Intelligent Disease Self-Management with Mobile ... - IEEE Xplore

4 downloads 0 Views 2MB Size Report
It has almost limitless potential to inform and engage the patient in treatment decisions, to monitor the patient's condition, and to alert caregivers about any.
COVER COVER FEATURE FEATURE THEME COMPUTING HERE IN HEALTHCARE

Intelligent Disease Self-Management with Mobile Technology Marina Velikova, Embedded Systems Innovation by TNO Peter J.F. Lucas and Maarten van der Heijden, Radboud University, Nijmegen

Cost-effective mobile healthcare must consider not only technological performance but also the division of responsibilities between the patient and care provider, the context of the patient’s condition, and ways to implement patient decision support and tailored interaction.

M

obile technology is a promising way to transfer aspects of clinical care support from the caregiver—physician, nurse, nurse practitioner, or physical therapist— to the patient, thus enabling disease self-management. It has almost limitless potential to inform and engage the patient in treatment decisions, to monitor the patient’s condition, and to alert caregivers about any unexpected changes.1 However, continuous and active patient involvement in mobile healthcare (m-health) requires considerable

Selected CS articles and columns are also available for free at http://ComputingNow.computer.org.

32

r2vel.indd 32

CO M PUTE R P U B LISH ED BY TH E I EEE COMP UTER SOCI E T Y

insight into disease self-management as a process, which requires accommodating a shift in the patient’s role. As Figure 1 shows, in traditional care, the patient is largely passive, but in an m-health system, the patient is actively engaged in decision making about treatment. Involvement to this degree requires sufficient knowledge about disease-related conditions and problem-­ solving skills that lead to behavioral changes and coping strategies. Active day-to-day self-management means regular monitoring and reporting of signs and symptoms, enhanced medication adherence, and appropriate responses to health changes. Inherent in all these requirements is a close patient–­caregiver partnership. Most research to date emphasizes the technological aspects of m-health and other forms of patient-­centric healthcare2,3 or focuses on specific m-health applications.4 0018-9162/15/$31.00 © 2015 IEEE

1/22/15 5:05 PM

In contrast, we built two m-health ­systems­—one for patients with chronic obstructive pulmonary disease (COPD)5 and one for patients with pregnancy complications—­ primarily to explore how mobile technology can support the patient. Our focus was on embedding disease-­specific Bayesian network models within a smartphone to enable on-the-fly patient data interpretation. In this way, the system could assess the patient’s health status and advise the patient on what action to take— all without the care provider’s direct involvement. Experiments showed positive results for both m-health models. The COPD model, which we developed from clinical data in close cooperation with clinical experts, correctly detected 91 percent of COPD exacerbations from only patient-measured biosignals and reported symptoms. Tests of the pregnancy-related model, which we developed in cooperation with gynecologists,6 on actual pregnancy data reliably predicted hypertensive complications for 60 percent of pregnant patients at least four weeks before they received an actual diagnosis. As part of our work, we identified four foundational aspects of m-health for disease self-management—­support for shared care, context awareness, embedded intelligence, and personalized interaction—and explored how best to integrate these in mobile technology. A major challenge in disease self-management is determining the optimal complexity level for representing and interpreting context-­ specific clinical data to support patient-­tailored interaction, decision making, and responses. We believe that an integrated approach like ours is the best way to evolve cost-effective m-health systems that can augment or even

r2vel.indd 33

Traditional self-management

Disease self-management Report risk factors

Contact a professional

• Age • Gender • Weight • Smoking • Diseases

Patient

Adhere to medication

Follow/adjust activity plan • Exercising • Resting • Sleeping

Report symptoms • Headache • Pain • Coughing • Fatigue • Vomiting Monitor vital signs • Blood pressure • Glucose level • Temperature • SpO2 • Pulse

FIGURE 1. Traditional disease management versus disease self-management. In the traditional model (green), the patient passively follows a treatment plan, but in a mobile healthcare (m-health) system that supports disease self-management (yellow), the patient is actively involved in monitoring and decision making, and the caregiver receives continuous updates on the patient’s condition. Communication is tailored to patient preferences, whether through voice, text, video, or email.

replace traditional disease management and thus lead to greater numbers of individuals who enjoy improved health at a lower cost. We also believe an integrated approach will stimulate the scientific, technological, and business development of mobile, patient-oriented decision-­support systems.

