Grounded in behavioral theory, powered by artificial intelligence, and nurtured with natural language processing, virtual health assistants (VHAs) are capable of conversing with a person about his or her health.
They can provide education, assist in motivation, break down tasks so people will actually consider doing them, track and trend health data, provide summary graphs and charts, and talk about goals, feelings, barriers, fears, and effectively emulate a nurse or health coach. VHAs will soon become tomorrow’s motivators.
With preventable and manageable chronic conditions such as diabetes or prediabetes affecting nearly 120 million Americans, and approximately two-thirds of the nation now qualifying as either overweight or obese, few would argue that our overall health has deteriorated in the past quarter century. Left unaddressed, this epidemic of lifestyle-induced disease will bankrupt our great nation.
VHAs have the ability to reach patients on a daily basis—multiple times a day even—significantly expanding the reach of healthcare professionals, and allow patients to be informed about their own health, and make data-based choices. Using motivational interviewing techniques, these tools can provide a daily coaching bridge between physicians, pharmacists, and patients.
Looking for Solutions
Pharmaceutical companies focus on drug discovery as a solution, and thousands of smartphone applications are being developed to change our destructive health behaviors. But lacking in both of these approaches is a true solution that draws on the basic human desire for a relationship or companion in our search for better health.
Health behaviors such as whether we exercise or make healthy food choices are the result of a complex interaction of influences on a number of dimensions. If you ask people why they behave in a certain way, many cannot answer, or their answer does not hold up to observed practice (eg, how often have we been told by patients that they are taking their medication when in reality we see a medication possession ratio of 50%?).
Negative influences on health are ever-present. Think of the millions of messages people are exposed to through advertisements during their lifetime, many of which are created by vendors of Big Tobacco and Big Food promoting unhealthy habits and unhealthy amounts of fat, salt, and/or sugar. These messages, advertisements, and even the taste and texture of unhealthy food use psychological manipulative techniques to influence daily behavior and therefore health; albeit in a less than desirable way.
Psychologists have been studying health-influencing factors for decades, including dozens of theories and models to understand health behaviors, and hundreds of articles to help physicians and other healthcare professionals design effective methods to positively influence health behaviors.
Health Behavior Theories and Models
Karen Glanz, PhD, MPH, a professor in the Department of Biobehavioral Health Sciences at the University of Pennsylvania Perelman School of Medicine, and lead editor of Health Behavior and Health Education: Theory, Research, and Practice, now in its fourth edition, suggested that the most dominant theories and models are the social cognitive theory, the transtheoretical/stages of change, the health belief model, social support and networks, and good, old-fashioned patient–provider communication.1
From a 30,000-foot perspective, these health behavior theories and their accompanying models—which, when implemented properly, have been proved to work in clinical trials—focus on individual and interpersonal factors as well as individual stress and coping mechanisms. The number one behavior-related disease, diabetes, has continued to escalate beyond our imagination providing proof that these theories and models—which primary care physicians and ancillary providers have used as a basis to educate their patients on the need for exercise, lifestyle changes and exercise—have not worked.
We need to be able to deploy a behavior change agent daily, not weekly or monthly as in the current models.
The Physician–Patient Relationship
Interpersonal communication is one of the most critical factors in changing health behavior, and communication between physicians and patients has been studied extensively. Patients demonstrate enormous trust in this relationship, and accept their physician as the most knowledgeable person with whom to interact when it comes to healthcare.
Psychologists have extensively researched this powerful relationship, and singled out several key factors affecting this bond between the physician and patient: fostering healing; information exchange; responding to emotions; managing uncertainty; making decisions; and enabling patient self-management.
Although the physician–patient relationship is powerful, it has failed to keep pace with the growing needs of our nation. There are many reasons for this failure, including the shortage of physicians—especially in primary care, and even more so in underserved populations—the access-barring cost of physician services, the dramatic change in the way patients gather Internet-based information, and the perceived conflict of physician financial incentives to “save money.”
Most physicians agree that a primary reason for the deteriorating physician–patient relationship is the lack of time—which typically consists of only 10 to 15 minutes—during a standard office visit. We must face the fact that to change behavior, a patient needs more than just 1 office visit with their primary care provider every few months, or a visit to their pharmacy to pick up their prescription(s) every month. People spend only a few hours per year in the presence of a trained professional, and that is simply inadequate. They spend >5000 hours a year without direct professional supervision, stumbling their way through their disorders; what they need is an automated hovering approach using an avatar.2
VHAs in the Future of Medicine
The United States needs a new delivery model for changing health behavior.
