Utilizing telemedicine in oncology settings: Patient favourability rates and perceptions of use analysis using Chi-Square and neural networks

Abstract

BACKGROUND:

Telemedicine is an alternative to traditional face-to-face doctor-patient office visits. Although telemedicine is becoming more prevalent, few studies have looked at the perceived favorability rate among patients utilizing telemedicine over the traditional office visit to a provider’s office considering data samples from more than 5 clinics in northern Louisiana.

OBJECTIVE:

This study aims to measure patient favorability of using telemedicine to receive care. This study looks at the perceived positive and negative favorability rates of patients in the oncology settings. The researchers analyzed how age, income level, and education level influenced the perceived patient favorability rates and their willingness to utilize telemedicine.

METHODS:

The investigators used Chi-Square analysis to identify favorability with respect to age education and income levels. In addition to this Artificial Neural Networks were used to identify the threshold for favorability with respect to age, income, and education.

RESULTS:

Chi-Square tests of association showed that of the variables analyzed, only education level had a statistically significant relationship with a patient’s favorability rate of telemedicine utilization. While our neural network analysis indicated that the threshold for income, age, and education are $34,999, 66 years, and a college degree.

CONCLUSION:

In this article the investigators have successfully demonstrated the use of Artificial Neural Networks in identifying favorability of telemedicine used in addition to the traditional statistical methods such as Chi-Square. Thereby, creating a path for future research using advanced computational techniques like Artificial Neural Networks in analyzing human behavior.

Keywords

Chi-Square analysis telemedicine teleoncology neural networks

1. Introduction

The idea of receiving medical care remotely, known as telemedicine, is no longer an unfamiliar practice in the health care system. With the continuous need to harness the power of health information technology, doctors and patients alike are benefitting from the increased access to health care services made possible by telemedicine. The adoption and use of telemedicine has increased dramatically since the 1990s [1]. A ubiquitous definition of telemedicine requires a geographic separation between the patient and the clinician providing the medical care [2]. Here we would like to inform our readers that this article illustrates an extension of the research described in Gurupur et al. [3]. The instrument and data used here is the same; however, the investigators have demonstrated a different approach in attempting to analyze the acceptability of telemedicine use. The instrument used has been successfully tested for internal validity using Cronbach’s Alpha as described in [3].

Telemedicine is used in various capacities from teleconsultation and telediagnosis to telemonitoring services, which these are also the most popular uses of telemedicine [3]. Further, telemedicine has been instrumental in various medical specialty including cardiology, dentistry, dermatology, hematology, emergency care, endocrinology, home care, medical education, pediatrics, psychiatry, ophthalmology, radiology, surgery, urology and oncology. There are also various modes of interaction in using telemedicine technology. Telemedicine can occur in real time with synchronous one-way or two-way interaction motion video, transmission of still images or video clips with voice interaction or text interaction or asynchronous interaction in which images, or videos are stored and viewed at a later time [1]. Telemedicine utilizes both health information technology and communication mediums to deliver health care services, health information and health education [4].

2. Literature review

The focus of this study will be on telemedicine in the field of oncology. Telemedicine is utilized in oncology as a means to increase access to appropriate care [5]. To date, much of the literature has focused on models for deploying teleoncology, specifically as it relates to rural patients who would otherwise have to travel extensively for treatment [5, 6]. This travel costs the patient not only in gasoline and accommodation charges, but also in time lost due to excessive commuting.

Age has been an important factor when it comes to acceptability of telemedicine. This has been ascertained by Cimperman et al. [7]. In this study the investigators analyzed the significant factors responsible for acceptance of telehealth for older adults. The denouement of which provided seven predictors: effort expectancy, perceived usefulness, social influence, computer anxiety, perceived security, physicians’ opinions, and facilitating conditions. Telemedicine implementation is extremely important for older adults given the presence of chronic diseases and the rise in occurrence of dementia and diabetes. Owing to this aforementioned reason successful implementation of telemedicine is pivotal for providing healthcare to older adults.

