2605_2018_AI_in_Healthcare_Research_Blog_Post4_blog_image-1This blog post is the final entry in a series on ZS’s 2018 AI in Healthcare study.

The U.S. healthcare system is expensive and wasteful, and is one of the least efficient in the world. More than half of doctors are burned out, and medical errors are the third leading cause of death. And if the healthcare experience doesn’t kill patients, it most definitely frustrates them: Patients in U.S. metropolitan areas reportedly wait an average of 24 days for a doctor’s appointment—up 30% since 2014.

While the healthcare system is failing, health tech innovation is thriving. Disruptors are working hard to increase healthcare efficiency and reinvent healthcare delivery, better empowering patients and enabling doctors. Many of these tech solutions are driven by artificial intelligence, and AI certainly has been heralded for its transformative powers.

Recently, we conducted the 2018 AI in Healthcare study, surveying 400 doctors and 400 patients (and several payers and IDNs) to gather their thoughts on AI’s use in healthcare delivery. Based on our research, we see AI-enabled tech potentially making inroads to help fix our broken healthcare system. The question is whether AI will enable a return to the human-focused care delivery of the past (a traditional, Norman Rockwell-esque healthcare model) or accelerate a move toward one that’s more impersonal and efficiency-minded (an industrial model of medicine).

Healthcare’s AI-Enabled Evolution

The modern healthcare model—one that has been around since the early 1800s and is built on the doctor’s knowledge, experience and gut feeling—can certainly be improved upon. When, for example, did doctors start spending more time documenting care than providing it? The hope is that when AI is used to improve the efficiency and effectiveness of operations and administrative tasks, there could be more emphasis on the doctor-patient relationship. The concern is that when AI transcends administrative roles and lands in the realm of diagnosis and treatment decisions, the healthcare experience could become robotic.

That’s a legitimate concern, and one that appears to be quite common among HCPs. As one primary care physician employed by a regional health system in New England told us: “One of the potential negatives of some of these technological innovations is that it makes the delivery of healthcare more impersonal. You lose the human touch, the intuition that a practitioner can get from being in front of a patient, looking at her, assessing her tone of voice, how she presents herself—you know, her emotional state.”

Some health systems are trying to retain elements of traditional medicine despite the emergence of tech solutions. For example, One Medical and Iora Health have differentiated themselves with tech-based, patient-focused primary care models. On the other hand, market pressures continue to drive mergers and acquisitions, and the overall corporatization of healthcare, which is precipitating the move to industrial medicine, and AI is accelerating the transition. As this trend continues, we anticipate the following aspects of industrial medicine to infiltrate the current healthcare system:

  • The standardization of care and centralized decision-making: Provider organizations are adjusting workflows and redesigning hospital infrastructures to increase efficiency and effectiveness across all sites of care. Their goal is to systematize healthcare delivery so that its processes function more like an assembly line. For example, Cleveland Clinic is designing its departments around the treatment workflow for a given condition. Administrators and HCPs at Methodist Le Bonheur Healthcare follow uniform treatment processes for patients with stroke and sepsis, and cut costs in the process.

    And as providers pursue standardized care, AI is well positioned to help by analyzing data and processes to refine workflows, manage risks and gain visibility into system-wide cost drivers. Beyond the changes that AI will bring on the individual provider level, the tech likely will be used to establish standards across the industry. But what will the patient experience feel like as AI facilitates an end-to-end assembly line for heart surgery, for example—complete with a robot calling the shots in the OR? And is the quality of care at stake if AI-enabled standardization overlooks patient nuances?
  • The down-skilling of care delivery: Physician compensation is one the fastest-growing expenses in healthcare, and providers are taking action: They’re trimming budgets by employing AI-enabled, protocol-guided nurse practitioners and physician assistants as primary points of care. In addition, providers are using health coaches and health educators as front-line replacements for physicians. The goal is to emphasize prevention and wellness, and to reduce the overall costs of healthcare. By accessing decades of healthcare experience and decision-making on demand, AI is creating a new layer of patient engagement that requires less training and money, and reaches further into the wellness and prevention ideal than traditional physicians can accomplish alone.
  • Doctors as employees: There are differing opinions within the medical community about the benefits and risks of being employed by a health system. As the administrator of a New York-based academic hospital put it: “The business part of being a physician is the part they’re happy to give up. They just want to be doctors. They’re more than happy with being an employee. They don’t have to worry about the ups and downs of the industry. [However], they don’t want to give up control over the standard of care.” Nonetheless, we’ve seen an uptick in the number of physicians who are employed by medical centers. The low overhead costs and few administrative burdens associated with employment rather than ownership are an attractive prospect, particularly for millennial doctors. As doctors become part of larger corporations with increasing price and profitability pressure, will AI provide the opportunity to bring back bedside medicine?
  • Shifting sites of care: Tasked with curbing the rising cost of healthcare services and extending the primary care continuum, retail health clinics and other sites of care outside hospitals’ four walls are gaining prominence in the U.S. ZS analysis of American Hospital Association data across more than 2,000 hospitals shows that from 2011 to 2016, providers have increased their wellness, pre-acute and post-acute care services by as much as 42%. At the same time, they’ve decreased their inpatient and acute care efforts, which are viewed as cost centers. Telehealth also is on the rise, bringing with it the promise of improved care for rural populations and less strain on the overall health budget. As providers look to increase patient engagement in different manners and at different sites, AI is being used to optimize the reach into wellness and chronic care, where it can have a larger impact on population health.

