Artificial intelligence (AI) can be used in many different ways to improve the U.S. healthcare system. Many providers have already started implementing such technology into modern medical practice, and many more are expected to follow suit, as AI in medical imaging is expected to reach $2 billion by 2023. The potential benefits that AI can provide in healthcare are exponential and generally fall into four main categories.
The Benefits of Artificial Intelligence in Healthcare
First, AI can push frontiers in healthcare by doing things that aren’t possible for humans to do. For example, researchers have used 18-F-fluorodeoxyglucose positron emission tomography (FDG-PET) technology to train a deep learning algorithm to predict Alzheimer’s years before a conventional diagnosis could be made. AI has helped with early diagnosis with other illness as well, such as kidney failure, ALS, and cancer. Therefore, by helping physicians recognize health issues and start treatment earlier, AI can increase patients’ life expectancy and chances of survival by catching conditions much earlier on.
Second, AI can democratize expertise by making specialized care more available. For example, some specialists, such as dermatologists, are in great demand and short supply due to their expertise and the complexities of their patients’ health conditions. As a result, specialists don’t have time to keep track of current research, let alone implement it into their practice. The demands of such medical specialists also make it hard for them to take on more patients, and thus not all those who seek care are able to get it. AI networks can help by assimilating large amounts of medical research data and use cognitive computing systems to search for information related to physicians’ treatment inquiry and provide confidence-ranked answers. Therefore, the less time dermatologists have to spend researching complex treatment methods, the more time they can spend with current and future patients. Relatedly, AI can also give medical professionals more time to directly care for their patients by automating tasks such as record keeping and billing within the health care systems.
Lastly, AI can more accurately allocate healthcare resources by analyzing which patients will benefit most from such medical resources. For example, when a physician enquires about treatment options, the AI’s suggestion will indicate which medical resources are needed to provide a specific patient with the treatment that is most likely to succeed, such as a new kidney for a transplant. Thus, AI may specify which patients will benefit from treatment resources more than other patients. The role of doctors and AI in determining who gets what treatment can have significant implications for health law.
AI Implications in Health Law
These examples show how the healthcare system can improve from AI. Such improvements from AI may give rise to the need for fundamental changes in the standards that physicians are held to within their medical practice. For example, integrating decision support AI that provides suggested treatment methods could greatly affect medical malpractice claims, especially if the legal standard of care shifts to embrace this technology.
Currently, the standard of care for health professionals is that which a minimally competent physician with similar recourses would do under those circumstances. Thus, doctors have an incentive to treat a patient according to the method most widely accepted in the relevant medical community. If a physician is using AI that recommends a treatment method which deviates from the national practice, there is an incentivie to ignore that recommendation and continue with the more common method in order to avoid a potential medical malpractice claim, even if the latter is less likely to help the patient. That is to say, under current law, in a situation where a physician ignores an AI treatment suggestion that would be more helpful to the patient, and the patient is harmed by the choice to ignore the AI, the physician would not be liable for medical malpractice since the standard of care, i.e., the commonly accepted treatment method, was upheld.
However, as AI becomes increasingly more integrated into medical practice, and as a result overall medical care, including treatment results, continue to improve, we might imagine a shift in the law in order to accommodate and incentivize the adoption of AI. For example, as AI suggested treatments provide more accurate and consistent success in patient health outcomes, the standard of care may shift away from the most commonly used physician treatment methods, and move towards AI’s top suggested method of treatment, since it is faster and more effective in choosing the most successful treatment given the complexities of a specific patient’s condition.
Such a shift in the legal framework of health law would have significant liability implications for medical professionals. For example, a physician who ignores the AI’s treatment suggestion and harms the patient as a result would no longer be shielded from malpractice liability because at that point such treatment would fall below the standard of care. This shift is likely to affect legal implications for medical institutions as well. For example, although such a scenario may be a long way off, if treatment-suggesting AI systems become widespread enough, a hospital could be held liable for patient harms in cases where such technology is considered to be sufficient medical equipment and facilities for all medical institutions, yet that hospital has refused to integrate it. Based on recent indications from large healthcare organizations, this may be a welcomed yet gradual change, as some have expressed an interest in using AI but have some concerns about the effect this integration could have on things like accuracy and workflow.
Integrating AI into our modern healthcare regime is likely to affect many areas outside of just health law, such as insurance reimbursement, privacy law, discrimination claims, government regulation, property law, and tort law. While there are potential benefits that AI can provide in all these areas, such benefits are likely to bring potential issues as well. The law’s role in such integration is both a normative and substantive matter and will have significant impact on technology implementation. With the current trend of increased technology presence, at least in modern healthcare, the law must soon address the many benefits and potential dangers of largescale integration of artificial intelligence.
* Shelby Engelbrecht is an Associate Editor on the Michigan Technology Law Review.