Wootton and Bonnardot [9] analyzed the use of telemedicine usage in countries with lower income levels. The scenario of such telemedicine usage was described as a “low-resource setting.” Interestingly this study also analyzed education level which is another important factor applicable to the use of telemedicine considered in this article. To increase the acceptability of telemedicine usage the investigators pointed out that the education programs were pivotal in improving acceptance of telemedicine. Additionally, George, Hamilton, and Baker [10] analyzed the use of telemedicine among low level African Americans and Latinos. Some of the perceived factors included: confidentiality, privacy, and physical absence. In this study the investigators clearly pointed out that socio-economic conditions can be pivotal in the acceptance of telemedicine among its users. Based on the aforementioned arguments we can conclude that age, income, and education levels can be important factors in the acceptance of telemedicine.

In one recent study, Canada deployed teleoncology modalities as a feasible option for delivering care to hundreds of small, rural communities spread throughout its northern regions. This project began with simple videoconferencing used during appointments and later transformed into teleoncology consultations, follow-ups, and diagnostic evaluations [4, 6]. As of late, imaging and chemotherapy are being made available at local, regional centers, allowing patients to receive care at centers much closer to their home. Such realities can save patients significant money through decreasing travel costs [4, 6]. A similar study was conducted in rural Iowa at Visiting Consultant Clinics (VCCs) [6]. VCCs scheduled specialists and oncologists to visit outreach sites to assist in providing diagnostic or treatment services to cancer patients [6]. Nearly half of all Iowa-based oncologists practiced in the VCC rural community, increasing access to necessary care for cancer patients [8].

Other studies have assessed the perceived satisfaction rates of patient or provider telemedicine usage. Most literature suggests that physicians and patients alike report satisfaction in using teleoncology [7, 8, 9]. In the state of Kansas, a study found that all patients receiving care via telemedicine interactive video in a specialty oncology/hematology clinic reported high satisfaction with their care. Participants felt their access to specialty services was improved through telemedicine [10].

Although much of the literature on telemedicine, specifically teleoncology, addresses appropriate models of teleoncology and satisfaction rates with these models, few studies have addressed patient favorability rates, or intention of use. This study will address patient positive and negative favorability rates related to teleoncology utilization.

Dewar et al. [11] analyzed the motivation to engage with telehealth technology using t-tests by including the parameter of device use in their analysis. The idea of device use correlates to the notion of usability when it comes to telehealth and mHealth applications as described by Gurupur and Wan [12]. It has been commonly noticed that improvement in usability can be one of the deciding factors of telemedicine usage especially for especially for individuals above 50 years and older. With that in mind the investigators have taken into consideration the use of Theory of Planned Behaviour to this study. The application of this framework to analyze the specific research questions implies the fact that the analysis carried out here is different from the previous experiment [13] on the same dataset.

Gagnon, et al., [14] illustrated the use of the use of interpersonal behavior to analyze the adoption of telemedicine among physicians. In this particular study conducted in the state of Quebec, Canada the investigators identified the fact that physicians who had a perception of professional and social responsibilities towards the adoption of telemedicine had a stronger positive intention towards its adoption. Additionally, the investigators also found self-identity to be have a negative effect towards the intention of this adoption. These benchmark findings have led the investigators of the study delineated in this article to adopt Theory of Planned Behavior; although, in this study the investigators are not involved in analyzing the physicians.

Another interesting aspect of this study is that Artificial Neural Networks have been rarely used in analyzing acceptability of telemedicine and for general behavioral science studies. This article presents an interesting analysis of the usage of Artificial Neural Networks in analyzing the acceptability of telemedicine usage.

3. Methods and materials

3.1 Theoretical frameworks

As a guiding framework for this study, the authors have used the Theory of Planned Behavior [13] as well as the FITT framework. The key use of the Theory of Planned Behavior is that it predicts and explains behavior in specific contexts, or in this case, intention towards telemedicine favorability. Individual attitudes and norms will influence one’s intention and thus, ultimately, influence one’s behavior. In essence, individuals with positive attitudes toward telemedicine usage will likely have positive intention towards utilization. This intention then leads to the behavior: actually engaging in usage of telemedicine. Figure 1 displays the sequential components of the theory as they work together to influence one’s behavior.