SUPPORT FOR SHARED CARE

The success of intelligent disease self-management in m-health relies on how effectively the patient and caregiver share healthcare responsibilities. Systems with embedded intelligence must both automate patient data interpretation and feedback and support patient–provider interaction and decision making. Data delivery speed is a critical shared-care service, since real-time data transmission is essential to monitoring the patient’s condition. In a personalized m-health system, the patient could decide whether or not to transmit data and, if so, how much

data and how often. If regularly transmitted data ceases unexpectedly, the caregiver might take steps to check if this was intentional and investigate the need for intervention. Another service to support shared care is direct patient–caregiver communication through voice, video, text, or email, enabling a timely discussion of potential problems and possible treatment-­ plan adjustments. Patients would have the initiative to contact the caregiver, but the system would also advise the patient about the necessity of contact, thus lowering interaction degree and cost. Tailoring interaction in this manner will encourage both parties to partner in managing the disease. An m-health system that supports shared care is likely to motivate patients to self-manage their disease and possibly achieve a better overall outcome than is possible with traditional treatment. Easier access to personal clinical information, for example, can increase the patient’s desire to FEBRUARY 2015

33

1/22/15 5:41 PM

COMPUTING IN HEALTHCARE

explore evidence-based guidelines and relevant scientific studies to gain more knowledge about disease treatment.

CONTEXT AWARENESS

Because personalization is a strong benefit of m-health systems, mobile devices must be able to recognize, correctly interpret, and adapt to changes in patient, disease, and environmental contexts.

›› symptoms, subjective experi-

ences that the patient reports— for example, coughing, fatigue, and headache; ›› signs, biomedical data based on observations from tests and measurement devices—for example, blood pressure via a sphygmomanometer, and glucose or hemoglobin levels from biochemical tests; and

TO SUPPORT DISEASE SELFMANAGEMENT, MOBILE DEVICES MUST EMBED DECISION AIDS WITH ENOUGH INFORMATION TO MAKE INTELLIGENT HEALTH DECISIONS. Patient context

The patient’s age, gender, and personal and family disease history, as well as lifestyle choices about diet, alcohol consumption, smoking, and activities, can affect the risk of contracting a disease or the progression of disease that the patient already has. A mobile device is a convenient way to gather such patient-specific information—for example, by having the patient fill out a questionnaire to gain information beyond what health records contain. Personal and interpersonal factors such as prior health-related behavior, socioeconomic status, and social attitudes and support also influence the likelihood of health-promoting behavior and self-care. These factors determine the patient’s physiological and psychological status, which in turn determine both current and future health. Health status is based on three data types: 34

r2vel.indd 34

COMPUTER 

›› biosignals, repeatedly measured

biological signals collected using noninvasive technologies—for example, signals from electrocardiography, which reflect physiological processes.

Symptom reports are easy to collect through questionnaires on a mobile device, and measuring devices, such as body sensors, can collect signs and bio­ signals and wirelessly transfer them to a computing device. These measuring devices are compact, mobile, cheap, and easy to use, and measurements can be more representative than those obtained in a hospital or clinical environment. Remote measurements eliminate the white-coat effect—the patient’s altered state from being in a clinic or hospital—as well as decrease costs, since the patient makes fewer visits to the hospital. On the down side, ensuring that remote measurements are reliable

might be more difficult, since the patient might not follow procedures or have the required skills. For example, an elderly patient with impaired renal function could find it difficult to quickly process the color analysis of a urine strip test, which could lead to the incorrect reporting of that measurement.

Disease context

Whether or not a mobile platform can manage a disease depends on the personal, societal, and economic burden that the disease imposes. Disease management can be short or long, depending on the disease type and severity. Long-term disease management. Studies have shown that 75 to 85 percent of healthcare costs go to the management of chronic diseases, such as COPD, hypertension, and diabetes mellitus types 1 and 2.7 Many chronic diseases are well understood and also preventable, since they are closely linked to the patient’s lifestyle choices. For example, about 90 percent of COPD cases stem from the patient’s smoking history. Treatment obviously begins with smoking cessation, which makes COPD an excellent candidate for judging the cost-effectiveness of m-health systems in dealing with long-term disease management linked to lifestyle choices. Short-term disease management. Some diseases do not require indefinite management; for example, a few weeks of guided therapy are sufficient to treat many sports injuries. Pregnancy-­related disorders, such as gestational hypertension and pre-­eclampsia, require management only during pregnancy. Disease severity. Disease severity is a concern primarily in long-term management. In COPD, for example, W W W.CO M P U T E R .O R G /CO M P U T E R

1/22/15 5:05 PM

Age < 65 y FEV1 < 35% risk-E yes

Time t

≥ 65 y Activ

Dysp ≥ 35%

1-2

Hosp

risk-E no