With each new year, numerous reports, columns, and blogs are written to project the upcoming technologies and trends in health care. Several 2017 technology trends point to the growing use of artificial intelligence (AI) in health care. Gartner's Top 10 Strategic Technology Trends for 2017 (Cearley, Walker, & Burke, 2016) highlight three top trends: AI and advanced machine learning, intelligent apps, and intelligence things. Let's look at each.
First, certain technologies and specific techniques, such as deep learning, neural networks, and natural language processing, are encompassed within the AI and machine-learning concept. These techniques create software programs that are more than just rule-based systems. Rather, these systems can "understand, learn, predict, adapt and appear intelligent" (Cearley et al., 2016). Their ability to learn is key to their functionality. For example, a machine-learning system can analyze numerous electronic health records (EHRs) and recommend potential effective treatments. As more datasets are added, the system can learn and adapt the recommendations, for example, adding genomic data to the EHR database.
Intelligent apps, the second trend, refer to virtual personal assistants (VPAs). VPAs help users with everyday tasks, for example, sorting email or answering simple questions (just as SiriTM and CortanaTM do on our smartphones). VPAs will become more available in health care in the coming year. Intelligent things, the third trend, break down into three distinct categories: robots, drones, and autonomous vehicles.
In discussing health information technology trends for 2017, Health Data Management (2016) also named AI as the first trend, stating that, although AI exploded in health care in 2016, applications were typically very specialized. More general use is projected, which "will mean better access to actionable intelligence." Padmanabhan (2016) echoed this notion: "In 2017, we are more likely to hear terms such as 'cognitive computing' and 'artificial intelligence'[horizontal ellipsis]and less likely to hear the term 'big data analytics,' which now seems to be limiting in its description of the actual work being done in advanced analytics."
AI BASICS
The term artificial intelligence is not new. It dates back to the 1940s and 1950s, and Turing (1950) asked if machines can think. AI was even used in health care in the 1970s, for example, with a system called MYCIN, developed by Stanford University, that identified bacterial infections and recommended treatments (Shortliffe, 1976). MYCIN contained three components: a knowledge base created by experts, an inference engine with rule-based algorithms, and a user interface. Although there were other instances of expert system designed in that era, there was never a critical mass of users to adopt them for clinical practice. Most health care professionals did not see the need for machines to tell them how to practice.
Circumstances have changed. According to the Executive Office of the President, National Science and Technology Council Committee on Technology (2016, p. 6), the current motivation for AI was "driven by three mutually reinforcing factors: the availability of big data[horizontal ellipsis]dramatically improved machine learning approaches and algorithms[horizontal ellipsis]and the capabilities of more powerful computers."
COGNTIVE COMPUTING
Cognitive computing (CC) is an emerging term that some view as a subset of AI. Kelly (2016, p. 1) believes that the future of technology will be "cognitive and not artificial" and defines CC in terms of "systems that learn at scale, reason with purpose and interact with humans naturally." One other distinguishing factor is that CC can handle "unstructured data," whereas AI applications are typically based on structured or numeric data. Marr (2016) summarizes CC as "a mashup of cognitive science - the study of the human brain and how it functions - and computer science, and the results will have far-reaching impacts on our private lives, healthcare, business, and more." Kelly (2016) presents CC within a historical context: first, there was the tabulating era of computing (1900s to 1940s), followed by the programming era (1950s to 2011), and now the cognitive era (2011 to present).
Weber (2015), who describes how CC helps physicians analyze data (e.g., clinical notes, EHRs, lab results, images, scholarly journals, clinical guidelines), explains that CC is one flavor of AI. I like how he describes AI as "the ability of computers and computerized instruments to exhibit cognitive behaviors characteristic of human thought: to read and absorb knowledge; to sift data and learn from it; to reason; to plan ahead; to perceive the environment and its changes; to understand and communicate in natural, spoken language." He further describes this flavor as the "use [of] computational, data storage and analytical strengths of computers to undergird and enhance human capabilities in situations that are too complex for the brain we were born with to handle readily, unaided." In essence, AI will enhance, not replace, humans.