Figure 1.

Integration of Theory of Planned Behavior and FITT framework.

On the other hand, the FITT framework – or the fit between individual, task, and technology, is commonly applied to health information evaluation [15, 16]. The framework asserts that the adoption of health-related information technology tools is dependent upon the fit between attributes of the users (e.g. motivation), the attributes of technology (e.g usability and functionality), and the attributes of the “tasks” or clinical processes (e.g. task complexity) [15]. Figure 2 displays the three domains in the FITT framework, as adapted by [15]. This study will apply the FITT framework and analyze the results based on each model domain: individual, task and technology.

3.2 Research questions

This study looks at the favorability rates (positive and negative favorability) of patients using telemedicine in the treatment for cancer, specifically patients from both rural and urban regions of Northern Louisiana. Favorability rate is defined as a patient’s attitude towards the usage of telemedicine. Patients’ attitudes towards telemedicine can range from favorable to not favorable. As mentioned before, prior studies have focused on the prevalence and usage of telemedicine in oncology, types of telemedicine and barriers in the use of teleoncology [7, 8, 9, 10, 17]. Few studies have studied patients’ favorability in the acceptance and likelihood of using telemedicine as an alternative to the traditional face-to-face patient-doctor office visit for oncology care. This study looks to fill this gap in the literature on telemedicine and patients’ favorability for teleoncology and will focus on how patients’ gender, age, income level, and education level impact patients’ perceived favorability in using teleoncology. This exploratory study aims to answer the following research questions:

Table 1
Sample demographic characteristics [16]

Characteristics	Frequency	Percentage of sample ( $n=$ 147)
Gender
Male	71	49
Female	75	51
Age group
Under 40	11	8
40–65	66	45
66 and over	69	47
Household income
Low income (under $35,000)	75	51
Medium income ($35,000–$74,999)	38	26
High income ($75,000 or higher)	19	13
Educational attainment
Less than or some high school	25	17
High school or vocational graduate	65	44
Some college	24	16
Completed bachelor’s degree or higher	24	16

Note. Since not all participants elected to answer all demographic questions few variables total less.

Figure 2.

Algorithm used for artificial neural network analysis.

RQ1. How much influence does age have on the favorability rate of patients using teleoncology for their cancer care?

RQ2. How much influence does income level have on the favorability rate of patients using teleoncology for their cancer care?

RQ3. How much influence does education level have on the favorability rate of patients using teleoncology for their cancer care?

Figure 3.

Structure of the Artificial Neural Network used for experimentation.

In addition to the above data specific questions the investigators have also considered analyzing the aforementioned research questions using a neural network for the purpose of analyzing the results between Chi-Square analysis and Artificial Neural Networks.

3.3 Experimental design

This is an exploratory, cross-sectional study designed to look at the favorability rates of patients using teleoncology [19, 20]. The survey instrument used for this study was adapted from Gurupur et al. [3]. Using a 5-point Likert scale, patients were asked to rate their perception of their favorability in using telemedicine.

There were a total of 147 responses from a sample of cancer patients from oncology clinics located in both rural and urban Louisiana, nearby the city of Shreveport, Louisiana. The convenience sample of patients was selected from within clinics in the region who had not used telemedicine yet, and volunteered to participate in the survey. The demographics show that the survey sample is a good representative sample of the Louisiana population (Table 1) as compared to Louisiana demographics as per US Census data (http://quickfacts.census.gov/qfd/states/22000.html).

3.4 Applied measurement

Our study looked at the favorability rate of patients using telemedicine and whether gender, age, income level and education level influenced the favorability rate of patients using telemedicine, specifically for patients suffering from cancer. Appendix A provides the survey instrument used in the study. If a patient answered “agree” or “strongly agree” on the positive favorability rate questions and “strongly disagree” and “disagree” on the negative favorability rate questions, this indicated that the patients were favorable towards using telemedicine.

