Hello, welcome to the Introduction to Telemedicine Analytics my name is Asia Law. I am a Lead Business Intelligence Analyst for Johns Hopkins Medicine and today I will be talking to you about the basics of telemedicine reporting. This includes extracting data from the source, loading data for processing, transforming raw data into custom metrics, and visualizing data that tells a story. The learning objectives for this module is as follows : Identify various telemedicine modalities, explain the framework for building the JHM telemedicine data model, explain how different types of graphs convey different information, describe how graphical visualizations can lead to actionable data, improve operational outcomes using data, and understand the challenges of reporting. First things first, what is telemedicine? This definition is constantly changing and evolving but to paraphrase the American Telemedicine Association, it is the transfer of medical information via synchronous or asynchronous telecommunication technology. What are synchronous and asynchronous telecommunications? Synchronous is a real-time face-to-face clinical virtual visit. Asynchronous is a story-forward non-face-to-face clinical visit. Each of these modalities has subcategories. For synchronous we have TeleScreening which is a triage virtual visit between an ER patient and a remote provider. A video visit is a face-to-face virtual visit between a patient and a provider. A telephone visit is a scheduled real-time video visit that converts to an audio-only phone visit due to technical difficulties with video technology. For the asynchronous modality, we have E-Consult which is a consultation through messaging and the patient's electronic health record. The patient sends a request and the provider responds, E-Visit which is a non-face-to-face virtual visit where the patient fills out a questionnaire to report symptoms and the provider responds with a treatment plan or referral. E-Medical Second Opinion is a provider-to-provider clinical review and assessment of a patient's electronic health record. There are other modalities such as education and remote patient monitoring that have other subcategories as well. Now that we know a little about telemedicine modalities let's talk about the data framework for capturing and reporting patient information. There are three key factors to building the telemedicine data model framework. One factor is the input into the electronic health record. The EHR has things like patient data, visit details, provider information, and many more. The next factor is processing that data by the database management system extracting, loading, and transforming the data. The third factor is outputting the data to a reporting tool converting raw data points into visualizations. Now that we have a little bit of a better understanding of the process flow we can talk in more detail about what tools we use for this data framework. Epic is the input tool where data is entered and collected. Microsoft SQL Server is the processing tool where data is stored and transformed and Tableau is the Reporting tool where data is visualized and reported. Now that we have more detail into that telemedicine framework let's get into what each of these factors actually means. What is an electronic health record? An electronic health record is an electronic version of a patient's medical history that is maintained by the provider over time. It may include all of the key administrative clinical data relevant to that person's care under a particular provider including demographics, progress notes, problems, medications, vital signs, past medical history, immunizations, laboratory data, and radiology reports. The HR automates accessed information and has the potential to streamline the clinician's workflow. EHR is for the input in the patient care management. Next is the database management system. What is a database management system? A database management system is a software package designed to define, manipulate, retrieve, and manage data in a database. A database management system generally manipulates the data itself. Data format, field names record structure , and file structure. It also defines rules to validate and manipulate this data. The database management system is a space to build and manage any specific business logic that may require customization for workflow and nuances for caveats. Here is a real-world business example that requires transformation of data in the DBMS. Video visit conversion to phone visit. Typically a patient schedules a video visit, the patient and provider virtually arrive to that video visit and there could be some technical difficulties in that workflow where the video technology may not be working as expected. The video visit becomes an audio-only phone visit the provider completes that visit with the notation that that video has been converted to a phone visit. It is valuable information for the stakeholders to understand the difference between a successful video visit and a visit that converts to phone due to technical difficulties. With Microsoft SQL code built by the developers in the database management system, it's possible to identify video versus phone visits using that provider's notation in the electronic health record. Using that database code we have the ability to transform the data and create business illogic around that data to better meet the stakeholder needs. What is a reporting tool? A reporting tool is software that helps bring data from various