Thesis
saoajsylau
HCIN550TelehealthProject112.pptx
TELEHEALTH
PROJECT 1
Mehak Sharma, Shweta Patel, Na Zeng, Bolade Yusuf
Telehealth
Issues
Traditional Face To Face healthcare models have many limitations: inadequate mobility; regional distance; operating hours; parking limitation
Why telehealth?
Can provide education and self-management support
Data Mining methods can support clinical decision making
Benefit doctor-patient communication, patient-patient communication
Research authors’ goals
Comparing decision trees, data mining technology and clustering using in Telehealth
Introduction
Exchanging medical information electronically from one site to another. (Tuckson et al., 2017, p. 1591).
Used between clinician to clinician, clinician to patients as well as patient to mobile health technology.
Cost-effectively provide customized and preventive treatment.
Telehealth is defined as exchanging medical information electronically
from one site to another with the purpose of improving patients’ health
Telehealth is used between clinician to clinician, clinician to patients as well as
patient to mobile health technology.
The increasing global health spending has enabled healthcare organizations to adopt emerging health technology for chronic disease management and cost-effectively provide customized and preventive treatment.
Introduction
Provide preventive medicine and customized healthcare through value-based treatment models.
Although Telehealth and technology aspects have existed for decades.
Telehealth enables patients to be tracked remotely and their condition development controlled through constant evaluation
Although global health expenditures are expected to grow to $18.28 trillion by 2040, the future of Healthcare organizations' is poised to utilize developments in Telehealth technology and big data analytics to provide preventive medicine and customized healthcare through value-based treatment models.
Although Telehealth and technology aspects have existed for decades, the Covid-19 pandemic has taken Telehealth to the mainstream in the face of a worldwide crisis that is demolishing health facilities.
Telehealth enables patients to be tracked remotely and their condition development controlled through constant evaluation; whereas Big Data Analytics integrates data obtained from Telehealth modality covering both objective data (e.g. vital signs, ambient environment) and subjective detail (e.g. symptoms and patient behavior).
How telehealth and data analytics are making a difference in healthcare?
Telehealth has great potential to expand the capacity of healthcare
For example:
Apple is working on a wearable medical-sensor-laden device “iWatch'' to monitor blood through the skin.
Google announced the development of eye contact lenses that could analyze glucose levels through tears.
Telehealth has great potential to expand the capacity of healthcare to reduce risks, improve physicians-patients and patients-patients communication, and reveal unseen patterns or sensory features in a ubiquitous, personalized and continuous manner.
Data Review
Chronic conditions are the primary cause of ill health, affecting > 68 percent of all deaths around the globe. Many factors can lead to appointment non-attendance.
Additional obstacles to healthcare that can obstruct access to standard FTF services.
Telehealth technologies can be used to provide education, self-management support and have several advantages over traditional FTF models of care.
Telehealth approaches may help chronic condition patients to deliver comprehensive treatments and manage a shift in habits.
In clinical decision-support structures, data mining methods are rapidly being utilized to help doctors in decision making by analyzing factors, effects and characteristics of patients.
The factors such as Patient-centered barriers, including inadequate mobility and regional distance, operating hours and missing appointments, can lead to appointment nonattendance followed by increased rate of deaths around the world. Additional obstacles to healthcare that can obstruct access to standard FTF services including administrative negligence, inadequate access to clinic facilities, restricted parking and undesirable clinic operating hours.
Through telehealth services one could facilitate and sustain lifestyle changes by managing shift in eating habits and are adjustable in time and location, with the ability to deliver comprehensive treatments that may not be possible for conventional treatment models.
Analytical Techniques
Text Mining: is an artificial intelligence (AI) technology, uses natural language processing to transform the unstructured text in documents and databases into normalized, structured data suitable for analysis.
Regression analysis: is the process of identifying and analyzing the relationship among variables. It can help to understand the characteristic value of the dependent variable changes, if any one of the independent variables is varied. It is generally used for prediction and forecasting.
