Consumer Preferences for Telemedicine Devices and Services in South Korea

Abstract

Objective: The scope of healthcare has been expanding from caring for sick people to keeping people from becoming sick, and telemedicine will play a significant role in this new healthcare paradigm. This study investigated consumer preferences and willingness to pay for attributes of telemedicine services in South Korea. A market simulation was conducted to examine the market shares of alternative services and their relationships to the perceived usefulness of service types and preferred device types. Materials and Methods: Using a conjoint survey, we collected data on consumer preferences for six telemedicine service attributes. Data analysis used the Baysian mixed logit model. The market simulation estimated the probabilities of a specific service alternative being chosen using estimated model coefficients. Results: Wearable devices were the most preferred, followed by smart-home and smartphone devices. Consumers perceived managing blood glucose to be the most useful telemedicine service, followed by monitoring oxygen saturation and blood pressure. The market simulation indicated that consumer preferences for device types were associated with the types of chronic diseases for which management through telemedicine services is perceived to be useful. Conclusions: As the focus of healthcare moves from treating patients to keeping individuals healthy, a key factor for the successful deployment of telemedicine services is understanding consumer perceptions and attitudes. The results of this study revealed the dynamics of consumer preferences with regard to service attributes.

Introduction

Faced with healthcare costs that are growing at an unsustainable rate, an aging population, and innovations in healthcare technology, policy makers and academics have been searching for a new healthcare paradigm. One of the emerging trends in this effort is shifting the focus of healthcare from caring for sick patients to keeping people healthy. This approach is often referred to as P4 medicine, personalized medicine, or healthcare 3.0.^1
–3 The key concepts of P4 medicine are predictive, preventive, personalized, and participatory medicine. According to this approach, the scope of healthcare expands to include wellness care for healthy or high-risk individuals, and emphasis is placed on consumers rather than patients and providers. This is in contrast to traditional medicine, which is often referred to as “sick care.” Patient and consumer engagement in the care process is important in the new care paradigm.

The new healthcare paradigm could be enabled by telemedicine or telehealth technologies, which could play a significant role in implementing the prevention, personalization, and participation concepts of P4 medicine. A study by PriceWaterhouseCoopers estimated that the U.S. market for the personalized medical care portion of P4 medicine, which includes telemedicine, electronic medical records, and disease management, could reach over $100 billion by 2015.² Telemedicine or telehealth services help patients and consumers become engaged in their healthcare processes by informing them of their health status and providing them with information on their healthcare. They also enable providers to monitor patient health data and deliver healthcare services to consumers anytime and anywhere using information and communication technologies (ICT).⁴ Driving forces behind the development of various healthcare services within telemedicine are advancements in biometrics, materials, and ICT. These allow for sensors on bodies, clothing, and wearable devices to measure and monitor vital signs and then transfer information through an ICT network to remote healthcare providers such as healthcare management systems, hospitals, clinics, and pharmacies.

Telemedicine progams have been tried with various scopes and aims since the early 1990s in South Korea, which were mostly government-funded pilot programs.^5

–9 However, we have few surviving commercially successful programs so far mainly because of regulatory and payment issues. The terminology of “ubiquitous health (u-health)” has been used widely in South Korea in recent years to emphasize that the service uses a newer, better, and developed technology, although the concept is not much different from telemedicine or telehealth.

In recent years, Lin and Yang¹⁰ and Jen and Hung¹¹ have analyzed the factors affecting the acceptance of telemedicine services using the technology acceptance model. Lin and Yang¹⁰ attempted to understand and predict patient acceptance of the Asthma Care Mobile Service, which is a mobile phone-based care platform that monitors the real-time conditions of asthma patients. They found that the most critical factor affecting patient acceptance of the Asthma Care Mobile Service was user attitude, followed by perceived usefulness, subjective norms, perceived ease of use, and innovativeness. Jen and Hung¹¹ showed that attitude significantly affected the intention to adopt mobile healthcare services. They explained that perceived usefulness and perceived ease of use had indirect effects via attitude. Bradford et al.,¹² Qureshi et al.,¹³ and Ebner et al.¹⁴ also estimated willingness to pay (WTP) for a telemedicine service that replaces traditional face-to-face care. Bradford et al.¹² found that 55% of patients with chronic heart failure would be willing to pay $20 per visit to access telemedicine services. Qureshi et al.¹³ and Ebner et al.¹⁴ found that the majority of those choosing telemedicine for dermatology services were also willing to pay an average of $20–$25 per visit in out-of-pocket fees. They reported that dermatology patients preferred telemedicine to traditional care if this modality of care provided them with quicker access to their physicians.

