We developed the ThaiEpidemics app based on the mobile application development life cycle (MADLC) model [20, 21]. In creating the app, we obtained information about user requirements through in-depth interviews with three experts working at the Travel Clinic at the Hospital for Tropical Diseases, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand. Since opening in 2004, the clinic has serviced thousands of foreign travelers annually, providing healthcare services, vaccination, and consultation. We collected details concerning user requirements for the app prototype from five travelers who visited the clinic. We also collaborated with the director and individuals responsible for the surveillance databases of the Department of Disease Control, BoE. We developed ThaiEpidemics using React Native (a framework for JavaScript, and cross-platform technology). We selected React Native because it is an open-source software for building cross-platform apps that utilizes a single industry standard language.

The user interface was designed based on the main user requirements obtained when applying the MADLC model (Fig. 1). The app conveyed information related to 14 diseases: dengue, malaria, influenza, food poisoning, rabies, cholera, leptospirosis, tuberculosis, chikungunya, diarrhea, hepatitis A, hepatitis B, scrub typhus, and typhoid. The R506 data from the national disease surveillance report published on the BoE website was downloaded every week, cleaned, and mapped to the appropriate fields before uploading to the database [22] (Figs. 2 and 3), which is hosted on a Firebase cloud computing platform. The app is capable of synchronizing the R506 data stored in the cloud database to the local SQLite database installed on every user’s smartphone.

Design of the user interface in ThaiEpidemics based on main user requirements: (a) visualizing the disease prevalence and status for the current location; (b) visualizing the disease prevalence and status for a specified area

Screenshot of an R506 report and field mapping data for the database

Main tables in the database used for storing R506 data

The R506 reports published on the BoE website were not in a standard format for importing to the database. Thus, both automatic scripts and manual commands were created and used to manage the downloaded data files. The process of data transformation included preparing, cleaning, validating, and importing the validated, cleaned R506 data into the SQLite database before uploading to the Firebase real-time database hosted in the cloud for data synchronization to the mobile app (Fig. 4).

Steps of the downloaded R506 data preparation and transformation

In addition, ThaiEpidemics stores the search logs recorded from user interactions and querying. First, the search logs were stored on the local mobile SQLite database before synchronizing to the Firebase database for further analysis. Similarly, the search log and feedback data collected from the e-survey were stored on the cloud database (Fig. 5).

Conceptual diagram of the system and infrastructure

ThaiEpidemics, was made available on both the Google Play and Apple App Stores. Participants used the app to obtain information about the disease prevalence and status in an area they planned to visit. There were three parameters: province, district, and period. The app searches for data in the database and then displays the information based on the entered parameters along with the location of the province on the map. The participants received the information and maps regarding disease prevalence and status s according to their queries (Figs. 6 and 7).

Screenshot of ThaiEpidemics: (a) disease prevalence and status around current location; (b) situation information categorized by disease name and district

Screenshot of ThaiEpidemics at national-level visualization: (a) number of reported cases at the national level; (b) top five ranking of provinces with the number of cases of each disease; (c) specific queries for information about the disease prevalence and status in a particular area

The parameters specified in the app were as follows: period or duration, distance or buffer radius from the current location, disease name, district, and province. Apart from searching for information about diseases around their current location, the participants could also search for information in areas they wanted to visit. When the participants checked the disease prevalence and status in an area, the app recorded the query in the database.

The app prototype was tested on travelers in tourist attractions in Bangkok and travelers seeking medical services at the Travel Clinic. We invited the participants to read a brochure about the mobile app, which contained the following information: objectives of the mobile app, main features and functions of the app, its benefits, screenshots of the app interfaces, platforms supported, and the app name for downloading from the Google Play or App Store. These stores are public, so anyone—not just individuals at the Travel Clinic—could search, download, and install the app on their smartphones. To ensure that only study participants could access the app, we included a password in the brochure for app installation. If someone were interested in the app and willing to participate in the installation and app usage, they could download the app and install it on their smartphone.

We evaluated user interaction employing a descriptive cross-sectional survey. We collected data from the participants from December 2019 to January 2020. We collected three types of data: user engagement (number of log-in times, how long travelers stayed in the app), search logs (disease name, province, district, duration), and answers to an e-survey. The search logs provided data about all the queries the participants entered on the app when searching for disease surveillance information for a particular area, date, or disease. The logs were stored in the local database before synchronization to the cloud database. The log contained the following data: date, province, district, disease, and location. We informed participants that the app would collect usage and search log details.

For the e-survey, an electronic questionnaire module was built into the app. Following the second log-in, the app prompted the travelers to complete the e-survey. The e-survey covered opinions and preferences regarding awareness of disease prevalence and status by specifying their level of agreement with prompted question using a five-point Likert scale. The questionnaire had two main parts: demographics and app satisfaction. The demographics section contained questions about the participants’ country of origin, gender, age, and education level. Items related to app satisfaction were derived with reference to a study of mobile health-care applications [23]. This part of the questionnaire was based on the use of the Mobile Application Rating Scale, which is a quality measurement tool for assessing the quality of health mobile apps [24].

After the data collection was complete, descriptive statistics were utilized to analyze the characteristics of participants’ demographic data, including key variables of the three data sources. A chi-square test was used to examine the relationships among the following variables: improved awareness of disease prevalence and status, age-group, education level, and continent of origin. Binary logistic regression was also applied to determine whether the variables were significant predictors for improved awareness of disease prevalence and status after using the app.