Independent variables were recoded to represent dichotomous or trichotomous categories. Gender was already dichotomized. Age was recoded into 3 categories: under 40 years of age; 40–65 years of age; and over 66 years of age. Education level was recoded into 4 categories: less than or some high school; high school or vocational school graduate; some college; and a completed degree or higher. Lastly, income was recoded into 3 categories based on low, medium, and high annual household incomes: low income represented under $35,000; medium represented $35,000–$74,999; and high represented $75,000 or higher.

The respondent’s favorability rate toward telehealth utilization was the dependent variable for this study. To determine this rate, each respondent’s overall survey score was summed, using the formula shown below.

Respondent favorability

$\displaystyle\text{Rate}=\Sigma(P_{1}+P_{2}+P_{3}+P_{4}+P_{5}+P_{6}+P_{7}+P_{8% }+P_{9})$ (1)

$P_{i}$ (1) is defined as the favorability rate, per survey question. The survey uses a Likert scale for questions 1–9. The Likert scale ranges from strongly disagree, disagree, neither agree nor disagree, agree, to strongly agree. The Likert scale is coded where strongly agree $=$ 5 to strongly disagree $=$ 1. The higher the score, the more favorable a patient feels towards the utilization of telemedicine.

The respondent’s Likert scale responses to each survey question was computed (i.e. a strongly agree response was worth 5 points whereas a strongly disagree response was worth 1 point). A total score of 45 was the highest possible survey score (i.e. 9 survey questions $\times$ 5 “strongly agree” responses). Once respondent scores were summed, each score was categorized into one of three categories for the dependent variable – Favourability rate for telehealth utilization. The following were the categories: negative favor rate (0–18 Likert scale score total), average favor rate (19–27 Likert scale score total), and high favor rate (28–45 Likert scale score total).

From the survey, the positive favorability rate indicators include perceived benefits, which includes concepts such as the savings in travel cost and reduced wait times, perceived motivation which includes improved clinical feedback and quicker response time, perceived compatibility and advantage which includes increased accessibility and improved access to more specialists. The negative favorability rate indicators include perceived anxiety, which includes lack of comprehension and hindrance in communication, and perceived complexity, which includes the complexity in the telemedicine technology. Table 2 highlights positive and negative favor rate factors.

Table 2

Definition of favourability rate

Positive favour rate factors
Perceived benefits	Savings in travel cost
	Reduced wait times
Perceived motivation	Improved clinical feedback
	Quicker response time
Perceived compatibility and advantage	Increased accessibility
	Improved access to more specialists
Negative favour rate factors
Perceived anxiety	Lack of comprehension
	Hindrance in communication
Perceived complexity	Complexity in the telemedicine technology

The survey instrument had a Cronbach’s alpha coefficient of 0.84 (alpha $>$ 0.70) for the six positive readiness factor items, which suggests high internal consistency. However, the Cronbach’s alpha coefficient for the three negative readiness factors was 0.476 (alpha $<$ 0.70), which suggests that the items have low internal consistency. This limitation of the study may be a result of poorly worded survey questions on the negative readiness factors.

The responses were coded (and reverse scale coding for negatively worded questions) prior to statistical analysis testing. Both descriptive and inferential statistical analyses were performed using IBM SPSS 22. The research questions were analyzed based on models developed to measure patient telemedicine utilization favorability. We performed univariate analyses on each of the response variables to obtain descriptive statistics, central tendency and the pattern of responses.

Chi-Square test of associations was conducted to determine if there was an association between the independent variables (gender, age, income level, or education level) and the dependent variable (favorability rate for telehealth utilization). For this statistical analysis, statistical assumptions were met. These assumptions included variables that were categorical in nature with each category representing two or more independent groups.