Analytical Techniques
Decision tree: commonly used in operations research, specifically in decision analysis, to help identify a strategy most likely to reach a goal.
Clustering: identifies clusters of similarities and then forms groups of objects that are more similar in terms of certain aspects than other groups. Unlike classification, the groups (or clusters) are not predefined and can take different forms depending on the data analyzed.
Analytical Techniques
| STUDY INVOLVED | ANALYTICAL TECHNIQUE USED | DETAILS | TAGS |
| How to identify recipients of telehealth by deducing the most important attributes from a dataset of current patients. | 1. Data Analysis: Generate a target variable/attribute from patient data 2. Supervised Learning: Decision tree model algorithm: 3. Prediction: Process incoming data. | 1. Setting data points parameters (age, patient ICD codes, hospital size, patient location from hospital, hospital stays, telehealth monitoring capabilities) 2. C4.5 is a statistical classifier tool (aka J48) Classify data to Supervised, Decision Tree, C4.5, Classification. allow it to process and flow through decision tree. | Supervised, Decision Tree, C4.5, Classification. |
Analytical Techniques
| STUDY INVOLVED | ANALYTICAL TECHNIQUE USED | DETAILS | TAGS |
| Analyze Twitter tweets (location, volume, content) association with telehealth with Covid-19. | 1. Text Mining: Natural Language Processing - Breakdown and Analyze tweets 2. Unsupervised Learning: Descriptive Analysis: Generalized Linear Regression + K means Clustering Analysis (+ Elbow Method) 3. Geospatial Analysis: Visual geographical distribution of tweets correlation with cases. | 1. Tokenization and Stem-Rooting to tag and reduce noise and categorize words 2a. Generalized linear regression to study association between tweets and number of confirmed cases (P < 0.05) 2b. K means clustering to classify tweets into topics. | NLP, Unsupervised, K-means Clustering, Generalized Linear Regression, Geospatial, Association. |
Analytical Techniques
| STUDY INVOLVED | ANALYTICAL TECHNIQUE USED | DETAILS | TAGS |
| Impact of telehealth on a diabetic cohort over 12 months | Multilevel Models to assess impact with data from the self-reported questionnaires over a period. | Adjustments-covariate adjustment to control baseline variables | Supervised, Multilevel Model, Sidak, Covariate, repeated measures design, cluster randomized controlled trial, Classification. |
| Assessing whether telehealth had impact on glycosylated hemoglobin among type 2 diabetes. | 1.Mixed Effects logistic regression. 2.Sensitivity Analysis. | Repeated measures, cluster randomized controlled trial, mixed Effects Logistic Regression. |
Analytical Techniques
| STUDY INVOLVED | ANALYTICAL TECHNIQUES USED | DETAILS | TAGS |
| Assess novel clustering method based on graph models. | Clustering System based on Graphs compared it Mean Shift, K-means, ward hierarchical clustering, db. scan, birch clustering systems. | Supervised learning, clustering, classification, unsupervised learning, kernel trick. | |
| Assess the implementation of tele-PCMHI to new sites | 1. Generalized Linear Mixed Models - fixed for innovation and time and random effects 2. Mixed Logistic Model/or standard logistic regression model. | Standard logistic regression if intraclass correlation is insignificant; and mixed logistic model if it is significant. | Cross sectional design; multilevel model, intraclass correlation. |
Analytical Techniques
| STUDY INVOLVED | ANALYTICAL TECHNIQUES USED | DETAILS | TAGS |
| Assess telehealth impact on implementing dietary interventions via secondary studies | 1.Data analysis: Random effects meta-analysis (DerSimonian and Laird Method) + Fixed effects regression model 2. I-square to assess heterogeneity, variability between the studies 3. Sensitivity analysis 4. Egger's plot assess potential publication bias of studies used. | Meta-Analysis, Fixed effects regression, I-square, Sensitivity Analysis, Egger's |
Outcomes
Explored the models to identify the appropriate telehealth service candidates.
