Harnessing AI to reshape consumer experiences in healthcare

In the rapidly evolving healthcare landscape, AI has the potential to reshape how consumers engage with medical services. Today, consumers in the United States can struggle with everything from finding the right insurance coverage to understanding when they should see a doctor, what it will cost, and how to manage their health. Many spend hours researching, conferring with friends and family, and calling providers and payers to find answers to their questions. In fact, a quarter of consumers we surveyed say they are not able to get the care they need when they need it.¹McKinsey Consumer Health Insights Survey, 2024 (n = 2,133).

Meaningful improvements to address the complex healthcare ecosystem will need more than increased human engagement, particularly given labor shortages and increasing costs of care. Enter AI: the technology has the potential to reimagine the consumer experience (CX) and enhance engagement in ways that weren’t possible even a few years ago. It can help enable personalized care, boost transparency and simplicity, and ensure that consumers can take control of their health and healthcare-related decisions. A recent study in which healthcare professionals evaluated physician responses to questions from patients on social media forums in comparison to chatbot-generated replies found that evaluators preferred the AI responses, rating them higher quality and more empathetic.²John W. Ayers, Adam Poliak, Mark Dredze et al. “Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum,” JAMA Internal Medicine, April 28, 2023, Volume 183, Number 6.

Some good news: the healthcare sector understands the AI opportunity in CX. Sixty-two percent of respondents to McKinsey’s survey of healthcare leaders indicated consumer engagement and experience is an area where generative AI (gen AI) has the greatest potential. Yet only 29 percent of respondents have begun implementing gen AI in their organization for any purpose.

Organizations that use AI to tailor healthcare experiences to individual needs, preferences, and goals, while mitigating potential risks, have the potential to benefit from more trusting relationships with consumers. Adopting the latest technology, including AI, can also improve business outcomes: it is estimated that net savings could be 5 to 10 percent of healthcare spending, with exact percentages varying slightly across private and public payers, physician groups, and hospitals, according to a study coauthored by McKinsey and published by the National Bureau of Economic Research.³It is estimated that net savings could be 7 to 10 percent for private payers, 5 to 7 percent for public payers, 3 to 8 percent for physician groups, and 5 to 11 percent for hospitals; for more, see Nikhil Sahni, George Stein, Rodney Zemmel, and David M. Cutler, “The potential impact of artificial intelligence on healthcare spending,” National Bureau of Economic Research, October 2023.

Why AI now for healthcare consumers

Successful AI use requires data, and the industry has ample data for AI tools to dig into—roughly 30 percent of data globally is produced by the healthcare industry.⁴Linnie Green, “How healthcare data technology is leveraged by leaders,” Arcadia, September 6, 2023. The compound annual growth rate for healthcare’s data is expected to hit 36 percent by 2025.⁵David Reinsel, John Gantz, and John Rydning, The digitization of the world: From edge to core, IDC, November 2018. Healthcare organizations also have the distinction of more consumers willing to share their personal health-related information to support their health compared with employers, the government, or tech companies.⁶McKinsey Consumer Health Insights Survey, 2024 (n = 1,219); 74 percent definitely/probably would grant access to their personal health-related information to their primary care providers, 62 percent to their health system, and 61 percent to their health insurance company versus 32 percent to their employer, 22 percent to the federal government, 22 percent to state government, 18 percent to Apple, 18 percent to Google, 17 percent to Amazon, 17 percent to Best Buy, and 16 percent to their grocer.

Healthcare has, however, typically lagged behind other sectors in adopting digital technology.⁷McKinsey Digital Quotient assessment taken by more than 550 companies and 20,000 individual participants globally, 2019–20. In the past—apart from patient privacy concerns, which are paramount—the main technology challenge with utilizing AI has been the inability to extract valuable insights from unstructured data and curate experiences for what are highly variable consumer journeys. Just like in other industries, data is scattered across multiple systems, and while healthcare has used some of that structured data for AI, it has had limited success with unstructured sources such as call transcripts. Now, gen AI can help unleash the power of previously unusable data sources containing critical consumer information and make them compatible for broader AI use to learn behavioral patterns, providing a newfound ability to offer customization at a scale previously unattainable.

With the technology advancing at lightning speed, it has the ability to help optimize the healthcare consumer journey.

How AI can streamline the entire healthcare journey

AI has the potential to revamp the entire healthcare journey for consumers. Consumers experience care across several core journeys, and while these steps need not be chronological, discrete, or the same for each individual, each stage offers opportunities for AI to improve the experience.

Journey example 1. Enhancing engagement in proactive care and wellness

People often struggle to make healthy lifestyle choices that could minimize the risk of preventable health conditions. Providers and payers alike can adopt AI to facilitate a personalized preventative care experience. For example, machine learning (ML) models can predict clinical and behavioral risks for individuals, and gen AI can tailor wellness programs and personalize messages with the aim of increasing the likelihood of engagement. In particular, integrated delivery networks that have a well-rounded view of an individual can identify care gaps and use AI-powered models to suggest next steps. Beyond payers and providers, direct-to-consumer wellness companies are increasingly serving as health guides; for instance, the AI-powered Whoop Coach uses biometric data from a user’s Whoop wearable device and then uses ML on consumer-shared journey entries to generate individualized responses to health questions and to coach individuals to attain peak performance.¹“Optimizing athletic performance and recovery using machine learning at WHOOP,” Harvard Business School Digital Initiative, November 2018.

