² New York University Langone Health is using the technology to identify which patients are most likely to be readmitted within a month of discharge. It is also being used to predict in-hospital mortality, length of stay, and medical claims denials.³

Generative AI in the back office

When a medical insurance claim is denied, hospital billing staff can face a costly and lengthy process of reviewing patient records and medical policies to create an appeal letter. While many denied claims are recoverable, vague reasons for denial—combined with sometimes limited hospital billing resources—can mean only a fraction of in-network claims are appealed.⁴ This means millions of dollars might be written off by hospitals and health systems as an uncollectable loss each year. The ability to automate aspects of claims authorization and appeals could lead to a significant savings in revenue and time. Moreover, a retrieval model could be developed to reach across large volumes of medical policies and member plans to identify the necessary information for more accurate appeals. Using extractive algorithms, the organization could instantly consult unstructured medical notes, medications, lab results, and other electronic health records.

How might generative AI impact clinicians, staff, and patients?

Some health care leaders might be hesitant to embrace generative AI. There is sometimes a fear of making a mistake with the technology that could put the organization (or worse, the patient) at risk. There also are concerns that protected health information (PHI) could be compromised. More than 60% of health system executives said more time is required to assess the potential benefits and risks of generative AI, according to results from our upcoming 2024 Health Care Outlook Survey. Here are some other highlights:

• Clinicians: The information needed to make medical decisions (e.g., medical history, laboratory and imaging results, unstructured clinical notes) can be scattered across multiple records that exist in myriad formats and locations. Generative AI could be used to compile and organize this information—and put it into a format that is accessible and clinician-friendly—to accelerate and augment critical thinking. In addition, generative AI-enabled ambient documentation could pull information from clinician conversations and generate natural-sounding notes. The technology could also be trained to identify patterns that are too subtle for a human to recognize.
  
• Frontline workers: Accurate real-time audio and text messages could be generated instantly, and in different languages, as frontline workers interact with people for health care, social, and emergency services. Generative AI also could translate documents, websites, laws, regulations, and policies. Health advisories, for example, could make essential information accessible to a diverse population. Generative AI could also play a central role in optimizing and mitigating health and safety risks by generating worksite-specific safety training that replicates real-world settings and critical scenarios.
  
• Patients and customers: Operational inefficiencies or limited capacity in the call center can translate to decreased customer satisfaction. Generative AI could help to create hyper-personalized experiences with customers and patients. It could also help efficiently support customers while also reducing call volume handled by associates. The technology might also assist human staff in generating responses to customer questions about the claims process, insurance coverage, and other plan details. The customer service experience can have a direct impact on patient perception, even without any change in charged costs or appointment wait times. This is particularly relevant in the context of legacy payer call centers, where patients might spend significant time navigating Interactive Voice Response (IVR)-based responses.
  

Conclusion

Integrating generative AI likely won’t be the most challenging part of a digital transformation. A modern cloud-based data architecture will likely be necessary to build and maintain generative AI models at scale. Scaling will likely also require hospitals and health systems to continually maintain their models and guard against drift, introduction of bias, or hallucination. The use of standardized authorization rules and patient-specific medical history, alongside continuous monitoring and careful evaluation, should help to mitigate this risk and promote fairer and more equitable outcomes.

As recently as 2019, at least 70% of US hospitals were still relying on fax machines and standard phone lines to transfer and retrieve patient records or order prescriptions, according to a 2021 report from the Office of the National Coordinator for Health Information Technology.⁵ Generative AI could help clinicians and staff move beyond fax machines and other legacy hardware. Reaching the Future of Health will likely require hospitals and health systems to undergo a digital transformation that includes generative AI. This might not only help improve efficiencies and the bottom line, but it could also lead to better outcomes and improved patient and clinician experiences.

Latest news from @DeloitteHealth

Endnotes:

¹The fax is still king in health care — and it’s not going away anytime soon, Computer World, May 22, 2023

²Mayo Clinic to transform health care with generative AI, Google Cloud, June 7, 2023

³NYU Langone Health promotes generative AI innovation, Healthcare IT News, September 27, 2023

⁴How health insurers have made appealing denials so complicated, ProPublica, August 31, 2023

⁵Hospital use of certified HIT, The Office of the National Coordinator for Health Information Technology

This publication contains general information only and Deloitte is not, by means of this publication, rendering accounting, business, financial, investment, legal, tax, or other professional advice or services. This publication is not a substitute for such professional advice or services, nor should it be used as a basis for any decision or action that may affect your business. Before making any decision or taking any action that may affect your business, you should consult a qualified professional advisor.

Deloitte shall not be responsible for any loss sustained by any person who relies on this publication.

Return to the Health Forward home page to discover more insights from our leaders.