Previous studies have investigated the attitudes of patients with specific diseases toward telemedicine services designed to manage those diseases; few studies have attempted to understand the attitudes of healthy individuals toward telemedicine services. However, we need to acquire knowledge about healthy individuals to expedite the development of a consumer-centered care paradigm that promotes the concepts of prediction, prevention, personalization, and participation. This study attempted to examine consumer preferences for telemedicine devices and services with a particular focus on preferences related to chronic care. We analyzed conjoint survey data that contained information about consumer preferences for telemedicine service attributes using a mixed logit model. We also conducted a market simulation to assess the association between consumer perceptions of the usefulness of a certain type of service and their choices of device types.

Materials and Methods

Study Design

We focused on the management of chronic diseases and selected three prevalent ones to include in this study: diabetes, hypertension, and chronic obstructive pulmonary disease (COPD).¹⁵ Diabetes and hypertension are highly prevalent around the world, with prevalence rates of 9.7% and 26.9%, respectively, in a year. The prevalence of COPD is particularly high among individuals who are older than 45 years of age (16.2%) and is increasing because of the rapidly aging population.

A telemedicine service system is composed of three subsystems: agents (or medical devices) that sense and measure vital signs, managers who are responsible for communication, and application systems that operate in medical facilities.¹⁶ We studied consumer preferences for agents and selected three types of devices: smart-home, wearable, and smartphone devices. Smart-home devices are installed in a house and provide users with two-way services for chronic care by monitoring, transmitting, and analyzing the body signals of patients. For example, users can measure and share their vital signs such as blood glucose, blood pressure, and oxygen saturation levels concurrently with care providers through the Internet. In addition, care providers can monitor remote patients through information systems connected to patients' homes. Wearable and smartphone devices can monitor, analyze, and transmit the body signals of users to care providers while users are within a personal area network. For example, wearable devices made by Vivo Metrics (Ventura, CA) (Lifeshirt™) and mobile healthcare systems with smartphone devices made by IBM (Armonk, NY) use sensors to monitor, collect, and analyze the vital signs and posture of users while they are within the personal area network. The key issues with smartphone devices are that they do not offer the flexibility required to combine different functional modules and provide limited connectivity for additional sensors or interactive devices. In contrast, wearable devices, which are highly differentiated from smartphone devices, can be tightly integrated and possibly even built into textiles in most cases.¹⁷

Finally, in the conjoint analysis, we studied six attributes that define service alternatives: device type, device price, service type, service tailoring, reply time, and service fee. The definitions of the attributes and levels are presented in Table 1.

Table 1.

Definitions of Attributes and Independent Variables in the Conjoint Analysis

ATTRIBUTE/INDEPENDENT VARIABLE, LEVEL	DESCRIPTION
Device type
Smart home type	Type of devices that can monitor and analyze biosignals with sensors attached to bathtub, toilet, door handle, etc., in a user's house
Wearable type	Type of devices that can monitor and analyze biosignals with sensors attached to clothes, watch, ring, necklace, etc.
Smartphone type	Type of devices that can monitor and analyze biosignals with sensors attached to smartphone (cell phone)
Device price (USD)
46	Price of device
92
183
Service type
Management of blood glucose level	Management of the level of blood glucose (sugar) present in blood
Management of blood pressure	Management of blood pressure exerted by circulating blood upon walls of blood vessels
Management of oxygen saturation level	Management of relative measure of the amount of oxygen that is dissolved or carried in a given medium
Service tailoring
Yes	Service is tailored to satisfy a user's specific need by providing her or him with health information and instruction in addition to the routine management of the target vital sign.
No	No tailoring; routine management of the target vital sign only
Reply time (h)
6	Time elapsed from the transmission of vital signs and inquiry to reporting of analysis results and response to the inquiry
24
72
Monthly service fee (USD)
9	Monthly fee for using service
27
46

Survey and Data

We collected data on the stated preferences of consumers through a conjoint survey. Conjoint surveys have the advantage of analyzing consumer preferences for various combinations of product and service attributes,^18,19 and they have been widely used in marketing to evaluate attributes of new products and market segments. Recently, the method has been used to evaluate attributes of environmental goods and new technologies and products in combination with the consumer preference theory and econometric methodologies.^{20

–27}

The questionnaire used in the survey consisted of five sections. We started the questionnaire with an introduction of telemedicine products and services to provide respondents with knowledge about the subject, which described their types, functions, and future prospects (Section 1), and descriptions of six attributes and their levels, which defined service alternatives (Section 2). We presented alternatives to respondents (Section 3) and recorded their stated preference of alternatives (Section 4). Lastly, we solicited information concerning respondents' socioeconomic situation and demographic status (Section 5).