How Likely Is a Full Transition to Industrial Medicine?

Industry stakeholders, regulators and investors seem to be clearing the way for AI’s entry. In response, provider organizations are using AI to further standardize the care that they provide, which is accelerating the transition to industrial medicine. AI is moving beyond administrative or operational contributions (where it sits now) and is beginning to assist with clinical tasks and even augment human cognition. But what if AI could enable doctors to spend more time with their patients to reinvigorate the traditional relationship-based healthcare environment? What will happen to the doctor-patient relationship as AI enters the exam room? How can we keep tabs on healthcare evolution as the AI wave continues to roll in?

There are five major signposts to monitor AI’s impact on healthcare delivery:

1. Budgets feel the squeeze, but progress won’t slow. The financial pressures looming over every healthcare provider in today’s healthcare landscape continue to drive the conversion to industrial medicine, and AI will grease the skids. The prospect of leveraging the technology to automate tedious tasks and free up humans to explore more fruitful endeavors is too attractive for providers to ignore, despite the budgeting sacrifices that may need to be made. Use of AI has already shown financial value. As its applications grow, we should watch how physicians and administrators speak about future uses of AI, and its value to shareholders and the healthcare industry.

2. Healthcare investments heat up across provider organizations and startups. AI startups in the healthcare sphere are reportedly attracting more investor dollars than any other industry, raising $4.3 billion across 576 deals since 2013. As the money keeps flowing into AI and investors continue to make big bets, it’ll be interesting to watch how the investments change in terms of overall volume, deal size, specific AI applications, etc. On the provider side, administrators appear to be confident that AI will deliver on its promise, and justify the budget sacrifices—but they still have some convincing to do. Some of the more confident provider organizations are banking on seeing a return on their investments within the next three years.

3. Regulatory pressures lighten. We’re beginning to see signs of the transition to industrial medicine as the FDA warms to the idea of AI in healthcare: Commissioner Scott Gottlieb recently wrote that “artificial intelligence (AI) … holds enormous promise for the future of medicine.” The federal agency is developing a data science incubator to encourage the use of AI in drug development, collaborating with cancer-research platform Project Data Sphere to develop algorithms to improve tumor classification, and launching “Pre-cert 1.0” to foster innovation and to ensure that AI-related solutions reach the market quickly. Adam Boehler, director of the Center for Medicare and Medicaid Innovation, also issued a forward-looking statement, saying that redundant quality measures in healthcare should be replaced with artificial intelligence in EHR data.

Things seem to be evolving on a global regulatory front as well, with more tech-savvy countries such as Japan, Russia, Singapore and China working closely with the International Medical Device Regulators Forum to harmonize their regulatory approaches for digital health medical devices. The FDA seems to like what it has seen and is now following suit.

4. AI is more directly linked to better health outcomes. This is a goal that was supposed to be pushed along by the transition from healthcare’s fee-for-service model to value-based care. Will AI jump in and do the dirty work? Early use of AI has delivered improved efficiency and patient outcomes, and reduced hospital stays, waste and medical errors at various provider organizations such as Boston Medical Center.

5. The next generation of caregivers (and administrators) become more tech-savvy and tech-accepting. As it turns out, AI is already making an appearance on med school syllabi: Students at New York Medical College are working as interns at IBM Watson Research to learn how AI is reshaping medical care. At the undergraduate level, MIT has announced a $1 billion initiative to make artificial intelligence part of its curriculum for all students. For those already in the field, 42% of healthcare organizations are redesigning jobs to emphasize distinctive human skills and human-machine augmentation.

We can expect to see many more examples of AI’s entry into healthcare, but questions remain about who owns the patient data that fuels healthcare AI, and who has access to it, as well as how to overcome interoperability issues. There’s also the challenge of tapping a variety of data sources for real-world evidence that can be used to improve patient outcomes.

There’s no question that AI is powering changes to healthcare delivery, so pharma companies should continue to monitor the signposts. They’d also be well served by developing an evidence-gathering strategy now, and beginning to explore partnerships and collaborations that maximize their access to data-generating companies. The road ahead might be hard to see, but you won’t get where you need to go by standing still.

Topics: Paul Darling, artificial intelligence, healthcare delivery, AI in Healthcare study, industrial medicine, data and interoperability, standardization of care, traditional medicine