Journey example 2. Making care easier to find and schedule

When a consumer identifies a care need, finding care that meets their requirements and preferences can be frustrating and time-consuming. AI can serve as a 24/7 front door to conducting information intake to create initial suggestions on next steps. In doing so, it has the potential to speed up the process of pointing consumers to the right care, at the right place and time. For example, AI can couple patients’ preferences with provider-specific data to create a more streamlined search experience. Furthermore, gen AI can synthesize information to prepare consumers for visits and increase engagement by hyperpersonalizing appointment reminders.

AI may also help minimize appointment no-shows, which cost the US healthcare system upward of $150 billion annually.¹Sachin H. Jain, “Missed appointments, missed opportunities: Tackling the patient no-show problem,” Forbes, December 10, 2021. Delayed care can also lead to a patient’s condition worsening.²Jessica Buchter, Jenny Cordina, and Mark Lee, “Driving growth through consumer centricity in healthcare,” McKinsey, March 14, 2023. By analyzing past behavior, AI can help providers and payers proactively identify patients more likely to miss appointments and send reminders ahead of time, and, when appointments are missed, offer options to reschedule. For example, Total Health Care, a federally qualified health center in Baltimore, reported using eClinicalWorks’ Healow AI model to identify high-risk no-show patients, claiming that it reduced missed appointments by 34 percent.³“50-provider FQHC significantly reduces patient no-show rate by implanting healow no-show prediction AI model,” Business Wire, September 19, 2023.

Journey example 3. Making coverage and cost information more understandable

Before and after visits, patients often struggle to understand their insurance coverage and how much they owe. Healthcare organizations may be able to use AI to provide an easy-to-understand summary of expected costs and coverage modulated to each individual’s communication preferences (including language, reading level, and so on). For instance, health systems and payers can improve so-called cost estimator tools to account for the specific service the patient is searching for. And after services are provided, organizations can use gen AI tools to assist patients in understanding and resolving healthcare bills.¹Brian T. Horowitz, “Generative AI in healthcare: Google Cloud’s Amy Waldron on the tech giant’s health ambitions,” Healthcare Dive, August 17, 2023.

Journey example 4. Improving quality of care

Quality of care is one of the most important influencers of satisfaction; those who perceive that they received poor care are about three times as likely to switch providers.¹McKinsey Pre and Post Acute Experience Survey, 2024 (n = 5,026). One source of patient dissatisfaction is too-short doctor visits, of around 15 minutes, which patients view as insufficient to receive high-quality care. While there are many ways to work on this problem, AI offers intriguing solutions. For example, one way to improve this experience and to prioritize quality time is to use AI-powered ambient listening technology. These tools are designed to allow doctors to focus wholly on their patients concerns and questions, while the technology takes notes that a doctor can review for accuracy after a visit. Another example could be for AI to assist with in-patient care; this might look like a natural-language-processing model that identifies specific words uttered by bedridden patients and then notifies the nurses’ station to address their specific needs.

Journey example 5. Increasing adherence with follow-up care

It’s common for patients to leave healthcare visits unclear or confused about next steps. Gen AI can marry visit notes and discharge summaries with insurance coverage information and patient preferences to provide easy-to-understand takeaways from the visit, resources to manage stress, and tailored information on what’s needed next in a manner that’s free of medical jargon and shows empathy and understanding. This would not only make it easier for patients to navigate the healthcare system but also free up clinician time. For example, Kaiser Permanente reported that its AI-powered patient messaging system saw 32 percent of messages triaged without ever requiring physician input, ensuring that patients received the information they needed in a timely manner.¹Vincent X. Liu, Pamela Kaercher, Jennifer Manickam, “Content of patient electronic messages to physicians in a large integrated system,” JAMA Network Open, April 4, 2024, Volume 7, Number 4.

How to accelerate AI’s use to enhance consumer satisfaction

Despite AI’s potential to improve the end-to-end consumer healthcare journey, momentum has been slow. Organizations are caught between the excitement to quickly seize the opportunity and a lack of alignment on where to start, alongside general caution given the potential risks related to deploying AI. To accelerate the use of AI to upgrade the consumer experience, we propose five critical steps.

Tackle the 70 percent problem of data readiness

First, executives need to consider their organization’s data and tech readiness before allocating resources and funding. Delivering tangible value for healthcare consumers through AI requires integrated data ready for consumption, a challenging task that represents, on average, 70 percent of the work when developing AI-based solutions.⁸Eric Lamarre, Kate Smaje, and Rodney Zemmel, “Rewired to outcompete,” McKinsey Quarterly, June 20, 2023. For healthcare, the challenge is knowing what data to collect and how to connect those sources; data is fragmented across multiple platforms and in varying formats and levels of utility (for example, claims or electronic health records hosted on-premise, marketing information on a cloud platform, and call center information spread across multiple systems).

