Site takes the guesswork out of identifying pills

NLM's Pillbox lets users quickly narrow down possibilities from thousands of records

Agency: National Institutes of Health's National Library of Medicine

Project: Pillbox

Location: pillbox.nlm.nih.gov

Technology: LAMP stack; Adobe Flex; Asynchronous JavaScript and Extensible Markup Language for predictive text.; Chemical spell checker; In-house search engine; RESTful Web Service for application programming interface; and Open-source wrappers written by developers making applications with the Pillbox API, in both Ruby and Python.

With its Pillbox Web site, the National Library of Medicine has developed a fast way for medical personnel, law enforcement and anyone else to figure out the contents of a pill. By following prompts to submit descriptions of pills, users can narrow down from a list of thousands of drugs to identify the one they have in hand, with a clinical description of the drug, the information on the Food and Drug Administration label and, often, a high-resolution photo of the pill.

Poison control centers get 1.1 million calls a year to identify drugs in emergency situations, at a cost of about $50 per call, said David Hale, social media strategist at NLM and Pillbox project manager, at the Health 2.0 conference in Washington, D.C., in June.

The data to identify those drugs already existed at FDA and NLM, Hale said, but it was buried so deep in regulatory and nomenclature issues that it was largely unusable.


Related stories:

A renaissance of government Web apps

10 gov apps that get results

Great dot-gov Web sites 2009

Great dot-gov Web sites 2008


So, in essence, “FDA and the NLM hacked our own data, so you can hack us,” he said.

Pillbox is part of the Open Government Initiative and Community Health Data Initiative, which put government data into platforms that allow developers and communities to address health challenges, Hale wrote in an e-mail message. “Pillbox's data, images and search are exposed through an [application programming interface], allowing developers to create applications such as voice-activated pill ID, iPhone apps and even a Facebook game,” he said.

NIH Pillbox site

The API, like Pillbox, is still in beta, though Hale said the library will soon be removing the disclaimer and opening the API.

Users visiting the site can choose an Adobe Flex or HTML process. Click on one, and you first get a screen warning that the site is under development and not intended for clinical use. It also gives a phone number to call if you suspect a poisoning emergency.

In the Flex application, users see a set of tabs for identifying the imprint, shape, color, size and scoring on the pill. As you enter the information, the site shows high-resolution images of possible matches, narrowing down the list with each new bit of information.

If you think you’ve found a match, click on it to get information about that drug from two sources: FDA’s label for the drug, and MLN's Drug Information Portal, which provides links to drug information databases at FDA, NLM and the National Institutes of Health, such as NLM’s MedlinePlus and ClinicalTrials.gov.

The HTML tool provides greater functionality, Hale said, letting users search for active and inactive chemical ingredients. The latter search method is particularly beneficial for people with allergies to commonly used inactive ingredients, such as lactose or benzoate. “The HTML tool is also Section 508-compliant for accessibility, working with screen readers and text-based Web browsers,” he said.

NLM has been developing the site for about three years and has had a formal partnership with FDA for two years, Hale said. The beta version of Pillbox went live in September 2009, at first with 779 high-resolution photographs and 5,693 descriptions of pill-form prescription drug medications. Hale said a major upgrade will be coming soon, bringing the total to 1,200 images and 10,000 records, including over-the-counter and veterinary medications. Pillbox has had more than 200,000 visits from 117 countries.

“We’ve taken our entire dataset and exposed it through our API,” Hale said at the conference. And more importantly, he said, they exposed the Web service. “We’re driving traffic to our Web site, [but] that doesn’t really work. What you need to do is drive traffic to your information.”

And developers have taken advantage of the API. The voice-activated system, which helps a caller identify a drug while on the phone, was written in three weeks by a junior at George Washington University, who also produced instant messaging and Short Message Service versions, he said.

“Government shouldn’t necessarily be creating the resources,” Hale said. “We should be creating the environment and the tools that developers need.”

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above