sources to a centralized location to be presented in a meaningful way. There are various tools that provide businesses with the necessary framework to transform data into tables, charts, and graphs for making informed decisions. Examples of reporting tools include Tableau, Power BI, Cognos, Crystal Reports and there are many more. Let's talk a bit more in detail about reporting data and transforming data into visual representations of the information. Data visualizations are very important to easily convey information to various audiences. Certain audiences have different technology levels and understand data in different ways so with data visualizations you have the ability to better convey that data in a visual format. You can use things like line graphs typically to review trends over time. It allows the end-user to easily pinpoint decreases or increases in the metric over time. Bubble charts can be used to quickly see the relationship of data by group based on the size of the bubble. Larger bubbles indicate high value, smaller bubbles indicate low value. A bar chart can be used to compare categories. Let's say the total bar is 100 percent of the volume. In comparing two categories you can easily visualize the breakdown, for example, 80 percent video visits versus 20 percent phone visits, so over the entire 100 percent of that bar, you can see how each category makes up a percentage of the overall volume. Tree maps are used for analyzing nested groups. You can do data by size, color, and specific groupings. There are many other visualizations that you have at your disposal as well. There's an entire catalog that shows you different visualizations and gives you specific use cases for why you may use one visualization over another. Other visualizations include area graphs, box and whisker plots, histograms, pie charts, a geographic map, scatterplot, and more. As I mentioned there's an entire catalog that shows all of these different visualizations and the use cases for each so it's just a matter of exploring that catalog and determining which visualization is appropriate for your data. Various ways to present the data and the goal is to make data actionable. Ways to make data actionable is by understanding the questions that the business wants to answer. Say for instance the business wants to understand why there are video visits that have to be converted to phone visits. By using this bar graph you can see the percentage over time by age. Looking at this bar graph you may say that a patient's age potentially could impact the use of a video visit versus a phone visit. You can see as the age group increase, so does the use of the phone visit percentage increase as well. Potentially more senior patients don't have access or interest in video technologies. Another actionable data point is viewing data by geographic location. With the same example of video visits versus phone visits. Maybe the patient location could impact the increase in phone visits. You may have the hypothesis that more rural areas have an increase in phone visits and the reason could be potentially a weak internet connection in those rural areas. Plotting those points by zip code can give you a deeper understanding into the geographical location and how that impact of that location could either increase or decrease the use of a video visit. Another use case is comparing telemedicine and in-person visits. Visualizing data in this manner can quickly and easily highlight trends. Analyzing trends outliers and patterns to help make data-driven decisions to improve operations. Looking at this bar graph, giving you the date by week on the x-axis you can easily see pre-COVID which was March 2020 the second week or third week of March, you can see there's a lot of in-person visits. Very little to no video visits. As the pandemic picks up you can see that the video visits are now beginning to increase over time and the overall visit volumes have decreased pre-pandemic versus during the pandemic, so this bar graph gives you that easy representation to see how video visits to tele medicine visits increase over time and also how overall volume has shifted during the pandemic. With the bar graphs, you can also see the potential outliers on certain dates. You can see the week of the 25th there's a dip in the data and that can be due to a holiday. Typically memorial day is that week. That can be the attributing factor to that dip in the data. The week of June 29th could be contributed to the holiday the 4th of July, because of that holiday that could be a reason why that data has dipped. Then we see September which is typically Labor Day could be the contributing factor to that dip in the data. Seeing the overall breadth of the data over time you can see how the data fluctuates and potentially have a hypothesis for why the data is shifting balance. In some cases looking at the textual representation of the data is very impactful as well. Just seeing the numbers and seeing the percentages of different types of visits. You can see for in-person visits 77 percent of the volume, video visits is 23 percent of volume. You can easily see the percentage of video versus in-person from the organization. Data can also be used to drive change. Data analysis and a review of key indicators can improve operations. Visualizing an increase or decrease in indicators can highlight areas of success or improvement. This pie graph that we have here is taking a look at the operations efficiency. We have three slices of the pie one