3.5 Method to apply neural networks

The dataset consists 147 responses of which 117 were used for training the neural network and remaining 30 were used for testing since neural networks require a training data set. Here the investigators have used 80% of the dataset (93 participants) of the participants are used for training the Artificial Neural Network and the remaining 20% (24 participants) were used for validation. Four independent variables (answers to question Q1 through Q4) were used for analysis. The dependent variable used here is S4 (Counting the income of everyone in your household, which category best represents your annual household income?). A new output variable is defined such that if the value of S4 is greater than 4 (which indicates the annual household income is greater than $34,999) then it is assigned value “1” else if S4 is less than or equal to 4 (which indicates the annual household income is less than or equal to $34,999) it assigned value “0”.

This Artificial Neural Network takes the aforementioned 4 independent variables as inputs. These inputs are given to the hidden layer consisting of 3 nodes as illustrated in Fig. 2. Rectified linear unit is used as the activation function at the input and hidden layer. The output of the hidden layer is given to the output node which is the dependent variable. Sigmoid activation function is used at the output layer. The Artificial Neural Network is compiled using adman optimizer involving the calculation of cross binary entropy and 100 epochs. The accuracy is predicted by considering the threshold value as 50%.

4. Results

4.1 Chi-Square analysis

Unlike the previous experiment delineated in Gurupur et al. [3], the researchers used Chi-Square analysis for analyzing the acceptability of telemedicine usage. The Chi-Square test of association between age and favorability rate, gender and favorability rate, and income level and favorability rate all showed no statistical significance. This indicates that these variables had no bearing- either positive or negative- on one’s attitude toward using telemedicine in oncology care. However, education level and favorability rate proved to be statistically significant at 0.020, thus indicating an association between the two variables. Table 3 above indicates the statistical findings for testing the association between education level and favorability. As it relates to education, it is worth noting that of those who perceived telemedicine use with disfavor, seventy percent had only some high school or a high school diploma completed. As individuals pursued and completed high levels of education, their favorability rate towards telemedicine usage increased, to either average or high favorability. Table 4, above, illustrates the cross tabulation of respondent’s education level and perceived favorability rates: negative, average, or high. Of individuals with a negative favorability ( $n=$ 10 respondents), 5 had completed some high school, 2 had a high school or vocational degree, 1 had completed some college, and 2 had a college degree or higher.

Table 3
Test of association between education and favorability

	Value	df	$p$ -value
Pearson’s Chi-Square	15.023 ${}^{\rm a}$	6	0.020
N of valid cases	138

Table 4

Cross tabulation of education level and telemedicine usage favorability

			Education
			Some high	High school or vocational	Some	College graduate	Total
			school	school graduate	college	or higher
Favorability	Negative	Count	5	2	1	2	10
	favorability	% within favorability	50.5%	20.0%	10.0%	20.0%	100%
	Average	Count	3	23	3	5	34
	favorability	% within favorability	8.8%	67.6%	8.8%	14.7%	100.0%
	High	Count	17	40	20	17	94
	favorability	% within favorability	18.1%	42.6%	21.3%	18.1%	100.0%
Total		Count	25	65	24	24	138
		% within favorability	18.1%	47.1%	17.4%	17.4%	100.0%

Overall, most survey respondents had average or high favorability perception toward telemedicine usage. Sixty-eight percent of survey respondents had a high favorability rate towards telemedicine ( $n=$ 94/138). Twenty five percent of survey respondents had an average favorability rate toward telemedicine usage ( $n=$ 34/138). And, only ten survey respondents, or 7%, had a negative favorability rate towards telemedicine usage ( $n=$ 10/138). It is important to note that of the 147 total survey respondents, only 138 answered education related questions.

Figure 4.

Confusion matrix for all the three hypothesis.

4.2 Neural network analysis

While the research question 4 was sufficiently answered by the Chi-Square analysis it was not able to identify a significant correlation for questions RQ1, RQ2, and RQ3. Therefore, we attempt to apply Artificial Neural Networks to extract some answer to 3 by identifying relationships between annual household income and favorability of telemedicine use. The hypothesis and their obtained results are as follows (Fig. 4):

Hypothesis 1: If the value of annual household income is greater than $34,999 then the output variable is assigned value “1” else if annual household income is less than or equal to $34,999 it assigned value “0”.