The decision tree model was selected to solve the problem of telehealth patient classification for the
following reasons:
For the perspective of sensitivity, two models performed equally well.
For the perspective of accuracy, specificity, and precision.
Compared the differences between the telehealth services and usual care for different populations.
Explored the models to identify the appropriate telehealth service candidates.
After comparing the decision tree model provided by heuristic decision tree telehealth classification approach (HDTTCA) and the logistic regression, the authors selected the decision tree model to solve the problem of telehealth patient classification for the following reasons:
For the perspective of sensitivity, two models performed equally well.
For the perspective of accuracy, specificity, and precision, the decision tree model worked better than logistic regression.
Compared the differences between the telehealth services and usual care for different populations.
Different studies focused on the different populations.
Most studies indicated that there’s no significant difference.
One study showed the telehealth could modestly improve glycemic control.
Investigated the tweets contents to identify the contributions of telehealth during COV-19 pandemic.
Study investigated the rapid shift in telehealth adoption amidst the recent coronavirus Covid-19 pandemics.
Outcomes
Different studies focused on the different populations such as patients with type 2 diabetes.
Most studies indicated that there’s no significant difference between the telehealth services and usual care when comparing the life quality.
There’s one study showed the telehealth could modestly improve glycemic control among patients with type 2 diabetes, although it seems unlikely to produce significant patient benefit.
Investigated the tweets contents to identify the contributes of telehealth during COV-19 pandemic.
Study investigated the rapid shift in telehealth adoption amidst the recent coronavirus Covid-19 pandemics. The result showed the need for widespread implementation of digital health and the importance of supporting policy changes to unleash the power of this technology.
Comparison among different Analytical Methods
There’s only one article selected using natural language processing (NLP) due to the unstructured text data. This analytical technique would not be considered for our team’s topic.
Regression, decision tree and clustering were the most used analytical techniques in the studies. It is generally used for prediction and forecasting.
Decision tree is commonly used in operations research, specifically in decision analysis, to help identify a strategy most likely to reach a goal.
Clustering identifies clusters of similarities and then forms groups of objects that are more similar in terms of certain aspects than other groups.
Our topic is a little wide and different so when working on the topic at the very beginning, we could select to use clustering to have a quick look at the data and then use classification techniques to do a further exploration. Similar to one article we found we also need some data analytics to compare different methods to find out the better technique.
Only one article used natural language processing (NLP), which is text mining to explore their tweets data. It’s not a typical analytical technique for telehealth. It was selected due to the unstructured text data. This analytical technique would not be considered for our team’s topic.
Regression, decision tree and clustering were the most used analytical techniques in the studies. Regression analysis is the process of identifying and analyzing the relationship among variables and to understand the characteristic value of the dependent variable changes, if any one of the independent variables is varied. It is generally used for prediction and forecasting.Decision tree is commonly used in operations research, specifically in decision analysis, to help identify a strategy to reach a goal. Clustering identifies similarities and then forms groups of objects that are more similar in terms of certain aspects than other groups. Unlike classification, the groups (or clusters) are not predefined and can take different forms depending on the data analyzed.
Telehealth is a wide topic so initially we selected to use clustering to have fast access to the data followed by classification techniques and data analytics to compare various methods with the intense of finding better technique.
Summary
Telehealth steadily increases as it has become a viable modality to patient care, especially with Covid-19.
Using evidenced based self-management techniques targeting self-care and QoL delivered via telehealth, shall facilitate intervention delivery.
Telehealth technologies to manage chronic disease and deliver cost-effective personalized and preventive care.
Data mining techniques are increasingly used in clinical decision making for more accurate and effective decisions.
Summary
The classification Model is the most commonly practical.
Decision trees is a good approach in identifying the potential receivers of telehealth services.