The fractional factorial design of the six attributes generates a total of 486 alternatives, and we chose 25 alternatives for the survey using an orthogonal test. We divided the 25 alternatives into five subgroups and asked respondents to rank the five alternatives in each subgroup according to what they would most prefer to buy. We divided the alternatives into subgroups to help respondents pay a certain degree of attention to every alternative because respondents tend to evaluate low-ranking alternatives trivially when they are asked to rank a large number of alternatives at once. We asked them to make their decisions in consideration of their income.

We conducted face-to-face interviews with 400 adults residing in Seoul, Korea, in September and October 2011, which were carried out by Gallup Korea. We trained 10 interviewers of Gallup Korea on the conjoint survey with precise meanings of each survey items and specific guidelines on how to proceed with interviews in cases of various scenarios they might encounter. The interviewees were chosen by a purposive proportional quota sampling method to have a sample of demographic and socioeconomic characteristics that are similar to the characteristics of the population in Seoul. We used a quota sampling scheme based on age, sex, income, and education levels that was developed and used by Gallup Korea. It took about 20 min to complete an interview, and participants were offered a gift certificate worth $9 U.S. that can be used to buy cultural contents. The general characteristics of respondents are presented in Table 2.

Table 2.

General Characteristics of Study Subjects (n=400)

CHARACTERISTIC	MEAN/FREQUENCY (%)
Mean age (years)	44
Gender
Male	198 (49.5)
Female	202 (50.5)
Education
High school graduate or lower	170 (42.5)
College graduate or higher	230 (57.5)
Monthly household income
3,666 USD or less	208 (52)
More than 3,666 USD	192 (48)
Use of portable medical device
Use	41 (10.3)
Do not use	359 (89.8)
Knowledge of telemedicine
Know very well	12 (3)
Know to a certain extent	72 (18.0)
Do not know	316 (79.0)

Analysis

We used a mixed logit model, one of the discrete choice models based on random utility functions, to estimate consumer preferences and WTP for a certain level of service attributes. The mixed logit model can accommodate the heteroscedasticity of consumer preferences and can capture variations in preferences among individuals by incorporating stochastic terms into the coefficients and allowing those terms to be correlated with each other. The coefficients estimated by the mixed logit model can be interpreted as a pattern of relative preferences but do not represent any specific economic value. We estimated median marginal WTP (MWTP) and the relative importance of each level of attributes using estimates of the model coefficients (β_n ) extracted from a Bayesian process. Furthermore, we performed a market simulation to forecast the market shares of the alternatives using estimated consumer preferences and assuming that the alternatives are actually presented to consumers in a market. We derived the market shares from the probability that a specific alternative would be chosen from the 25 alternatives. Finally, we classified the respondents into three groups according to their preferences for service types and analyzed their preferences for device types to examine the relationship between preferences for service and device types. Details on the analytical methods are presented in the Appendix.

Results

Consumer Preferences and WTP

The estimated consumer preferences, MWTP, and average relative importance of telemedicine service attributes are presented in Table 3.

Table 3.

Coefficient Estimates of the Mixed Logit Model, Marginal Willingness to Pay, and Relative Importance

	COEFFICIENT ESTIMATE
VARIABLE	BASE MODEL	WITH INTERACTION	MARGINAL WILLINGNESS TO PAY (USD)	AVERAGE RELATIVE IMPORTANCE (%)
Device type
Smart home type	0.3^b	0.21^b	114.68	4.96
Wearable type	0.85^c	0.49^b	524.52	14.01
Device price	−0.05^c	−0.03^c	—	11.78
Service type
Blood glucose level	0.16^a	0.14^a	25.10	2.62
Blood pressure	−0.31^c	−0.19^a	−46.73	5.16
Tailored service	0.4^c	0.66^b	68.63	6.58
Reply time	−0.32^c	−0.21^c	2.96	14.55
Monthly service fee	−0.61^c	−0.64^c	—	40.34
Smart home type×sex		0.1^a
Wearable type×age		0.21^a
Tailored service×sex		−0.04^a
Smart home type×income		0.51^a
Wearable type×income		1.04
Tailored service×education		0.46^b
Tailored service×age		−0.62^a

p<0.1, ^b p<0.05, ^c p<0.01.

