What is AI?
The term "artificial intelligence," or in the abbreviated form "AI," is widely used nowadays. Let's see what it really means. Artificial Intelligence is intelligence demonstrated by machines, as opposed to the natural intelligence displayed by humans and animals. Artificial intelligence was initially conceptualized in the 1950s with the goal of enabling a machine or computer to think and learn like humans. Modern AI textbooks define the field as the study of "intelligent agents:" Any device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals. Colloquially, the term "artificial intelligence" is often used to describe machines (or computers) that mimic cognitive functions that humans associate with the human mind, such as learning and problem-solving.
How AI helps healthcare
The world population is aging, as life expectancy increases. It is predicted that by 2050, one in four people in Europe and North America will be over the age of 65. This means that health systems will have to deal with more patients with complex needs and, more importantly, that they will need to shift their philosophy from episodic care-based management to long-term care management. This shift is reflected in the value of AI for healthcare, which is now one of the world's highest-growth industries. In 2014, the AI sector was valued at $600 million, and it is projected to reach $150 billion by 2026. AI is so valuable because it simplifies the lives of patients, doctors, and hospital administrators by performing tasks that are typically done by humans but in less time and at a fraction of the cost. By definition, AI in healthcare is referred to as the utilization of algorithms and software to emulate human cognition in the analysis, interpretation, and comprehension of complex medical data.
Robots detecting cancer - no longer science fiction
The rapid development of Artificial Intelligence (AI) has opened up exciting new possibilities in medicine. In healthcare, AI has almost endless potential. Every day we read some news about it.
Clinical judgment and diagnosis are two of the main areas where AI can transform healthcare. There is a multitude of projects underway that employ AI in the early detection of macular degeneration, acute kidney failure, different types of cancer, sepsis, and Alzheimer's disease, among others.
In a 2017 study from Stanford University, it was shown that an AI algorithm was successful in detecting skin cancer against the diagnosis of 21 board-certified dermatologists. In its diagnoses of skin lesions, which represented the most common and deadliest skin cancers, the algorithm matched the performance of dermatologists in 100% of the cases 1. In 2020, Google's DeepMind technology successfully trained a neural network to detect more than 50 types of eye disease by analyzing 3D retinal scans, improving the detection of age-related macular degeneration (AMD), the most common cause of blindness in the developed world dramatically 2.
Most importantly, it is important to acknowledge and emphasize the benefit of combining the powers of the AI algorithms with the powers of the physicians. At the International Symposium on Biomedical Imaging, a competition was held for computational systems programmed to detect metastatic breast cancer from biopsy images. While the winning algorithm made the diagnosis with a 92.5% success rate, combining it with human pathologists' opinion and expertise increased that number to 99.5% success.
Everything that AI can do
AI can learn from large data sets and then use the obtained information to enhance clinical practice. It can quickly acquire useful information from patients' populations to assess - in real-time - risks for the general population (truer than ever now in the Covid era). It can carry out highly repetitive tasks such as analysis of tests, X-rays, CT scans, or inserting data into databases. It can lower the risk of diagnostic mistakes. It can help doctors and healthcare professionals to find the best treatment by updating them with the latest news from scientific publications, textbooks, and clinical reports. It can organize clinical charts and also analyze the performance of specific healthcare institutions. It can help to promote a so-called precision medicine approach, implementing drugs based on a rapid assessment of the mutations for the disease of interest. And these are just some examples 3.
The rapid development of AI techniques has fuelled an active and controversial discussion on whether human physicians will be eventually replaced by "AI doctors." This is probably not going to happen, at least not in the foreseeable future. However, AI can definitely assist physicians in making better clinical decisions or even replace human judgment in certain functional areas of healthcare (e.g., radiology).
AI-driven healthcare – why sharing is caring
Whether it's used to find new links between genetic codes or to drive surgery-assisting robots, artificial intelligence is reinventing — and reinvigorating — modern healthcare through machines that can predict, comprehend, learn, and act.
So, it is clear that AI has enormous potential to streamline and improve diagnosis, treatment, and patient care. To do so, though, it needs to be trained on very large data sets of CT scans, other imaging exams, and general patients' data. AI is the ability of a computer algorithm to approximate conclusions based solely on input data. Healthcare institutions have huge patients' data sets available, from the first presentation to diagnosis, treatment, and final outcome. And scientists who are going to develop AI-based tools need those data to achieve the full potential of AI for healthcare. Etta Pisano, chief research officer at the American College of Radiology, says that "to fulfill the promise of AI in healthcare, medical data will need to be treated as precious to our health as drinking water." However, one of the limitations for the advancement of AI-based tools has been the lack of consensus on an ethical framework for sharing clinical data.
In March 2020, a framework for the ethics of sharing clinical imaging data for AI has been published, adding a new dimension to the role of the patient and other stakeholders 4. Dr. David Larson and colleagues, authors of the report from Stanford University, argue that after clinical data are acquired for the primary purpose of diagnosing and treating the patient, any secondary use of the data should be regarded as a form of public good. Rather than debate whether patients or provider organizations "own" the data, the authors propose that clinical data are not owned at all in the traditional sense, but rather that all who interact with or control the data have an obligation to ensure that the data are used for the benefit of future patients and society. The implications are that the patients do not need to consent to secondary use of their clinical data, and that data should not be sold but rather made available for the development and implementation of knowledge and tools that fulfill societal benefit.
To conclude, we know that in the evolving landscape of health information technologies, AI can transform medicine by improving diagnosis, treatment, and the delivery of patient care. However, this transformation can only happen with proper sharing of healthcare data5. We need dialog and joint effort, both from the fields of medicine and computer science, as well as from policymakers, to navigate the ethical questions surrounding AI and medical data-sharing and to thoughtfully translate ethical consideration into regulatory and legal requirements. Dr Larson's recently published framework is the first promising step in that direction.