The details of the confusion matrix is as follows: a) True Negative (TN) value $=$ 6 which indicates the values of the predicted output that is correctly considered as 0 as per the hypothesis (Result $=$ 0 when annual household income is less than or equal to $34,999), b) True Positive (TP) value $=$ 8 which indicates the values of the predicted output that is correctly considered as 1 as per the hypothesis (Result $=$ 1 when annual household income is greater than $34,999), c) False Negative (FN) value $=$ 3 which indicates the values of the predicted output that is wrongly considered as 0 as per the hypothesis (Result $=$ 0 when annual household income is less than or equal to $34,999), and d) False Positive (FP) value $=$ 8 which indicates the values of the predicted output that is wrongly considered as 0 as per the hypothesis (Result $=$ 1 when annual household income is greater than $34,999). This clearly establishes the fact that favorability of telemedicine use is high when the annual household income is equal to or greater than $34,999.

Hypothesis 2: If the education level is greater than 3 the output variable is assigned value “1” else if it is less than or equal to 3 then it is assigned “0”, 3 indicates high school education.

The details of the confusion matrix is as follows: a) True Negative (TN) value $=$ 9 which indicates the values of the predicted output that is correctly considered as 0 as per the hypothesis (Result $=$ 0 indicates high school education), b) True Positive (TP) value $=$ 9 which indicates the values of the predicted output that is correctly considered as 1 as per the hypothesis (Result $=$ 1 indicates higher than high school education), c) False Negative (FN) value $=$ 3 which indicates the values of the predicted output that is wrongly considered as 0 as per the hypothesis (Result $=$ 0 indicates high school education), and d) False Positive (FP) value $=$ 3 which indicates the values of the predicted output that is wrongly considered as 1 as per the hypothesis (Result $=$ 1 indicates higher than high school education).

Hypothesis 3: If age is greater than or equal to 66 the output variable is assigned value “1” else if the age is less than 66 it is assigned “0”.

The details of the confusion matrix is as follows: a) True Negative (TN) value $=$ 5 which indicates the values of the predicted output that is correctly considered as 0 as per the hypothesis (Result $=$ 0 when the age is less 66), b) True Positive (TP) value $=$ 2 which indicates the values of the predicted output that is correctly considered as 1 as per the hypothesis (Result $=$ 1 when the age is greater than or equal to 66), c) False Negative (FN) value $=$ 7 which indicates the values of the predicted output that is wrongly considered as 0 as per the hypothesis (Result $=$ 0 when the age is less 66), and d) False Positive (FP) value $=$ 10 which indicates the values of the predicted output that is wrongly considered as 1 as per the hypothesis (Result $=$ 1 when the age is greater than or equal to 66).

5. Discussion

Our study assessed telemedicine favorability rates for patients receiving oncology care. As the field of teleoncology continues to grow [21, 22, 23], it is important to understand patients’ perceptions on utilization and assess various favorability factors towards this usage.

Revisiting the study’s research questions, it was determined that variables such as gender, age, and income level are not associated with one’s perception towards telemedicine usage. Which provides a negative result for research questions RQ1, RQ2, and RQ3. Based on the Chi – Square analysis described so far education was the statistically significant factor that lead to a positive perception towards the use of telemedicine.

Education level has a statistically significant relationship towards association with telemedicine favorability. In assessing those with negative favorability views towards telemedicine usage, the majority of respondents who were unlikely to use the technology did not have a high school diploma. As such, it is important to consider telemedicine systems that are easy to use and understand by patients of all educational backgrounds. It is also necessary to apply efforts to educate all patients, regardless of education status, on proper use, function and benefit.