Claburn (2016) noted that IBM refers to its "work as being focused on augmented intelligence, systems that enhance human capabilities, rather than systems that aspire to replicate the full scope of human intelligence." These thoughts are echoed by the dean of Carnegie Mellon's School of Computer Science. Moore, in an Information Week interview (Claburn, 2015), stated that "98% of AI researchers are focused on engineering systems that can help people make better decisions rather than simulating human consciousness." Dean Moore was also interviewed by Charlie Rose (https://charlierose.com/videos/29644).
EXAMPLES OF AI IN HEALTH CARE
A report from the Executive Office of the President (2016) states that humans will use augmented intelligence tools as a way to assist and expand their productivity. The report provides an example from health care: "AI technology such as IBM's Watson may improve early detection of some cancers or other illnesses, but a human healthcare professional is needed to work with patients to understand and translate patients' symptoms, inform patients of treatment options, and guide patients through treatment plans" (p. 18). So let's look at examples of AI in health care.
Perhaps the best known example is IBM's Watson - you probably remember when Watson was a winning contestant on the TV game show Jeopardy. IBM Watson Health uses CC to ingest, combine, and understand massive volumes of structured and unstructured data. It can reason and learn from its interactions with the data. For a population health example, see the webinar at http://www.ibm.com/watson/health/population-health-management/resources/welcome-. And see how Watson uses an application (Phytel) to examine your data and makes recommendations to reach out to patient populations for their mammograms. A clinical nurse specialist describes how the program helped detect an early-stage breast cancer for one patient (http://www.ibm.com/watson/health/population-health-management/resources/using-ou). Watson also helps clinicians, specifically in oncology, to facilitate evidence-based treatment decisions for their patients. Learn more at the Watson Oncology website (http://www.ibm.com/watson/health/oncology/).
Watson is also a student, learning from the cancer experts at Memorial Sloan Kettering Cancer Center (MSKCC). IBM Watson Health has a collaborative project with MSKCC "to train Watson Oncology to interpret cancer patients' clinical information and identify individualized, evidence-based treatment options that leverage our specialists' decades of experience and research" (http://www.mskcc.org/about/innovative-collaborations/watson-oncology).
Hallmark Health Care Solutions offers Einstein II (http://einsteinii.com), a workforce optimization scheduling system that uses AI and machine learning to construct and use rules that meet acuity needs, regulatory requirements, skillsets, and employee status. The system uses a predictive engine to provide real-time information as well as future staffing needs.
AlME, a virtual assistant that can interact with patients using natural language, is targeted for use with patients who have chronic diseases. Personalized to the patient, it helps patients better manage their health (http://www.nextithealthcare.com/learn.php) with medication alerts and coaching.
Babylon, a UK company (http://www.babylonhealth.com), uses AI in its symptom checker, which processes billions of symptom combinations to provides a likely diagnosis and course of action (Middleton et al., 2016). The product has many other features such as a virtual visit with your general practitioner, digital assistance with medication adherence, and clinical record and pharmacy services. The National Health Service provides more details about its many services (http://www.babylonhealth.com/uploads/home/babylon-NHS-brochure.pdf).
To summarize, there is no doubt AI will continue to have an impact on health care and will be used more and more by health care professionals, even nurses. I would like to leave you with two thoughts for the new year.
The first is from the Moore interview with Charlie Rose on AI and robotics: "Interestingly, something like a nurse or a teacher[horizontal ellipsis]whose real job is understanding the people that they are interacting with all the time will be much harder to automate. Don't see these positions to be eliminated for decades."
The second is from an editorial by Klasko (2016): "The future of healthcare will not be found in AI, although AI will contribute more than we imagine. The future of healthcare will not be found in augmented reality (AR), although AR will transform our experiences. My prediction is that healthcare will be driven by augmented intelligence that places the locus of meaning back at home - back with the patient - and allows the physician to get back to the role of guide."
As always, you can email with your thoughts at mailto:[email protected].
REFERENCES