Artificial intelligence is often touted as the next big thing in medicine, with the potential to make healthcare more efficient, more affordable and more personalized. Yet, so far, most uses of AI in healthcare support business as usual, incrementally improving the system and making insurers, hospitals and pharmaceutical companies more money without fundamentally changing the experience for patients.
We are missing out on AI’s revolutionary potential to reshape our healthcare system by directly serving patients and putting them back at the center of care.
Consider the information that comes out of every doctor’s visit. There has been a huge push in recent years to digitize health records. Companies like Nuance Communications, acquired by Microsoft earlier this year for nearly $20 billion, offer AI tools that help doctors create and manage electronic medical records and enable healthcare systems and insurers to make better use of the data.
Yet few patients directly benefit from these systems. As Seema Verma, the former administrator of the Centers for Medicare & Medicaid Services (CMS), said in an interview, “We’ve spent $36 billion [taxpayer] dollars to create electronic medical records, and those records today are locked in your doctor’s computer.”
What if we leveraged the same machine learning technology Nuance and others employ but instead used it to help patients access and understand their health data? For example, AI could provide an annotated transcript of every doctor’s visit and extract details about treatment plans, medication regimens and future appointments. It could even track symptoms and provide nudges and incentives to help patients stick to their treatments.
AI tools like this can empower patients to take greater control over their health. A study where patients were granted access to their doctors’ notes found that 77% felt more in control of their care and over 60% were better at following their medication regimens. More broadly, research shows that patients want greater support in understanding their health, weighing treatment options and navigating important medical decisions. In an era when providers have limited time for patients, AI can fill an important gap by serving as a sidekick that helps patients navigate the system and manage their care.
A handful of companies are starting to use AI to make healthcare more accessible to patients, but they are still in the minority. Curai Health and Babylon incorporate AI into platforms that allow users to easily and affordably access primary care and connect with providers from home. Ada offers an app driven by AI and clinical research that helps patients monitor their health and track symptoms. Abridge (a company one of us co-founded) uses AI to help patients record, understand and follow through on conversations with their doctors. Woebot Health offers a therapist chatbot that can provide emotional support and monitor mental well-being. Research found that patients can establish and maintain a bond with the Woebot that is comparable to a human therapist.
These companies offer a compelling glimpse into what AI can do. Healthcare innovators should be looking for more opportunities to use technology to put patients in the driver’s seat. Research shows that patients who are engaged and empowered are more satisfied with their care and more attentive to and involved in managing their health, leading to better health outcomes.
Patient engagement also makes physicians’ jobs easier and can potentially reduce healthcare costs through improved preventive care, better adherence to treatment plans and decreased use of unnecessary, expensive procedures.
Now is the time to start building patient-first AI systems. The pandemic sparked a newfound openness to telehealth and more opportunities for patients to manage their health from home. Healthcare apps and wearable devices like the Fitbit and Apple Watch are increasingly popular and are producing a flood of valuable data. As patients have more channels for managing their health and gain access to more of their healthcare data, they can benefit from AI-powered tools to make sense of it all.
To get there, health technology innovators need to change how they think about what AI can do in medicine. They should engage in user-centered design driven by patient needs and explore new use cases to deploy AI in service of patients (while also recognizing the sensitive nature of health data). They should involve patients in defining problems and developing tools to solve them.
None of this is possible unless patients’ personal health data is accessible to them. There is a growing movement by federal agencies like CMS and patient groups like OpenNotes to secure patients’ rights to their medical data. But we need to incentivize or require providers, hospitals and EMR systems to share the data in a clean format that can be transferred and used across platforms (with the patient’s permission, of course). Giving people access to their data helps ensure that the money we have poured into digitizing health records serves patients, not just systems.
We also need to develop mechanisms for ensuring AI tools are safe and effective and can do what their creators claim they can. Earlier this year, the FDA released its first plan for how it might regulate AI-based products. Existing frameworks for approving medical device hardware may not work well for AI software, but we will need some way to ensure that AI systems are safe without throttling innovation. We also need to educate patients so they can decide which AI systems to trust, understand how to use these products and be clear on what AI tools can and can’t do.
Instead of using technology to further business as usual, AI offers us an opportunity to give people more control over their health. The ultimate goal of our medical system is to efficiently improve patients’ health, so let’s put them back at the center as we develop AI for healthcare.