Pocket doctor: smartphone-based chronic disease monitoring

Diane Strong, Emmanuel Agu, Peder Pedersen and Bengisu Tulu of Worcester Polytechnic Institute’s Healthcare Delivery Institute dial up Practical Patient Care to discuss their vision for smartphone-based chronic disease monitoring.


As chronic disease management becomes a more intense problem in the US, the healthcare delivery system is struggling. Patients with diseases such as diabetes can see their primary care physician for 15 minutes or so, only a few times a year, yet their condition likely requires daily attention.

This means that those with chronic conditions are essentially responsible for their own disease management, as much as, or even more than, their physicians. So, how well are they doing? With the rising prevalence of type 2 diabetes and heart disease, and associated contributing factors such as obesity and physical inactivity, statistics indicate they are not doing well.

A key part of the solution could be technology. While your doctor is only occasionally available, your mobile phone is usually in your pocket. If it is a smartphone with significant computing and storage capacity, it could function as a personal health assistant that is always there to manage health information. It could be capable of recording your health status in real time and providing personalised guidance for managing chronic conditions.

It is already possible to develop smartphone apps for supporting chronic disease management. For example, physical activity-tracking apps use a smartphone's accelerometer and its GPS tracking features to collect data automatically. Similarly, a smartphone's Bluetooth feature can gather data from nearby Bluetooth-enabled devices such as scales and glucometers. The phone's camera can collect images for analysis or scan barcodes at supermarkets to provide automatic entry of nutrition values for daily diet entries. These features are important because users of apps generally prefer automated data entry over manual input.

While some of these apps have a large user base, most are still at a prototype stage, even if they are already available. For example, many available diabetes apps do not include the core functionality needed to support those with diabetes. Of the 162 diabetes apps available from Google's Android Market, only six supported these tasks. Clearly, there is much more to do to deliver apps that truly meet the needs of those who want to, and need to, better manage their health, including their chronic diseases.

The HDI diabetes app

The Healthcare Delivery Institute (HDI) at Worcester Polytechnic Institute is broadly tackling the problem of creating better designs for delivering healthcare that take advantage of the many newly available technologies. One of HDI's programme areas is designing and developing a variety of mobile apps to realise the vision of the smartphone as a personal health assistant. Faculty members and their students are designing and developing apps to help manage different conditions such as advanced type 2 diabetes, obesity and weight loss, and weight gain during pregnancy. These apps use various devices to collect health measures and algorithms in order to analyse changes over time.

HDI's diabetes wound analysis app illustrates what the Institute has learned about design and development. Currently, it is a prototype intended for use by those who have advanced type 2 diabetes. Such patients commonly have diabetic foot wounds that take months to heal and require daily care, which patients (or their caregivers) must deliver themselves. They should be tracking their glucose levels, managing their weight, engaging in physical activity and eating a healthy diet. The app can provide support for all of these issues.

In the process of designing and developing these apps, HDI discovered three broad principles that are guiding its work. Based on these principles, the Institute works differently from other healthcare app teams and, as a result, expects to provide apps better aligned with achieving an improved healthcare delivery system.

Principle 1: apps must be medically sound

A medically sound app is clinically accurate and behaviourally appropriate. While this may seem obvious, from talking with other healthcare app and device development teams, HDI was surprised to find many of these teams comprised entirely technologists. The Institute's diabetes app team consists not only of four professors and students, but also two physicians, a clinical psychologist and a wound specialist from the University of Massachusetts Medical School. They provide expertise in diabetes care, wound care, and behavioural medicine. With their advice and active participation in this project, the app is medically sound.

Medical soundness shows up in the app in a variety of ways. The health measures collected and the intervals they are collected at are those that physicians recommend. For example, the wound measures computed from wound analysis algorithms mirror the ones that the wound specialists record when a patient visits a wound clinic. The functionality partly depends on what physicians believe is important, while also taking into account patients' views on what will help them better manage their care. For example, HDI is adding functionality to capture what users eat because diet is critical for managing diabetes. The frequency of collecting measures, for example blood glucose and weight, is based on medical advice received from clinical partners. These medically informed design decisions make the app clinically sound and reliable.

