Year: 2016 | Clemson University | Undergrad Team Members: Ben Williams, Michael Smith, Freddie Pope | Skill: Heuristic Evaluation, Survey, Interview, Cognitive Walkthrough, Rapid Prototyping, Usability Testing | Role: PhD Leader
When dealing with health-related matters, there are a plethora of available resources to help people understand medical information and track their own personal health. If these resources are delivered through the use of technology, they are collectively known as examples of a telehealth service. However, there are some issues with the telehealth, such as wide-spread information from the internet, the validity problem of the information, and inconvenient usage. Therefore, there exists a clear market for a central health-related application to deal with these issues. An application that combines all of the various resources previously required to keep up with health would make individuals much more likely to actually utilize these resources. Launching one application to look up reliable information about an illness, or to search physicians or medicines’ information, would drastically reduce the amount of time required to complete each of these tasks.
HCC CYCLE STAGE 1: DEFINE USERS' NEEDS
Heuristic Evaluation From the heuristic evaluation on existing healthcare applications and comparison analysis, the researchers found that (1) the diagnostic procedure was not intuitive; (2) the applications could not provide an accurate diagnosis without users’ medical record and history; (3) the applications included unnecessary information; (4) they still lacked the functions and information users needed based on our survey results. The researchers resolved these problems with the creation of the HealthAdvisor mobile application.
Design Requirements First, the design will be a mobile application, because users are able to use it anywhere, and anytime. Based on previously defined target users’ needs, the design should meet the following requirements:
Be time saving and convenient. From the survey result, saving time and convenience are two most attracting points for using health app. 1) Users expect the health app include all the information they need (medicine, symptom, condition, and physician information), and 2) it should be convenient for them to set medication reminder and make appointments with physicians.
The diagnostic procedure should be intuitive, matching the real world. Consequently, agent-based interaction should be considered. The task of parsing the information available to an individual has become complicated, especially since there is a wide array of medical information that would need to be sifted through to get what is applicable to the situation. In addition, during the process of narrowing down what is relevant to a user, the questions that the conversational agent asks is based off of information already provided by the user. These aspects highlight what makes a conversational agent a better choice over alternative methods of data input for an application similar in function to ours.
The app should provide more accurate diagnosis, diagnosing based on users’ previous health history, physician synchronized record and health tracking devices (sleeping tracking etc.).
Users should be able to search local common diseases where they plan to travel.
Users should be able to ask questions about their conditions, or search questions and answersother users created.
HCC CYCLE STAGE 2: DESIGN LOW-FIDELITY PROTOTYPE
Low-fidelity Prototype Screen Samples
STAGE 3: EVALUATION & REDESIGN ITERATIONS ON LOW-FI
In this stage, the low-fidelity prototype was evaluated by six experts and redesigned for two iterations. They were HCC, CS PhD or the user experience designer who worked in industry. The researchers tested the prototype with two experts in each iteration.The method the team used was cognitive walkthrough.
STAGE 4: DESIGN HIGH-FIDELITY PROTOTYPE
After fixing all problems in low-fi prototypes, we built a high-fi prototype based on the last version of low-fi, using Just In Mind software.
High-fidelity Prototype Screen Samples
STAGE 5: USABILITY TESTING
We collected both quantitative data and qualitative data in high-fidelity prototype evaluation section, including the time, error number, successfulness and user satisfaction (using post-test questionnaire). Using the post-task questionnaire, the researchers were able to know the experts’ attitudes, feedback and problems for each single task, which helped the researchers focused on the problems in detailed. Six task scenarios are shown below:
Suppose you are experiencing headache, chat with the conversational agent Molly to diagnose diseases.
After getting the report, make an appointment with physician.
Suppose you will travel to Puerto Rico this week, search the local common diseases there.
Set medication reminders for Ibuprofen or Cafergot.
Check the first discussion in Migraine group in patients’ forum.
Check the scheduled appointments, medication reminders, and the health history.
Quantitative Data Performance Data Measurements:
Time spend for each task
The number of errors occurred for each task
The successfulness for each task
Preference Data Measurements:
Rate the ease or difficulty of performing this task (1very difficult-5very easy)
Rate the time it took to complete this task (1more time than expected-5less time than expected)
Rate the likelihood that you would use this feature/task (1not likely at all-5very likely)
SUS (System Usability Scale)
Qualitative Data The participants were also asked to think aloud when they performed the tasks, so that the researchers could collect qualitative feedback.
The SUS scores of the high-fidelity prototype is 75. According to Bangor, products with SUS scores above 70 are passable.