And while healthcare organizations may have a leg up on data volume compared with other sectors, they nonetheless face gaps that prevent a holistic view of consumers. For example, breaks in care continuity make it hard to fully understand a patient’s needs, habits, and preferences. AI outputs could also be biased unless they’re built on demographically diverse data. And to surface meaningful insights, organizations can complement their clinical and patient data with information on social determinants of health, patient-reported outcomes, retail purchases, and wellness trackers.

Zero in on consumer experience priorities to ensure AI success

In parallel with assessing data readiness, leaders can assess and prioritize areas for AI investment based on importance to their overall priorities to improve CX, opportunities, strategies, and feasibility. For example, AI could optimize administrative processes to reduce consumer touchpoints, thus lowering the cost to serve. For providers, this may mean fewer cancellations as a result of a better overall experience, while for payers it may lead to fewer follow-up calls to answer questions about benefits or coverage.

This is a critical step to avoid trying to do too much at once, which can limit meaningful progress. To prioritize areas of focus, it is imperative to engage cross-functional leaders in the organization. For example, clinical leadership in particular has first-hand insight on patients’ pain points and what exactly isn’t working in care delivery and CX.

Optimize real-time insights for AI-powered interventions

Once the data foundation is established and priorities are set, organizations can start to figure out what else is needed to properly contextualize the gathered data. Delivering truly personalized AI-powered insights involves stringing together multiple touchpoints across data sources into a personalized consumer journey. By combining details about, for example, doctor visits (frequency, types of doctors seen, or where appointments are made), patient outreach efforts, and a patient’s interactions and experiences, AI models can develop a closer representation of consumer behavior, which is critical for building predictive analytics to inform future interventions.

By analyzing details such as a patient’s appointment preferences and how or when they have responded to outreach, AI can tailor the timing, frequency, and message themes to provide recommendations most likely to resonate. Gen AI can further enhance the effectiveness of these timed interventions with hyperpersonalized message content.

Map AI risks in healthcare and develop mitigation plans

Compared with other industries, healthcare leaders face unique challenges given consent requirements, privacy risks, potential health implications, and regulatory oversight. While consent mechanisms in place during member enrollment or appointment scheduling give organizations the go-ahead to use some consumer data, consumers have no easy way to review or adjust these consents. They should be able to learn not only about data usage when signing new consent forms but also about changes to privacy policies on previously provided consent, with clear opt-out instructions.

Besides data-use transparency, organizations can establish governance processes that are also anchored in AI-use and algorithm transparency. They can provide consumers with clear logs and documentation on AI systems, including bias mitigation strategies and training protocols such as details on the population profiles used. As consumer expectations shift, as in other industries, toward easier access and control of their data—for example, some companies enable customers to decide which of their purchases are used to train ML recommendation models—healthcare organizations will face increased pressure to do the same.

Adding to an already complex patient-privacy environment is rapidly evolving AI-specific regulations. The Department of Health and Human Services’ AI Task Force has been developing policies to protect patients as part of the White House’s Executive Order on AI Safety.⁹“Fact sheet: Biden-Harris administration announces key AI actions following President Biden’s landmark executive order,” The White House, January 29, 2024. Finally, more mature, integrated data repositories built to power AI can become valuable targets for cyberattacks. 2023 broke the record for healthcare data breaches, logging some 725 breaches of 500 or more records, more than twice what was reported in 2017. ¹⁰Steve Alder, “December 2023 healthcare data breach report,” The HIPPA Journal, January 18, 2024.

Level up your team’s AI capabilities

In the long term, provider organizations and payers will have to invest in their capabilities and talent to fully capture the AI opportunity. They must carefully balance upskilling existing talent and hiring for AI-specific skills and then organize tactical teams to act on chosen initiatives. Partnering with third-party AI vendors is also an option and might allow an organization to move quickly.

One way to increase the likelihood of success in AI implementation is to employ a copilot model, where employees work alongside AI tools to make incremental process improvements. This capitalizes on AI’s speed and capacity with the checks and balances of human skill and intuition to mitigate errors and risks. Importantly, this process includes periods of capability testing and learnings collection within a small set of users prior to scaling across the enterprise. Such a test-and-learn tactic allows organizations to de-risk scaling and to measure impact and adoption within existing workflows.

Today, interacting with the healthcare ecosystem is often clunky and lacks the personalization consumers expect. AI has the potential to reshape the healthcare journey by enabling consumer-centricity. Building successful AI solutions that scale requires an iterative approach, a defined controlled launch strategy with a clear plan of how to integrate with existing and reimagined workflows, and key performance metrics to amplify what’s working well. Executive commitment is also key to capturing the power of the flywheel effect. Although enabling this revolution requires targeted investments, data advancements, and risk mitigation, we expect the hard work to pay off. Healthcare AI implementation can benefit organizations’ bottom lines, as well as operational and administrative functions, while consumers can enjoy greater ownership of their health and wellness journeys and better overall health.