We found that of the three device types, consumers preferred wearable devices the most, followed by smart-home and smartphone devices. This result implies that consumers prefer devices that are easy to use and allow them to have continuous access to information. Consumers perceived services for the management of blood glucose levels to have the highest utility, followed by services to manage oxygen saturation and blood pressure. Although the prevalence of hypertension is higher than the prevalence of both diabetes and COPD according to Organization for Economic Co-operation and Development data from 2009,¹⁵ consumers perceived the need to continuously manage blood pressure to be lower than the need to manage blood glucose and oxygen saturation. This result implies that the magnitude of demand for telemedicine services is not strictly proportional to the prevalence of certain diseases. Consumer perceptions about the usefulness of managing certain diseases through telemedicine services play a role in determining demand.

The monthly service fee appeared to be the most important among the studied attributes, and the result indicates that consumers care more about ongoing payments (service fee) than onetime payments such as the expenditure to buy a device. In addition, we found that consumers also considered reply times and service tailoring to be important attributes. Respondents were willing to spend an additional $69 U.S. per month to receive personalized services. Moreover, respondents indicated that a 1-h reduction in reply times would be worth $3 U.S. to them. We found that consumers were willing to pay $115 U.S. to change from a smartphone device to a smart-home device and $525 U.S. to switch from a smartphone device to a wearable device. The estimates may have upward biases due to the property noted by Train and Weeks,²⁸ who observed that the distribution of WTP in preference space where WTPs are derived from coefficients tend to have a large variance rather than in WTP space, and therefore WTP estimates tend to get large.

We found significant interactions between consumer preferences and demographic and socioeconomic variables, implying heterogeneity in preference patterns among different demographic and socioeconomic groups. Male respondents exhibited higher preferences for smart-home devices than female respondents did, age was associated with higher preferences for wearable devices, and preferences for smart-home devices became stronger with higher levels of income. However, wearable devices were the most preferred in all income groups. With regard to service personalization, female respondents showed higher preferences for it than their male counterparts did, age was negatively associated with this attribute, and education was positively associated with it.

Market Simulation

The alternatives with the top five market shares are shown in Table 4. We found that the attribute shared by the top five alternatives was the monthly service fee of $9 U.S., which is the lowest level among the three levels presented to respondents. Furthermore, four out of the top five alternatives proposed the management of blood glucose levels and offered personalized telemedicine services for consumer health needs. The alternative with the largest market share (12%) was a telemedicine service that allows for the management of blood glucose levels, provides personalization and 24-h reply times, uses a wearable device priced at $92 U.S., and charges a monthly service fee of $9 U.S.

Table 4.

Five Alternatives with the Highest Market Shares

DEVICE TYPE	DEVICE PRICE (USD)	SERVICE TYPE	TAILORED SERVICE	REPLY TIME (H)	MONTHLY SERVICE FEE (USD)	MARKET SHARE (%)
Wearable	92	Blood glucose level	Yes	24	9	11.86
Wearable	46	Blood pressure	No	6	9	9.51
Smart home	46	Blood glucose level	Yes	6	9	9.21
Smart home	46	Blood glucose level	Yes	72	9	8.31
Smartphone	92	Blood glucose level	Yes	6	9	7.52

Table 5 presents the market shares of device types estimated for each subgroup of respondents, defined by the type of service they perceived to be the most useful. We found that respondent perceptions about the most useful telemedicine services influenced their choice of device type. Smart-home devices were the most preferred among respondents who indicated that the management of blood glucose through telemedicine services would be the most useful even though wearable devices were the most preferred type overall. The market shares of the wearable devices were the largest in the two other subgroups.

Table 5.

