Mobile Computing and the Decentralization of Health Care

Contact Our Team

For more information about how Halldale can add value to your marketing and promotional campaigns or to discuss event exhibitor and sponsorship opportunities, contact our team to find out more

 

The Americas -
holly.foster@halldale.com

Rest of World -
jeremy@halldale.com



Eliot Grigg, MD, hypothesizes on the decentralization of health care and the use of big data to predict epidemic outbreaks.

Eliot Grigg, MD, discusses the evolving role of hospitals, doctors and patients through the use of technology and meaningful data collection.

Ever since the valetudinaria (hospital precursors) of Ancient Rome, the provision of health care has become increasingly centralized. There was a time when house calls were the norm and hospitals were reserved for segregating the incurable. Today the vast majority of patient encounters occur in doctors’ offices or specialized medical centers. Data-gathering requires the physical proximity of a doctor or nurse, and the fidelity and volume of information is directly related to the acuity of care; intensive care units gather data continuously, but outpatient visits occur only every few months or years. Once a patient leaves the hospital or office, little or no information is collected until the patient returns. For most clinicians, the only source of information from the intervening period is the patients themselves.

With the advent of mobile technology, there now exists an opportunity to gather information from patients in real-time, far from any medical setting. There is an emerging trend – to decentralize health care. Today, nearly two-thirds of Americans own smartphones and 19% rely on them to access online services and information (Smith, 2015). A more accurate description is that the majority of Americans now possess wireless, Internet-connected computers on their person, which presents an enormous opportunity for health care.

The home monitoring trend – the so-called "quantified self" among technology circles – began in the consumer fitness realm. Today, individuals can monitor parameters like their physical activity, calorie intake and sleep quality with wrist-worn pedometers and gyroscopes. Some companies are creating scales, blood pressure cuffs, electrocardiograms, pulse oximeters and blood glucose monitors that connect to smart phones, and researchers are beginning to use these systems to gather data from patients at home (Pare, 2007). With recent releases like the Apple Watch and the Microsoft Band even consumers are now able to continuously and routinely monitor a whole variety of health parameters.


The physiologic sensors on the back of the Apple Watch.

Propeller Health (formerly Asthmapolis) uses a small sensor affixed to a traditional metered-dose inhaler for asthma medication to wirelessly record the time and GPS location of each administration (Aldridge, 2011). The Scanadu Scout, a small handheld device able to measure temperature, blood pressure, and blood oxygen levels, has been likened to a modern Star Trek ‘Tricorder’


The Scanadu Scout mobile monitoring device

(Smith, 2015). Someday in the not-too-distant future, patients – both healthy and sick – will be continuously monitored 24 hours a day, seven days a week, from the moment they are born until the day they die. The question then becomes what to do with all of this data, and how will this new data stream change the doctor-patient relationship?

Feedback for Behavioral Change

When executed thoughtfully, preventive medicine is both beneficial and cost-effective, and the role of lifestyle choices in patient outcomes is considerable (Goddell, 2009). Unfortunately, most preventive care relies on behavioral change, which is notoriously difficult. Part of the challenge is the timing of incentives: patients struggle with diet, exercise and smoking on a daily basis, and yet the reinforcement from health care providers occurs only during periodic encounters. The outcomes are often many years down the road, and the effects of many chronic interventions (like blood pressure medications) are not immediately perceived.

The fitness industry has made inroads with continuous feedback and personalized rewards. Connected pedometers count the number of steps taken daily, the progress towards a pre-set goal, and rewards for accomplishments. The goal is to make patients aware of the consequences of everyday choices and empower them to take control of their daily routines. From the real-time fuel economy data on the dashboard of a Toyota Prius to the sobriety coins of Alcoholics Anonymous, numerous entities are using badges, leader boards, rewards, challenges and virtual currencies to motivate consumers. Feedback and gamification are by no means panaceas, but they are more consistent and constructive than inquiring about weight loss at an annual physical. There is an opportunity to not only track everyday habits, but also motivate patients throughout their normal routines, because many chronic health states result from the accumulation of numerous small, daily decisions.