The study’s theoretical framework – both the Theory of Planned Behavior and the FITT model – support the study results. As the Theory of Planned Behavior predicts, those with positive intentions (or high favorability perceptions) toward telemedicine usage will likely behave as such and utilize the technology. In this study, over half of respondents (68%) had a positive favorability reception (or intention) towards telemedicine. This holds positive implication for telemedicine innovation to serve as a suitable alternative to face to face care delivery. Because intention leads to behavior [13, 17], those study participants with high favorability (i.e. intention), will likely utilize telemedicine (i.e. behavior).

Technology adoption and use is dependent upon a number of variables: the user, the task and the technology at hand. The FITT framework [15]captures this construct and assesses utilization based on these overlapping factors. As the study results conclude, those who view telemedicine utilization with an unfavorable attitude are users primarily without a high school diploma, or those with only a high school/vocational school diploma. This particular user reveals a disconnect with the technology (i.e. usability and functionality) and task (complexity) which leads to unfavorable perception of use [3, 17].

It is necessary to reduce barriers to teleoncology utilization, such as fear, lack or education or lack of understanding, so that patients and providers, alike, can successfully adopt and utilize telemedicine tools. Exploring the reasons why individuals perceive negative favor towards telemedicine usage will be key in eliminating barriers of use. As it relates to education, patients should receive proper guidance and training concerning the use of telemedicine tools; ease of use should also be communicated to patients to alleviate their concerns and increase positive favorability. Future research opportunities, as suggested by the researchers, include expanding the sample size in an effort to analyze the variables once more, with a larger, more generalizable population.

It is important to note that Artificial Neural Networks displayed a higher level of accuracy when compared to Chi-Square analysis for the dataset used for the study. This analysis allowed us to predict the threshold identifiers as well. The analysis delineated here may form the basis for further exploration in experiments where traditional methods used in behavioral science may not sufficiently answer questions that may be answered by Artificial Neural Networks.

5.1 Limitations

There are several limitations of this study. First, the Cronbach’s Alpha for the negative readiness factor questions was 0.476 (alpha $<$ 0.70), which is lower than the acceptable value of 0.70. This indicates that the questions may be poorly worded leading to inaccurate responses on the survey questions.

Second, although the sample of respondents are representative of the population of Louisiana, the small sample size is a limitation of the study, and thus, the study results are not very robust. The study sample proved homogeneous and instead, a more heterogeneous sample should be considered. Further, the demographic of populations within the 50 states is very different, thus the results from this study may not be generalizable to other states in the United States.

Third, with respect to usage of neural networks the amount data used for the analysis was relatively smaller proportion when compared to traditional neural network analysis that consumes large amount of data to provide accurate results. Also, the analysis would have improved by modifying the program using category variables.

Lastly, self-reported answer may contain bias if respondents feel that they should answer the survey questions as to how they think the researchers would want them to answer. Nevertheless, the limitations of this study do not de-tract from the significance of the findings from this study.

6. Conclusion

The results from this study will provide policy makers, clinicians and health care systems with information as to the favorability of telemedicine utilization. Telemedicine offers an alternative method in which to deliver care that can be more accessible and affordable to patients. As the United States struggles with access and cost issues within our health care system, it would be prudent to look towards telemedicine as another method to deliver health care services. Those individuals who are not yet ready to accept telemedicine as a means in delivering health care services, policy makers, clinicians and administrators, alike, may want to boost educational campaigns to increase these patients’ knowledgebase and understanding of the benefits of using telemedicine.

Future research opportunities include measuring patient favorability for telemedicine based on their specific disease/condition, health care outcomes, or for basic primary care needs. The decision to utilize telemedicine is impacted by many factors. With the rapid advancement of healthcare technology, telemedicine will be incorporated in the health care field in many different capacities. Thus, it is important to understand patient favorability to the use of telemedicine.

Footnotes

Acknowledgments

The authors would like to express thanks and appreciation to Dr. Thomas Wan. Without his guidance and mentorship, this study would not have been possible.

Conflict of interest

The authors declare that there is no conflict of interest.

References

Perednia

Allen

. Telemedicine technology and clinical applications. JAMA 1995; 273(6): 483-488.