Market Share of Device Types for Groups by Preferred Service Types

PREFERRED SERVICE TYPE, DEVICE TYPE	MARKET SHARE (%)
Management of blood glucose level
Smart home	45.94
Wearable	35.04
Smartphone	19.02
Management of blood pressure
Wearable	39.61
Smart phone	30.21
Smart home	30.17
Management of oxygen saturation level
Wearable	40.32
Smart home	36.25
Smart phone	23.43

Discussion

This study analyzed consumer preferences and perceptions regarding the value of the various attributes of telemedicine services, with a specific focus on device and service types, to draw implications for service design and diffusion strategies in the era of P4 medicine. All six attributes studied significantly influenced consumer preferences and utilities, and demographic and socioeconomic characteristics significantly interacted with these preferences and utilities, indicating that service strategies should differ according to these characteristics. This implies the need for consumer segmentation, with different segments being targeted through distinct marketing strategies.

The study's findings largely confirmed previous findings and theory. The financial burden was the most important factor affecting consumer preferences for telemedicine services, particularly monthly service fees rather than onetime payments for devices,^29

–33 although consumers were willing to pay significant service fees to receive services that were tailored to their health needs. This finding implies that providers of telemedicine services should expect price competition. In addition, an effective strategy to improve market share could be to provide service subscribers with fee subsidies from employers, insurers, the government, or other stakeholders. The study results also confirmed previous findings that the ease of use and usefulness perceived by consumers affect their attitudes toward the services as derived with the technology acceptance model.^10,11 Consumers preferred services based on wearable devices most because these are perceived to be easy to use. Therefore, device manufacturers should improve their products to make them easier to use and should avoid product and service designs that complicate usage.

It should be noted that consumers have specific perceptions about the usefulness of telemedicine services designed to manage certain chronic diseases and that these perceptions affect their preferences for service attributes. It appears that consumers perceive the management of blood glucose levels or diabetes through telemedicine technology to be more useful or effective than the management of blood pressure or hypertension. Furthermore, consumers who indicated that the management of blood glucose levels through telemedicine services would be the most useful had the strongest preference for smart-home devices, whereas wearable devices were the most preferred among the group who indicated that the management of blood pressure would be the most useful. Therefore, the usefulness perceived by consumers with regard to service types should be considered when service design and diffusion strategies are formulated, and further studies must attempt to better understand the perceived usefulness of consumers with regard to various diseases.

This study has limitations and provides suggestions for future studies. First, the study did not include the influence of medical professionals, which is known to significantly influence consumer acceptance of medical services and products even though decisions to use services or products can be made by consumers themselves.³⁴ Second, the study is limited in terms of the generalizability of the findings because the subjects were drawn from residents of Seoul, which is a large metropolitan area with good access to information technology and medical services. However, 79% of respondents indicated that they did not know about telemedicine. Finally, we did not consider situations in which a device is capable of managing multiple chronic diseases. These types of devices are expected to become available through advancements in biometric technology. We need further studies to model this framework.

In summary, we attempted to obtain knowledge about the attitudes of healthy consumers with regard to the attributes of telemedicine services aimed at managing and preventing chronic diseases. This information is required to promote the concept of P4 medicine through telemedicine technology in the short term and improve quality of life and contain the growth of healthcare spending in the long term. Future studies should acquire more knowledge and improve understanding of this topic.

Footnotes

Disclosure Statement

No competing financial interests exist.

Appendix

The mixed logit model assumes that an individual n has his or her own utility function for each alternative j in a choice set t based on the random utility theory. The utility of individual n, based on alternative j in a choice set t, U_njt , can be denoted by , which consists of a deterministic part V_njt and a random part ɛ_njt ^35,36: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}U_{njt} = V_{njt} + \varepsilon_{njt} = \beta_n^{ \prime} X_{njt} + \varepsilon_{njt} , \beta_n \sim N ( b , \Sigma) \ \forall j \tag{1}\end{align*} \end{document} where x_jt denotes the vector of attributes associated with alternative j in choice set t, β_n represents the vector of the attributes' coefficients, which are distributed in a multivariate normal distribution with a mean b and a variance covariance matrix Σ, β_n ∼ N(b,Σ), and ɛ_njt is a random disturbance having an independent and identically distributed extreme value distribution.