Just-In-Time Feedback

Behavioral change for long-terms goals is valuable, but are difficult to quantify on an individual basis. For patients with active medical conditions, the consequences for more immediate prevention using real-time feedback are much more apparent. A heart failure patient tracking daily weights can receive immediate feedback about sodium intake, and a diabetic patient tracking blood glucose can receive real-time advice about the next insulin dose. These interventions can prevent very costly and morbid hospital admissions – with the associated risks of errors or infections – within days or weeks, rather than years. The benefit to patients is immediate, and the potential for health care savings is considerable.

Initially, generic algorithms (e.g. insulin sliding scales) can be used on all patients. Over time, after enough data is gathered, individually-tailored algorithms can be developed based on a patient’s response to past interventions. When such a system is in widespread use, large datasets can be used to develop predictive algorithms that anticipate patients’ needs, rather than react to them. Someday, a heart failure patient will step on a scale, and a mobile application will advise him to take an extra dose of diuretic because based on prior experience, and with Thanksgiving a few days away, his risk of a hospital admission in the next two weeks is doubled.

Patient-Centered Care

Effective medical decision-making depends on reliable data, yet when patients arrive in clinic, the primary source for information from the intervening months is their own recall. While subjective complaints have been the cornerstone of diagnostic medicine for centuries, patients are not reliable as the sole source of information for a myriad of reasons: (1) they are inherently biased and anxious, (2) they are not trained to describe symptoms accurately, (3) they may be extremely young, old or incapacitated, and (4) they rarely track and consistently record changes. Clinic visits would be much more efficient and effective if they were based on data – either objective or subjective – gathered on an hourly, daily or weekly basis. Some practitioners enlist diaries for pain or asthma, but they usually rely on voluntary patient input. To be a long-term solution, devices need to prompt and motivate patients or, better yet, gather data automatically. Initially, patients will download data periodically in clinic, but eventually the data will be continuously transmitted to physicians from the patient's home, to allow for “over the air” corrections without a clinic visit.

Such data will allow for better-informed and more personalized decisions. Despite advances in evidence-based medicine, individual providers never know precisely how any one patient will react to a specific intervention. Instead of the statistical gamble that providers enlist today (or waiting for pharmacogenomics to prognosticate), data from individual patients should inform patient-centered care, which will allow providers to tailor decisions based on the variations between specific patients real-time. For example, the rate of metabolism of methadone varies widely between patients, and being able to monitor an individual’s response from home would not only increase safety but also save the patient and the health care system extra days in the hospital.

Epidemiological Surveillance

When a stream of high-fidelity data is continuously gathered throughout the population, epidemiological surveillance will be able to achieve a timeliness and fidelity that is currently impractical. Much surveillance today requires contact with a health care institution, depends on voluntary reporting and cannot be synthesized with multiple data streams. In situations where timeliness is key, like an infectious outbreak, waiting for patients to present to their doctors consumes valuable time. A sudden increase in meter-dose inhaler use in an area might signal an air quality issue, or a sudden increase in over-the-counter flu medications may provide an early warning for a brewing influenza outbreak. Combining datasets like over-the-counter flu medication sales with online searches for “influenza” could increase the specificity of early warning systems.

On a more routine basis, such a pervasive data stream would allow longitudinal studies to be performed on large populations that is not possible today. For example, phase four clinical trials (after FDA approval) could help detect rare side effects from medications that would only be seen after millions of patients started using the drug. A group at Microsoft combing through web searches uncovered an interaction between medications leading to high blood sugar that was discovered only after both medications were already brought to market (White, 2013).

Practice-Based Provider

One of the challenges in health care today is judging individual provider performance. Administrators, regulators and patients obviously have reasons to compare physicians, but even the physicians themselves get little concrete feedback about their own performance. In a given institution, colleagues may share anecdotal experience, but they rarely have concrete data with which to identify best practices. For example, a group of anesthesiologists may have different strategies for preventing post-operative nausea, but they are unlikely to have access to local data (rather than generic evidence-based guidelines) to identify whose strategy is most successful. Equally important, the physicians receive little or no feedback once the patient leaves the hospital. Real-time feedback regarding the experience of local patient populations can help continuously improve performance. Once this data is being gathered at multiple institutions, it can be used to rigorously evaluate local practice models and compare performance between providers or institutions. Current metrics for comparison tend to be overly specific (e.g. beta-blocker administration rates) or overly general (e.g. mortality rates). Somewhere in-between is a much richer dataset based on real-time patient outcomes, gathered both in the hospital and in the home.