Bashshur

. On the definition and evaluation of telemedicine. Telemedicine Journal 1995; 1(1): 19-30.

Gurupur

Shettian

, et al. Identifying the Readiness of Patients in Implementing Telemedicine in Northern Louisiana for an Oncology Practice. Health Informatics Journal 2016. DOI: 10.1177/1460458216639740.

Roine

Ohimaa

Hailey

. Assessing telemedicine. A systematic review of the literature. Canadian Medical Association Journal 2001; 165: 765-771.

Gatti

Pravettoni

Capello

. Tele-oncology. Switzerland: Springer International Publishing, 2015.

Palkhivala

. Canada develops models of teleoncology. Journal of the National Cancer Institute 2011; 103: 1566.

Cimperman

Brencici

Trkman

Stanonik

. Older Adults’ Perceptions of Home Telehealth Services, Telemedicine and e-Health 2013; DOI: 10.1089/tmj.2012.0272.

Gruca

Nam

Tracy

. Trends in medical oncology outreach clinics in rural areas. Journal of Oncology Practice 2014. DOI: 10.1200/JOP.2013.001350.

Wootton

Bonnardot

. Telemedicine in low-resource settings, Frontiers in Public Health, 2015. DOI: 10.3389/fpubh.2015.00003.

10.

Hede

. Teleoncology gaining acceptance with physicians, patients. Journal of the National Cancer Institute 2010; 102: 1531-1533.

11.

Dewar

Bull

Sproat

Reyes

Malvey

Szalma

. Testing the reliability of a measure of motivation to engage with telehealth technology, in: Proceedings of the Humans Factors and Ergonomics Society 2016 Annual Meeting, 2016, pp. 1114-1118.

12.

Gurupur

Wan

TTH

. Challenges in implementing mHealth interventions: a technical perspective, Mhealth, 2017, DOI: 10.21037/mhealth2017.07.05.

13.

Ajzen

. The theory of planned behaviour. Organizational behaviour and human decision processes 1991; 50: 179-211.

14.

Gagnon

M-P

Godin

Gagné

. Fortin

J-P

Lamothe

Reinharz

Cloutier

. An adaptation of the theory of interpersonal behaviour to the study of telemedicine adoption by physicians, International Journal of Medical Informatics 2003; pp. 103-115.

15.

Tsiknakis

Kouroubali

. Organizational factors affecting successful adoption of innovative eHealth services: A case study employing the FITT framework. International Journal of Medical Informatics 2009; 78: 39-52.

16.

Honekamp

Ostermann

. Evaluation of a prototype health information system Using the FITT framework. Informatics in Primary Care 2011; 19: 47-49.

17.

George

Hamilton

. Baker

. How Do Low-Income Urban African Americans and Latinos Feel about Telemedicine? A Diffusion of Innovation Analysis, International Journal of Telemedicine and Applications 2012; DOI: 10.1155/2012/715194.

18.

Mair

Whitten

May

, et al. Patients’ perceptions of a telemedicine specialty clinic. Journal of Telemedicine and Telecare 2000; 6: 36-40.

19.

Satcher

Bogler

Hyle

, et al. Telemedicine and telesurgery in cancer care: Inaugural conference at MD Anderson Cancer Center. Journal of Surgical Oncology 2014; 110: 353-359.

20.

Weinerman

Barnett

Loyola

, et al. Telehealth – A change in a practice model in oncology. Telemedicine and e-Health 2012; 18(5): 391-393.

21.

Ammerwerth

Iller

Mahler

. IT-adoption and the interaction of task, technology and individuals: a fit framework and a case study. BMC Medical Informatics and Decision Making 2006; DOI: 10.1186/1472-6947-6-3.

22.

Jhaveri

Larkins

Sabesan

. Telestroke, tele-oncology and teledialysis: A systematic review to analyse the outcomes of active therapies delivered with telemedicine support. Journal of Telemedicine and Telecare 2015; 21(4): 181-188.

23.