Systematic review and narrative analysis explore telehealth and patient satisfaction association from effectiveness and efficiency view.
Lack of consistency in the telehealth literature in the study methodologies and data analysis techniques used.
Longer-term studies could examine impacts of telehealth on complications of diabetes and acute MI.
References
1. Massaad E, Cherfan P (April 26, 2020) Social Media Data Analytics on Telehealth During the COVID-19 Pandemic. Cureus 12(4): e7838. doi:10.7759/cureus.7838
2. Chern, C., Chen, Y. & Hsiao, B. Decision tree–based classifier in providing telehealth service. BMC Med Inform Decis Mak 19, 104 (2019). https://doi.org/10.1186/s12911-019-0825-9
3. Hirani, S. P., Rixon, L., Cartwright, M., Beynon, M., Newman, S. P., & WSD Evaluation Team (2017). The Effect of Telehealth on Quality of Life and Psychological Outcomes Over a 12-Month Period in a Diabetes Cohort Within the Whole Systems Demonstrator Cluster Randomized Trial. JMIR diabetes, 2(2), e18. https://doi.org/10.2196/diabetes.7128
4. Russell, T. G., Martin-Khan, M., Khan, A., & Wade, V. (2017). Method-comparison studies in telehealth: Study design and analysis considerations. Journal of telemedicine and telecare, 23(9), 797–802. https://doi.org/10.1177/1357633X17727772
References
5. Caffery, L. J., Martin-Khan, M., & Wade, V. (2017). Mixed methods for telehealth research. Journal of telemedicine and telecare, 23(9), 764–769. https://doi.org/10.1177/1357633X16665684
6. Steventon, A., Bardsley, M., Doll, H. et al. Effect of telehealth on glycaemic control: analysis of patients with type 2 diabetes in the Whole Systems Demonstrator cluster randomised trial. BMC Health Serv Res 14, 334 (2014). https://doi.org/10.1186/1472-6963-14-334
7. Kruse, C. S., Krowski, N., Rodriguez, B., Tran, L., Vela, J., & Brooks, M. (2017). Telehealth and patient satisfaction: a systematic review and narrative analysis. BMJ open, 7(8), e016242. https://doi.org/10.1136/bmjopen-2017-016242
8. Jelinek, Herbert & Cornforth, David & Kelarev, Andrei. (2016). Advanced Clustering Method for Neurological Assessment Using Graph Models. International Journal of Computer & Software Engineering. 1. 10.15344/2456-4451/2016/109.
References
9. Owen, R.R., Woodward, E.N., Drummond, K.L. et al. Using implementation facilitation to implement primary care mental health integration via clinical video telehealth in rural clinics: protocol for a hybrid type 2 cluster randomized stepped-wedge design. Implementation Sci 14, 33 (2019). https://doi.org/10.1186/s13012-019-0875-5
10. Jaimon T Kelly, Dianne P Reidlinger, Tammy C Hoffmann, Katrina L Campbell, Telehealth methods to deliver dietary interventions in adults with chronic disease: a systematic review and meta-analysis, The American Journal of Clinical Nutrition, Volume 104, Issue 6, December 2016, Pages 1693–1702, https://doi.org/10.3945/ajcn.116.136333
11. AbuKhousa E (2017) Analytics and Telehealth Emerging Technologies: The Path Forward for Smart Primary Care Environment. J Healthc Commun. 2:67. doi: 10.4172/2472-1654.100108
12. Akhondzadeh Noughabi, Elham & Ameri, Kimia & Alizadeh, Somayeh. (2017). Application of Data Mining Techniques in Medical Decision Making: A Literature Review and Classification. 10.4018/978-1-5225-2515-8.ch01
13. Tuckson, R. V., Edmunds, M., & Hodgkins, M. L. (2017). Telehealth. New England Journal of Medicine, 377(16), 1585–1592. https://doi.org/10.1056/nejmsr1503323