B.J. Fogg, PhD, Director of the Persuasive Technology Lab at Stanford University, CA, uses a formula that simplifies the most effective theories of behavioral change.3 He states that to change behavior, a person must have motivation; the desired behavior must be broken down into achievable, bite-sized pieces; and then the behavior must be triggered. Again, using the case of obesity, prediabetes, diabetes, and other similar conditions—motivation, capability, and activation—need to occur simultaneously and repeatedly.
I propose that a VHA can provide this model for change. What is a VHA? Film buffs will remember HAL 9000, the heuristically programmed algorithmic computer and artificial intelligence from 2001: A Space Odyssey, and J.A.R.V.I.S. (Just A Rather Very Intelligent System) from the Iron Man and Avengers franchises. Siri, a personal assistant and knowledge navigator found on Apple, Inc.’s iPhone operating systems is also an example of a similar type of technology.
Getting people to adopt healthy lifestyle choices requires a VHA that is focused on healthy behavior. Loading the VHA onto a smartphone and giving it the ability to process natural language opens up all kinds of opportunities to influence behavior.
So, to describe the near future, a VHA will talk to you; remind you to take your medication; instantly calculate how many calories are in a meal; track your caloric intake and exercise; give you a breakdown on your overall nutritional status as measured by free sugar, saturated fat, vitamin, mineral and fiber content; prompt you to exercise; provide a healthy “tip of the day”; and encourage healthier substitutes at restaurants, fast food joints, and stores. It can break down your health goals into smaller, intermediate goals (eg, taking an additional 1000 steps daily vs running a marathon).
VHAs and Behavior Health Models
The convergence of raw computational power, cloud-based data sharing, smart devices, and natural language processing technology that makes up a VHA gives providers a powerful new tool to improve health; one that takes advantage of a limitless number of opportunities to develop a relationship with each patient that is rooted in the most effective methods of behavior change: conversation.
Drawing on the theories and models discussed above, VHAs have the unprecedented opportunity to change the behaviors of large populations, and change the course of our collective health. Through conversation and contextually based programming, a VHA can establish a “personal, health-fostering relationship” with a person, and assist him/her with understanding their condition, managing uncertainty, responding to their emotional needs, determining barriers to behavioral change, self-management, and can even embed itself into our social structure to provide access to other patients working toward the same goals. These are forward-looking applications for a VHA, but currently apps such as fitbit are doing this. The VHA will do the same. Much of what I discuss is in development but could be available by the end of this year (see Figure).
The VHA can also incorporate other wearables and other data sources into its functionality. For example, a VHA could incorporate data from a glucometer, pedometer, sleep monitoring device, and smart pill bottle and graph adherence to a variety of goals that patients agreed to with their provider. It can also be integrated with copay assistance programs to assist with financial issues that can affect adherence. In addition, the VHA can include gamification, competition, and social support functions that allow people to encourage one another in a variety of ways.
Large-scale clinical trials proving that VHAs will work are lacking, but numerous academic studies have been done that demonstrate their efficacy. The leading US researcher in this field is Timothy Bickmore, PhD.4
The VHA is limited only by the imagination of the organization willing to engage in this strategy to improve health. The science and technology are advanced and available, and the need is present. The onus is on early adopters to develop these initiatives.
Our medical system is going through enormous changes because the status quo is not sustainable. We have tried for nearly half a century to change behavior and have succeeded as individuals, but not as a population. The traditional behavioral modification approach of one-on-one is not scalable. We cannot continue to throw more money and professionals at a problem that requires more than what physicians, nurse practitioners, and other healthcare providers can provide. Only through the use of VHAs can we hope to solve the health behavior problems we are facing.
- Glanz K, Rimer BK, Viswanath K, eds. Health Behavior and Health Education: Theory, Research, and Practice. 4th ed. San Francisco, CA: Jossey-Bass, Inc; 2008.
- Asch DA, Muller RW, Volpp KG. Automated hovering in health care—watching over the 5000 hours. N Engl J Med. 2012;367:1-3.
- Fogg BJ. What causes behavior change? www.behavior model.org. Accessed May 26, 2015.
- Relational Agents Group. Projects. www.relationalagents.com. Accessed May 26, 2015.