It has been said that the practice of medicine is both an art and a science. For example, different patients react differently to the same medication for the treatment of anxiety, and what works for one patient will not work for another. It can be a trial-and-error process to find the correct medication and optimum dosage. Why do side effects to medications or complications to treatments arise in some patients and not others? Every patient is unique, and the art of medicine is the component of the practice that addresses such uniqueness with compassion and care. Will artificial intelligence (AI) change all of that?

Of course, medicine is also a science, and health care is no stranger to technology. Health care is and has been for many years highly dependent upon technology, from x-ray machines and laboratory equipment; to more advanced medical equipment such as MRIs, ultrasounds and CT scanners; to systems that dispense medications within the hospital; to electronic medical records.

A hospital’s electronic health record may connect with the electronic health record of other hospitals and physicians to create a “health information exchange” (HIE), so that providers have access to all of a patient’s medical information regardless of where he or she may have been treated. The goal of providing access to a patient’s comprehensive medical information is to decrease medical errors as well as eliminate the provision of duplicate tests or procedures that may occur when a specialist may not have access to another provider’s records.

Access to all of that information, however, is creating an information overload for many physicians—yet under the current reimbursement system, where providers are paid a set fee per patient encounter or procedure, providers are being pressured to see more patients in a given day, thereby having less time to spend on any given case. Similarly, hospitals are paid a set amount for the entire hospital stay for a specific patient diagnosis. Accordingly, hospitals are incentivized to discharge patients much quicker than they once did.

Medicare has also created the Medicare Shared Savings Program (MSSP), which in effect focuses on the quality or value of the care that is delivered. Providers are still paid pursuant to a fee schedule but have the opportunity to earn additional payments if they save the Medicare program money and improve patient outcomes. Commercial insurance companies have adopted similar types of payment approaches.

The MSSP and other value-based payment programs, in essence, require providers to practice differently. Primary care physicians are essentially assigned to track and coordinate their patients’ care across all types of providers, whether it be specialists, therapists, home health providers or whoever is involved in that patient’s care. The providers must meet certain quality benchmarks for various medical conditions such as chronic diseases like asthma or diabetes in order to receive their additional payments.

The goal is for the physicians to keep their patients healthier, prevent lengthy hospitalizations or re-hospitalizations; identify where the patient has gaps in care (for example, if he or she is in need of seeing a specialist, or is missing appointments); utilize treatment protocols based on best practices for a given condition; and work to make sure a patient complies with her treatment plan.

The use of artificial intelligence will be critical to the success of such MSSP and other value-based programs. Artificial intelligence in this context involves data analytics coupled with the ability to predict outcomes and decision-making capabilities for the development of treatment plans based on empirical evidence. Such applications are geared toward saving costs as well as improving patient outcomes.

In order to achieve these goals under MSSP or other value-based programs, physicians must have ready access to up-to-date information about all aspects of the patient’s care, across all providers, as well as aggregate data across a patient population that can be used to predict outcomes, risk factors for disease and to develop evidence based treatment protocols.

Large amounts of patient clinical data, along with claims and financial data, must be aggregated, analyzed and reported to providers and payers in a timely and meaningful way. Data analytics and artificial intelligence with predictive analytics and decision-making tools in particular are essential to generating information that enables providers to make quick decisions without contributing to data overload.

Artificial intelligence may also be used in hands-on delivery of care. There are several robotic systems, such as the da Vinci robots that assist—but do not replace—surgeons in operating on prostate cancer patients, enabling them to better reach tissues and resulting in improved outcomes. Certain medical devices operate autonomously—implanted defibrillators, for example, which monitor a patient’s heart rate and may automatically deliver a shock if the patient’s heart rate is too fast or irregular. However, such devices are inserted by, and used under the direction of, physicians. The question remains as to whether we will see robots that not only assist the physician in surgical or other procedures, but which operate solely independent of and without the need for a physician medical judgment.

Regardless, these clinical applications raise interesting legal questions. For example, if there is an untoward outcome, who is liable—the physician, the manufacturer of the robot under a product liability claim, or both? These machines and devices also store patient data, leading to HIPAA privacy and security concerns in connection with the sharing and storing of such data. On the other hand, artificial intelligence may help health organizations in addressing cyber security by predicting ransomware attacks and other types of unauthorized activity.

Another issue: if a robot is delivering care, how is it regulated? The Food and Drug Administration currently regulates devices, but the practice of medicine is regulated at the state level through licensure by boards of medicine. Will additional licensure or regulatory scrutiny be required if the devices supplant the physician’s practice of medicine? Will such technology become the “standard of care,” such that its use is expected in patient care and the failure to use it could result in malpractice?

Currently, the art remains in medicine as physicians remain hands-on in the ultimate decision making about patient care. Will medicine go the way of the driverless car, or will it be more like today’s state-of-the art automobile, where a human remains in control but with added safety features in the form of driver assistance technology?

Time will tell.