Navigating healthcare’s next frontier: AI

July 28, 2023

A recent Yale School of Management survey showed close to half of 200 top CEOs across sectors believe health is the field in which AI will make the most transformative contribution. 

  • Healthcare spending on AI software is expected to grow 40% to nearly $6.2 billion in 2023, compared with $4.4 billion in 2022
  • Rules-based predictive AI has been around for a while, powering myriads of applications, whether identifying hospital readmission risks for patients or predicting clinical outcomes in drug trials. 

Why it matters: Now, with breakthroughs in generative AI, disparate sources of unstructured data like clinical notes, diagnostic images and medical charts have turned into assets for the data-hungry technology. 

The bottom line: The untapped potential is huge for the field of healthcare, which currently generates 30% of the world’s data volume. 

What’s next: In this issue, we explore three major pain points in healthcare that are ripe for an AI solution, specifically for providers, hospitals and health insurance companies. 

  • Our next edition will focus on the pharmaceutical, biotechnology and medical devices industries.

1. Admin tasks detract from high-value work

The big picture: Clinician burnout and healthcare staff attrition are worldwide concerns, leading to lost productivity, increased costs and a decline in patient care quality.

Using AI to tackle paperwork isn’t glamorous, but can reduce a huge burden for providers. For years, doctors have used simpler admin tools like speech recognition software to help with documentation. AI can do far more – such as summarizing, organizing and tagging conversations between physicians and patients. 

Why it matters: Physicians are spending an estimated 4.5 hours a day completing electronic health records required for treatment and billing. 

  • In dollar terms, admin expenses accounted for up to $1 trillion in the U.S., or 25% of the total national health expenditures in 2019. 
  • Automating paperwork means providers can spend time connecting with patients.

What else? When the quality of patient care declines, hospital readmissions tend to rise, resulting in an even heavier burden on the healthcare system. 

  • Clinician burnout is not unique to the U.S., and it was a problem even before the pandemic. In a 2019 study, Germany estimated an average of €9 billion in lost productivity annually, while Switzerland averaged an annual loss of €5.8 billion to replace doctors who left the field due to stress and exhaustion.

Go deeper:

However, a fragmented landscape with multiple AI programs could perpetuate the industry-wide issue of interoperability.

2. Prior authorization inefficiencies lead to delays or coverage denials

The big picture: Prior authorization is the process through which a doctor files a request with a patient’s health insurer before treatments, tests or prescriptions. It has remained a largely manual process on both sides, with humans sorting through a patient’s health plan and medical history via emails, phone calls and faxes. 

Because prior authorization is based on data exchange, AI will be able to automate up to 75% of the manual work and reduce the approval window to days, if not hours, per McKinsey. That would present a massive improvement over the current average of 10 days. 

Why it matters: Data show that 93% of physicians said prior authorizations delay patient care, and 82% said the process is so complicated that it causes patients to abandon treatment altogether.

  • Automation is difficult as payers have no standardized method for receiving and approving requests. Prior authorization has the lowest electronic adoption rate (about 26%) among all admin tasks for payers. 
  • The manually-intensive nature exposes the process to errors. A 2022 report found that 13% of prior authorization denials by Medicare Advantage plans were for benefits that should have been covered.

The Centers for Medicare & Medicaid Services (CMS) recently proposed a rule that would require certain payers to implement an automated process, meet shorter time frames and be more transparent about their decision-making.

Go deeper: 

  • Florida Blue partnered with Olive, a healthcare automation company, to issue approvals while a patient is still at the doctor’s office. Rather than deny requests that it cannot immediately approve, the tool instead routes them to clinicians for human review. Health Care Service Corp. has implemented a similar automated tool.
  • Startups like Cohere offer AI solutions that can be customized for individual health plans’ specific prior authorization needs. 

In June, however, the American Medical Association called for more oversight of how AI is used for prior authorizations to ensure it does not result in more coverage denials.

3. Physicians struggle to stay on top of latest medical knowledge

The big picture: A 2022 survey showed that 95% of physicians are interested in learning about new trials, treatments or procedures, but 68% said they feel overwhelmed by the amount of information they have to keep up with. 

Large language models, or LLMs, can help healthcare professionals stay up-to-date by quickly summarizing and analyzing new research findings, and suggesting relevant studies based on the provider’s specialty and patient population. 

Why it matters: “Medical knowledge is growing so rapidly that only 6% of what the average new physician is taught at medical school today will be relevant in 10 years,” according to a National Bureau of Economic Research report. “Technology such as AI could provide valuable clinical data to the clinician at the time of diagnosis.” 

That said, “adoption of AI for decision-making in medicine is outpacing efforts to oversee its use,” per STAT News. A research collective called Health AI Partnership – which includes leaders from New York Presbyterian, Mayo Clinic, and other major institutions – published a guide to help health systems overcome challenges, address biases and prioritize equity while implementing AI tools. 

Go deeper:

  • New York-based Northwell Health integrated Aidoc’s AI system into 17 of its hospitals, while also launching an AI-enhanced pregnancy chat app earlier this year to screen for common symptoms and give users personalized advice. New York City Health + Hospitals and NYU Langone Health are also proactively using AI for patient care.
  • Mount Sinai is incorporating AI for more timely diagnosis of eye disease and risk assessment of systemic health conditions. The hospital also has a chatbot that guides anxious patients who are trying to decide between making a regular doctor’s appointment, visiting a local urgent care or heading to an emergency room.

Yes, but: Generative AI’s tendency to hallucinate is a major red flag in the high-stakes realm of patient care. LLMs “should never replace humans in the diagnosis and treatment of patients,” said Dr. Karen DeSalvo, chief health officer at Google and a former Obama administration health official. 

4. Upcoming health events + conferences to monitor

  1. Healthcare Automation and Digitalization Congress
    September 25-26, 2023 // Zurich, Switzerland
  2. HLTH 2023
    October 8-11, 2023 // Las Vegas
  3. Reuters Total Health
    November 7-8, 2023 // Chicago
  4. FT Global Pharma and Biotech Summit
    November 7-9, 2023 // Digital & In-Person
  5. FT Health Technology Summit
    November 30, 2023 // Digital
  6. 2023 Forbes Healthcare Summit
    December 4-5, 2023 // New York City

Share your feedback with us at