Flattening the Healthcare Pyramid

When a similar amount of data is gathered from patients at home as in the hospital, the relationship between providers and patients will change, and patients will spend as little time in the hospital as possible. Patients used to spend a week in the hospital after an appendectomy, but today some leave the same day. Hospitals will enjoy the cost-savings, but the more significant pressure will come from patients themselves. Nobody wants to go to the hospital, change into a dehumanizing gown and attempt to recover surrounded by sick people. Hospitals are a utilitarian compromise – an aggregation of resources from an analog age. Almost everything in health care is structured for the convenience of the providers, not the patients. A noteworthy example is hospital-acquired infections – a major source of preventable morbidity – that are the direct result of congregating sick patients in the name of efficiency and accessibility.

Just as clinical responsibilities are being pushed down the provider hierarchy – medical assistants are performing tasks previously done by registered nurses, nurse practitioners are performing tasks previously done by doctors – the location of care will be pushed down the healthcare acuity pyramid.


Where most health care is provided will change in the future as more and more health information is gathered from outpatients.

Care currently provided in clinics will be done in the home; much of the care provided in hospitals will instead be done in outpatient clinics (including most surgeries). Inpatient facilities will be reserved for only the most complex procedures requiring expensive machinery and physical proximity. Birth and death will largely return to the home, which is where they started. The once-daily medical rounds still pervasive in medicine will become a 24-hour cycle of decision-making, made possible by being continuously connected to patients. As the majority of health care services move away from fixed facilities, a new industry will emerge for same-day delivery of physical resources – like an Amazon.com for health care – which will eventually be supplanted by 3-D printers.

Conclusion

Computers will not replace doctors for a number of reasons (such as risk, uncertainty, empathy and complexity), but their role will evolve. As hospitals transform into information hubs – with massive data warehouses and tele-surgery centers – doctors will become the chief information officers for their patients. Ultimately, the reasons that patients seek out doctors will change. Once they are not the sole sources of medical knowledge, doctors will become more like guides, helping patients navigate the mountains of medical information accessible online – and, of course, performing technical procedures. Patients will likely present armed with potential diagnoses from online algorithms, and doctors will be tasked with vetting sources, filtering irrelevant data, and aligning options with patients’ goals. Rather than simple diagnosticians, doctors will become arbiters of uncertainty – thriving in the gray areas and willing to take responsibility when the answers are conflicting or unclear, and pushing medical knowledge ever forward into the unknown.

About the Author

Eliot Grigg, MD received his BA in Government from Dartmouth College. He spent a few years as a Research Associate at the Thayer School of Engineering at Dartmouth, doing research in mobile health technology and consulting for the government regarding bioterrorism and the future of health care. He received his MD at George Washington University and completed his residency and fellowship in pediatric anesthesiology at the University of Washington (UW) in Seattle. He is currently an Assistant Professor of Anesthesiology and Pain Medicine at UW, and practices at Seattle Children's Hospital, with research projects involving information technology design, human factors, and engineering in the operating room.

REFERENCES

  1. Smith A. Pew Research Center. US Smartphone Use in 2015.

http://www.pewinternet.org/files/2015/03/PI_Smartphones_0401151.pdf

  1. . Published April 1, 2015. Accessed October 8, 2015.
  2. Paré G, Jaana M, Sicotte C. Systematic review of home telemonitoring for chronic diseases: the evidence base. JAMIA. 2007; 14(3): 269-77.
  3. Aldridge D. Project Red and Asthmapolis. CIN: Computers, Informatics, Nursing. 2011; 29(10):542-3.
  4. Smith C. Scouting for Approval: Lessons on Medical Device Regulation in an Era of Crowdfunding from Scanadu’s “Scout”. Food Drug Law J. 2015; 70(1): 209-35.
  5. Goodell S, Cohen J, Neumann P. Robert Wood Johnson Foundation. Cost savings and cost-effectiveness of clinical preventive care. http://www.rwjf.org/content/dam/farm/reports/issue_briefs/2009/rwjf46045. Published September 2009. Accessed October 8, 2015.
  6. White RW, Tatonetti NP, Shah NH, Altman RB, Horvitz E. Web-scale pharmacovigilance: listening to signals from the crowd. J Am Med Inform Assoc. 2013; 20: 3.

Related articles



More Features

More features