being MyChart active, the other being eCheck-In incomplete, and MyChart inaccurate. Mychart is an epic tool that provides online access to health information. Through MyChart a patient has the capability to eChek-In to their scheduled visit. This saves time and resources and positively impacts patients' experience. If the organization has a goal to get all patients active on MyChart. Visualizing such statistics can help leadership focus on areas of improvement. If we look at this pie chart, we can easily see that 59 percent of patients are active on MyChart, 14 percent are inactive, and 27 percent are active and are completing their eCheck-In. If the goal is to have 100 percent of patients using MyChart, the first area that we can focus on is that 14 percent of patients who are not active on MyChart. Knowing this information can lead to outreach and understanding the patient group and why they may not be using MyChart. Maybe they don't know that that tool is available or maybe they just don't know how to use it or could be intimidated by the technology and just need to be educated and walk through how to use it. Twenty-seven percent of patients are active on MyChart but are not using the eCheck-In functionality. Knowing this information can lead to reaching out to users, educating them on the tool and the functionality, and maybe just making them aware that it exists. Maybe the patients aren't aware that eCheck-In is a functionality they can use or maybe they just don't know how to use it. By visualizing this data in this manner you can see the percentages and the patient groups that you want to focus on for meeting the overall organizational goal. Data is powerful, but there are challenges to reporting. I'm going to talk to you a little bit about the challenges of reporting and potentially overcoming some of those challenges. Reporting outcomes are susceptible to manual data error, workflow inconsistencies, and data processing delays. Manual data entry error is always a factor because there are people who input the details into the electronic health record. It's human error, it's human entry. None of us are perfect. In having that manual data entry there's always room for error and once that data is input into the electronic health record, it is then stored and processed in a database, and subsequently, that data is reported in the reporting tool. Workflow inconsistencies can also impact reporting. There are so many different workflows for so many different things and if there are certain people who follow workflow A and other people follow workflow B, when it comes to reporting a unified metric or unified data it's difficult to have a baseline to report because of the workflow inconsistencies. Another challenge is working with data processing delays. Because the data is being stored in a database management system and the data is tracking and storing millions and millions of rows of data, there are sometimes cases where the data processing is delayed because of a failed process or there is downtime or maintenance, so subsequently if that data processing is delayed, the reports will be delayed. If someone is used to seeing their data refreshed at 09:00 a.m. If there are processing delays that might have to be pushed to 10:00 a.m. or 11:00 a.m. or 02:00 p.m. There's downstream impact based on those data processing tools. Another challenge is the rapidly evolving workflows that can impact data integrity. Workflows are constantly changing. There are new workflows that are implicated in the process and that can ultimately have an impact on the data integrity. If workflows are constantly changing and definitions are changing, it could interfere with end-users trust in the data because things are always happening and there are so many moving parts. Rapidly evolving workflows can have an impact to that data integrity. Constantly changing requirements can interfere with version control and stability of the data model because things are constantly changing, the requirements are changing. There's always rapid development that has to happen and if you don't have a core stable version control process sometimes it's difficult to track those changes and just in case there's an issue with something that's implemented if you don't have good version control it's difficult to roll back to your previous version so that you don't have any downtime in your reporting. As requirements change it's difficult to manage those version control processes so you wanna make sure you can have a good version control setup for this data. One of the major challenges is telling a compelling story with the data points that is dynamic and keeps the audience invested. Data is very powerful but if you don't tell a story with the data and you don't have the proper narrative end-users, stakeholders, leadership, aren't going to be as interested and invested in the data. You want to be able to tell a dynamic story, you want to be able to give people the flexibility to look at the data in a different way. You come up with different narratives for different data points, so telling a compelling story is very important but it's also the most challenging. These are some of the challenges of reporting. There are other challenges but as you continue to work through the processes you have your certain lessons learned and you can prepare better for each step of your reporting process. It's been a pleasure providing you with an introduction to Telemedicine Analytics. Thank you for your time, and have a great day