To reflect the transformation process of the distribution of β_n from a normal distribution to a log-normal distribution and to consumer utility, we used the transformation C(β_n )=exp(β_n ). The likelihood function reflected the distributions of β_n is as in : \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}L ( d_n \mid \beta_n ) = \left( \prod_ { t = 1 } ^T \frac { e^ { C ( \beta_n ) { \prime } { { x } _ { jt } } } } { \mathop \sum \limits_ { k = 1 } ^J e^ { C ( \beta_n )\prime { { x_{ kt } } } } } \right) \tag { 2 } \end{align*} \end{document} where d_n denotes individual n's choice on the choice set of the total of T×J alternatives. In this research, we used the Bayesian estimation method by using Gibbs sampling to resolve calculation complexity. The Bayesian estimation method is known to have advantages in resolving complexity in calculation, initial point, and local optimal convergence problems.^20,35,37,38 Furthermore, it also has the advantage that its results can be interpreted from the Bayesian perspective as well as from the classical perspective.³⁹

The estimation results from Bayesian estimation method provide only a pattern of relative preferences. We further estimated the MWTP of each attribute k using the extracted coefficient sample, β_n , from Bayesian inference process to obtain economic meanings from the estimation results. Theoretically, MWTP is compensating variation in microeconomics and can be estimated by Eq. 3. In addition, we estimated relative importance of each attribute using : \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} { \rm Median } \ MWTP_k = { \rm Median } \left[ - \frac { \partial U_n / \partial x_ { nk } } { \partial U_n / \partial p_n } \right] = { \rm Median } \left[ - \frac { \beta_ { nk } } { \beta_ { n ( { \rm price } ) } } \right] \tag { 3 } \end{align*} \end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*} \begin{split}\hbox{Average relative important percentage of attribute} \\ k = {1 \over N} \mathop\sum_{n - 1}^N \left( { \hbox{part worth}_{nk} \over \sum \limits_k \hbox{part worth}_{nk}} \times 100 \right)\end{split} \tag{4}\end{align*} \end{document} where part worth_nk is the interval of attribute k's level×C(β_n ).

References

Hood

, Balling

, Auffray

. Revolutionizing medicine in the 21st century through systems approaches. Biotechnol J, 2012; 7:992–1001.

PricewaterhouseCoopers LLP. The new science of personalized medicine: Translating the promise into practice. New York: PricewaterhouseCoopers LLP, 2009.

Koh

, Choi

, Kang

, Lee

. Healthcare 3.0: The new paradigm of healthy life [in Korean]. Samsung Economic Research Institute. 2012. Available at www.seri.org/db/dbReptV.html?menu=db01&submenu=IN&pgno=1&pubkey=db20120822001 (last accessed March 8, 2013 ).

Haux

, Howe

, Marschollek

, Plischke

, Wolf

. Health-enabling technologies for pervasive health care: On services and ICT architecture paradigms. Inform Health Soc Care, 2008; 33:77–89.

Chae

, Heon Lee

, Hee Ho

, Ja Kim

, Hong Jun

, Uk Won

. Patient satisfaction with telemedicine in home health services for the elderly. Int J Med Inform, 2001; 61:167–173.

Kwon

, Cho

, Kim

, Song

, Ko

, Lee

, Kim

, Chang

, Kim

, Cha

, Lee

, Son

, Lee

, Yoon

. Establishment of blood glucose monitoring system using the internet. Diabetes Care, 2004; 27:478–483.

Joo

, Kim

. Mobile phone short message service messaging for behaviour modification in a community-based weight control programme in Korea. J Telemed Telecare, 2007; 13:416–420.

Park

, Jeong

, Jung

. Application of u-health under special situations [in Korean]. J Korean Med Assoc, 2009; 52:1164–1172.

Ryu

, Lee

, Kim

. Current state of u-health and its developmental strategies in Korea [in Korean]. J Korean Med Assoc, 2009; 52:1141–1147.

10.

Lin

, Yang

. Exploring key factors in the choice of e-health using an asthma care mobile service model. Telemed J E Health, 2009; 15:884–890.

11.

Jen

, Hung

. An empirical study of adopting mobile healthcare service: The family's perspective on the healthcare needs of their elderly members. Telemed J E Health, 2010; 16:41–48.

12.

Bradford

, Kleit

, Krousel-Wood

, Re

. Willingness to pay for telemedicine assessed by the double-bounded dichotomous choice method. J Telemed Telecare, 2004; 10:325–330.

13.

Qureshi

, Brandling-Bennett

, Wittenberg

, Chen

, Sober

, Kvedar

. Willingness-to-pay stated preferences for telemedicine versus in-person visits in patients with a history of psoriasis or melanoma. Telemed J E Health, 2006; 12:639–643.

