The phrase “AI in healthcare” conjures countless different images and feelings. Some may picture an advanced algorithm tirelessly and rapidly combing through data to discover novel drugs or inspire creative treatments. Others may imagine a futuristic dystopia with an apathetic robot taking vitals, or perhaps using the equivalent of a Google search to diagnose a patient. In reality, AI has been utilized in healthcare since before the evolution of ChatGPT and the subsequent AI craze in early 2023. Its applications are endless and sensational. As described in a review by Kanadpriya Basu and colleagues, it can automate administrative and secretarial tasks like documentation, keep a record of hospital protocol, and handle prescriptions and reports. Streamlining these time consuming tasks gives more freedom to clinicians to focus on patients. AI aids in physician-patient communication and expands the possibilities for virtual appointments, making it possible to receive care in remote areas. AI platforms are also crucial for research and continuing education. Some of the more exciting applications of AI include improving diagnoses, drug discovery and development, imaging interpretations, and contact tracing for diseases like COVID-19.

Powerful, accurate AI-driven tools are changing the game for neuroscientists, cardiologists, and biologists, just to name a few. The United Kingdom’s National Health Service (NHS) has widely distributed an AI CT software that can triple the rate of stroke recovery. The device can analyze a patient’s stroke and quickly determine the severity, whether there is a need for emergency surgery, and recommend an appropriate treatment. Since strokes cause significant damage at a blistering pace, this tool revolutionizes the speed of intervention and eventual recovery. Another recent breakthrough was the creation of an AI stethoscope that can detect major heart conditions in a staggering fifteen seconds. It works by recording ECG data while simultaneously analyzing differences in heartbeat and blood flow that would be undetectable with an ordinary stethoscope. Early diagnosis is critical to give patients with heart conditions the best care possible, and the timing of intervention is often key to saving lives. A third application of generative AI is a new antibiotic that may be able to combat MRSA. MRSA is a potentially deadly bacterial infection that is resistant to most common antibiotics. Researchers used an AI algorithm to design and test over thirty possible compounds. This approach allowed them to discover antibiotic structures that haven’t yet been used in medicine, unlocking a multitude of possibilities for future research and development.

Patient-facing technologies are being developed as well, in the form of wearable technology and online platforms. For example, Empatica created Embrace2, a seizure monitoring device in smartwatch form. It’s the world’s first epilepsy-related wrist wearable approved by the FDA. Embrace2 tracks physiological signals, detects possible seizures, and alerts loved ones when they happen. It also includes an app where wearers can track their seizures, medications, sleep, and activity to have a more complete view of their health. Another example is Outcomes4Me, an AI-powered online platform for cancer patients where they can track symptoms, get treatment guidance, access education resources, and be matched with clinical trials. Patient-facing technologies like these are crucial for the expansion of AI in health, making quality care accessible to more people. A study led by Md Faiazul Haque Lamem found that AI can address workforce and medical resource shortages. This is most plainly seen by the way online systems can expand access to diagnosis and early detection tools, an aspect of care that is often difficult for patients in remote areas to get. Additionally, many technologies are more cost-effective than traditional modes of care. Remote health monitoring systems, for example, reduce the need for healthcare facilities, which in turn saves money that would ordinarily be spent on staff and facility upkeep. Healthcare providers in remote areas also benefit from AI tools, given their isolation from a connected professional community. Clinical decision support systems can provide advice on diagnoses, treatment, and patient management.

Of course, the use of AI in healthcare has its fair share of criticisms. As Basu’s review mentions, it can be difficult to find large amounts of high quality data to train models due to issues of patient privacy. Data sharing is often limited or even prohibited in certain contexts, and patients have a right to choose what information is released. This can result in incomplete or unreliable datasets. Models should also be trained on the best possible representation of the intended population, so that the algorithm isn’t biased. A review written by Margaret Chustecki reminds readers that AI can make mistakes just like humans, which is a major reason why people are hesitant about its applications in a field as important as healthcare. Chustecki’s review also addresses the concern that AI will take over too many jobs in the health industry; while it may eliminate some roles in administration, secretary services, and research and development, it won’t replace clinician roles entirely. AI proficiency may actually be a resume booster, in the same way that programming skills used to be. Another challenge involves liability and accountability. For example, if an algorithm-driven treatment or decision has detrimental effects, who is responsible? The clinician? The hospital as an organization? The AI developer? These are just a few questions that keep many people wary of AI’s role in healthcare.

There is still uncertainty surrounding AI use in healthcare, and many experts agree that clinicians will likely have to adjust their scope of knowledge to adapt to future changes. One goal is to expand the use of AI-powered tools, both in scaling and access. What might this look like? It would be the norm for hospitals to have accurate, robust imaging and analysis tools, as well as early detection and screening technology. It might massively improve vitals technology for nurses; AI could increase the accuracy of heart rate, oxygen, and blood pressure readings. All of these expansions could have an even greater impact when considering how AI can make personalized medicine possible.

The healthcare industry has an opportunity to change millions of lives by harnessing technology in order to provide fast, quality treatment and support tailored to the individual.

Dr. Aarti Sathyanarayana, an assistant professor at Northeastern University, directs the SATH Lab, which uses device data to inform on and improve health. Using a two-pronged approach, the lab analyzes daily behaviors and lifestyles along with physiological patterns helpful for diagnosis in order to get a clearer picture of an individual’s overall health. A career in digital and health sciences has enabled her to appreciate the powerful potential of AI in healthcare, even in the face of its risks. “The fear is all natural. We should be scared, we should be asking questions, we should be pushing back, and we should be really critical of these tools, but they can be used to help us,” said Sathyanarayana. “It’s a really exciting time to be in [this] field…while we should be skeptical, we should be really excited for how this can change our lives.” The question isn't whether AI will transform healthcare, because it already has. The question is whether this generation will actively participate in shaping that transformation responsibly.