Principle 2: apps must be patient-centric

Being patient-centric increases the likelihood that an app will be used properly. Consistent use is critical for achieving the health behaviour changes needed to manage a chronic disease. Unfortunately, a common problem with apps, especially earlier versions developed for the web rather than for phones, is that people use them for a while but then stop. The lifelong nature of chronic diseases means that the app must go beyond the typical evaluation criteria of being usable and useful, to also engage patients and become part of their daily life. For HDI, this means involving patients as it designs the apps and ensuring that the apps interact with patients frequently by providing meaningful feedback.

HDI has been running focus groups of patients with advanced type 2 diabetes as it develops design ideas to test. For example, the main screen of the app prototype (see image above) has large screen buttons as requested by a focus group. These are necessary for elderly, sometimes vision-impaired patients with advanced diabetes. While the need of elders for larger buttons is known, apps are still often designed with the typically small buttons of most smartphones in mind.

"A technically sound app must ensure that data in the phone is copied to a secure place, with sufficient back-up, and transmitted securely."

Another request from patient focus groups was to avoid keyboard data entry. The app is designed to use Bluetooth when possible to automatically gather data. For example, it works with a Bluetooth glucometer and a Bluetooth scale. Users can enter their weight and blood glucose manually, but a number picker feature is provided, rather than keyboard data entry.

An important aspect of healthcare apps designed for patients is the feedback the app provides to that patient. HDI is working with clinical experts and patients to design a feedback mechanism that is consistent with medical advice and with practices for helping patients change their health behaviours, and that satisfies patients' expectations. For example, the app will include goals for weight because goal setting is important for achieving weight loss. This will allow the app to interact with patients frequently and provide meaningful and helpful feedback to encourage continued use.

HDI learned through interactions with patients that any feedback they receive from the app should be positive and encouraging, even when results are negative. This design criterion is critical because, if users do not like the feedback, they will simply stop using the app. The Institute's goal is for patients to see this as a personal health assistant and continue using it as long as their conditions persist.

Patient involvement during design is essential, but it is not enough. HDI will carry out the standard usability and usefulness testing of its apps as part of the development process. While such testing is important for healthcare apps, evidence of usability problems for existing apps indicates that such testing may not be common. Even usability and usefulness testing is not enough. The Institute also plans to conduct a longer-term trial to test patient engagement with the app and how patients integrate it into their life. Through studying how patients use the apps, it will be better able to design apps they will continue to use, which is the only way of encouraging the better behaviour that leads to improved health.

Principle 3: apps must be technically sound

A technically sound app is reliable and technically correct. For example, HDI's app includes complex algorithms for diabetic wound analysis. These algorithms start from an image of a wound, such as the one shown in the image below, from which the algorithms determine the boundary of the wound and then conduct a process called colour segmentation, which divides the area inside the wound boundary into red, yellow and black. Red tissue is actively healing, yellow is infected tissue that is not yet ready to heal, and black tissue is problematic. By tracking the changing sizes of each colour area, the app can track wound healing progress and report that to the user.

This brief summary of the wound analysis algorithms indicates the potential for apps as health assistants, but the apps must be developed in ways that are medically and technically sound.

In addition to the medically technical aspects of an app, a technically sound app must ensure that data in the phone is copied to a secure place, with sufficient back-up, and transmitted securely. In reading reviews of apps, a surprising number of apps transfer data into the cloud without using secure transmission, which violates Health Insurance Portability and Accountability Act rules.

Beyond securing the data and its transmission, and providing sound algorithms, an app should have a sound internal architecture. In the healthcare context, apps should also be designed for integration into the broader healthcare delivery system. For example, while the goal of HDI's diabetes app is to support patients, and thus be useful for them without any interaction with their physicians, it is also designed so that patients may send their data to providers if they so desire.

Conclusion

Deploying mobile technology to improve healthcare delivery to patients that require more frequent care than physicians and a healthcare team can provide, and ultimately to improve the health of these patients, apps need to be medically sound, patient-centric and technically sound to deliver on the promise of an improved healthcare delivery system.

This research was supported in part by the National Science Foundation under Grant IIS-1065298. Any opinions, findings, conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views of the National Science Foundation.