14.

Ebner

, Wurm

EMT

, Binder

, et al. Mobile teledermatology: A feasibility study of 58 subjects using mobile phones. J Telemed Telecare, 2008; 14:2–7.

15.

Organization for Economic Co-operation and Development. Health at a glance 2009: OECD indicators. Paris: OECD Publishing, 2009.

16.

Korea Food and Drug Administration. Guideline for standardization of home health care devices. Seoul: Korea Food and Drug Administration, 2007.

17.

Amft

, Lukowicz

. From backpacks to smartphones: Past, present, and future of wearable computers. Pervas Comput IEEE, 2009; 8(3):8–13.

18.

Green

, Srinivasan

. Conjoint analysis in consumer research: Issues and outlook. J Consum Res, 1978; 5:103–123.

19.

Green

, Srinivasan

. Conjoint analysis in marketing: New developments with implications for research and practice. J Mark, 1990; 54:3–19.

20.

Huber

, Train

. On the similarity of classical and Bayesian estimates of individual mean partworths. Mark Lett, 2001; 12:259–269.

21.

Hensher

. The valuation of commuter travel time savings for car drivers: Evaluating alternative model specifications. Transportation, 2001; 28:101–118.

22.

Layton

. Random coefficient models for stated preference surveys. J Environ Econ Manag, 2000; 40:21–36.

23.

Batt

, Katz

. A conjoint model of enhanced voice mail services. Implications for new service development and forecasting. Telecommunications Policy, 1997; 21:743–760.

24.

Kim

, Lee

, Koh

. Effects of consumer preferences on the convergence of mobile telecommunications devices. Appl Econ, 2005; 37:817–826.

25.

Byun

, Lee

, Ahn

. Korean wireless data communication markets and consumer technology. In: Cooper

, Madden

, eds. Frontiers of Broadband, Electronic and Mobile Commerce. Heidelberg: Physica-Verlag, 2004:149–164.

26.

Kim

, Park

, Lee

. Using stated-preference data to measure the inconvenience cost of spam among Korean e-mail users. Appl Econ Lett, 2006; 13:795–800.

27.

Ahn

, Lee

, Kim

. An analysis of consumer preferences among wireless LAN and mobile internet services. ETRI J, 2006; 28:205–215.

28.

Train

, Weeks

. Discrete choice models in preference space and willingness-to-pay space. In: Alberini

, Scarpa

, eds. Applications of simulation methods in environmental and resource economics. Dordrecht, The Netherlands: Springer, 2005:1–16.

29.

Park

, Chon

, Lee

, Choi

IeJ

, Yoon

. Service design attributes affecting diabetic patient preferences of telemedicine in South Korea. Telemed J E Health, 2011; 17:442–451.

30.

Bryce

, Zickmund

, Hess

, McTigue

, Olshansky

, Fitzgerald

, Fischer

. Value versus user fees: Perspectives of patients before and after using a web-base portal for management of diabetes. Telemed J E Health, 2008; 14:1035–1043.

31.

Bradford

, Kleit

, Kousel-Wood

, Richard

. Willingness to pay for telemedicine assessed by the double-bounded dichotomous choice method. J Telemed Telecare, 2004; 10:325–330.

32.

Cocosila

, Archer

, Yuan

. Would people pay for text messaging health reminders?. Telemed J E Health, 2008; 14:1091–1095.

33.

Whitten

, Buis

. Private payer reimbursement for telemedicine services in the United States. Telemed J E Health, 2007; 13:15–23.

34.

Simon

, Kotler

. Building global biobrands: Taking biotechnology to market. New York: Free Press, 2003:99–100.

35.

McFadden

. Conditional logit analysis of qualitative choice behavior. In: Zarembka

, Frontiers of econometrics. New York: Academic Press, 1973:105–142.

36.

Train

. Discrete choice method with simulation. Cambridge, UK: Cambridge University Press, 2003.

37.

Edwards

, Allenby

. Multivariate analysis of multiple response data. J Mark Res, 2003; 3:321–334.

38.

Allenby

, Rossi

. Marketing models of consumer heterogeneity. J Econometrics, 1998; 89:57–78.

39.

Train

, Sonnier

. Mixed logit with bounded distributions of correlated partworths. Applications of Simulation Methods in Environmental and Resource Economics. Dordrecht, The Netherlands: Springer, 2005:117–134.