More than just Skin-Deep: Pedsdermatlas

Christoph U. Lehmann, MD, Michael R.G. Cohen and Bernard A. Cohen, MD

Anyone who has generated online medical education resources, knows how difficult and time consuming the gathering of content and its transfer into a web compatible format can be. We would like to describe an online educational resource that has minimized these problems.was designed to minimize these tasks.

Pedsdermatlas was conceived when two developers who have offices on the same floor realized that each had something the other was interested in: The dermatologist had an unlimited number of dermatology images and the medical informatician had a tool to publish these images online.

Pedsdermatlas was designed using a Microsoft Access database. The most difficult portion was to select the fields necessary to contain all needed information and then to designWe designed a form-based interface that would allowallows domain experts with little or no computer knowledge to enter image information. Using the replication feature available with Microsoft Access, all potential local contributors and editors received a replica of the database, which can be easily be synchronized with the master database from one of the developers’ computers or through the Johns Hopkins LAN. This allows multiple editors to work simultaneously on this project.

The database contains information on the image name and file type, its location, as well as patient and disease information. This includes patient age, skin pigmentation and body part as well as diagnosis, category of disease, name of the contributor and comments and keywords. Using Cold Fusion the database is connected to the Internet. The "on-the-fly" generation of all pages "on-the-fly" allows the content of the website to be altered immediately after a change in the database. All pages are based on dynamic queries to the database:

1. Alphabetical index.

When a user selects the link to the alphabetical index, the database is queried and all categories, diagnoses and body sites are returned in an alphabetical order and displayed as links on the page. By selecting any of these links (for example the diagnosis ‘Bite, Insect’), this information is relayed to another database search where only images with the diagnosis are returned.

2. Quick Search.

The quick search feature lets the user choose keywords. The database is then searched based on these keywords (Boolean ‘OR’ if more than one keyword) in the fields for diagnosis, category and comments.

3. Advanced Search.

This feature allows the user to determine body site and age range and pigmentation for the patient, do more sophisticated Boolean searches, as well as search by image name (important for parents who would like to follow their child’s skin condition over time)

4. Contributor index.

The results of any of these four queries are identical. A list of images is displayed with a small thumbnail of the image on the left and pertinent information to the right. If the user selects the thumbnail, the full image and its information is displayed as well as any related images in thumbnail format. The user may select the related images for the full image view. When an image is displayed, the diagnosis is listed and highlighted. Selecting the diagnosis opens a direct query into the National Library of Medicine's Pubmed for articles relevant to this diagnosis in a new browser window.

Another feature of the pedsdermatlas is the Quiz. If a user selects this mode, an image is randomly selected from the database (no duplications during a session) and displayed. The user is offered a choice of up to 20 multiple diagnosis options (extended multiple choice design). The list of multiple-choice options is based on a dynamic query to the database, which selects diagnoses of other images that display the same body site. Once a user selects an answer, it is evaluated for correctness and he has the option to review the correct answer, to retry or to move on to the next image. As long as a user stays in the quiz mode, the number of questions tried and the number of correct answers are tracked and stored in a separate database.

The design of pedsdermatlas is based on a series of dynamic queries to the database, which generate the resulting pages "on-the-fly". This design has enabled us to continuously update the pedsdermatlas with new images without the need for any new html coding, link management and page updating. This design frees the Cold Fusion developer from maintenance tasks and allows him to continuously add new features.

The design based on a database and Cold Fusion has two additionally benefits not yet discussed:

1. Monitoring user searches especially those that did not result into a match in our database, we were able to identify specific user interests and to modify our image additions to more closely match those specific needs. The most commonly requested images are in the area of viral exanthemas.
2. We encourage users to submit their own images for incorporation into pedsdermatlas. A series of simple forms allows potential contributors to provide information about themselves and the image they would like to submit. They can then upload the image to our server where it is processed. The database entry for any image submitted this way does not become active, unless until it was approved by one of the editors. The submission feature has been very successful in the past and lead to international contributors.

In summary, using a dynamic design based on a database approach, we quickly generated an educational tool, which has been well received (JAMA reference here). Allowing editors and contributors to manipulate the database resulting in immediate changes in the web page resulted in fast growth, up-to-date content as well as significant satisfaction for contributors who can review their submissions immediately after the database has been updated.