Algorithm For Reading Emotions

Advancing a decision support system to improve physician practice and patient outcomes

By Ashley R. Smith

There are smart phone apps to tell us if a movie is a blockbuster or a flop, if a restaurant is two stars or five, and if it’s T-shirt weather before we roll out of bed. And if we can’t remember when our bills are due every month, we have to-do lists, calendars and reminders for that as well. Yet we believe — and often expect — our doctors to know and remember everything when, in fact, they need to know far more information than they could ever memorize.

Joshua Steinberg, MD, a family physician at Wilson Hospital in Binghamton, is humble enough to affirm that this information deluge is a very real challenge within the medical profession.

As Steinberg explains, it’s inevitable for questions to arise over the course of a day that a physician cannot answer on the spot. They typically involve uncertainty related to a diagnosis and treatment strategies — such as the precise dose of a medicine or what imaging study can rule out a kidney stone. Studies have shown that while doctors are quick to turn to textbooks and colleagues for support, they usually spend less than two minutes seeking an answer. And therefore, some questions go temporarily unanswered.

No worries, though. They’re not rolling the dice with your health. “Maybe we order an extra test, or refer you to a specialist,” Steinberg explains. “Or sometimes we have a general idea, or there’s not a precise answer.”

He’s the first to admit that is not the best practice for good patient care. “We’ve all had those times when we have five patients in rooms grumbling that they’ve been waiting for 40 minutes to see the doctor,” Steinberg says. “And there’s no time to go read a chapter in a textbook.”

Acutely aware of his needs in certain topics, Steinberg adopted the traditional practice of keeping pocket manuals and guides on hand.

When the Palm Pilot debuted, he became a self-proclaimed power user and was able to make a few apps for personal use.
Then the iPhone arrived. UHS, General and Wilson hospitals went all-in, developing custom software to plug into the hospitals’ patient information system. Soon, every physician, every resident and every student would have a hand-held device.

Steinberg redirected his attention, but unlike with the Palm Pilot, there’s no simple kit for constructing iPhone software. “After buying three different books — one literally called iPhone Application Development for Dummies — I was convinced I couldn’t write a lick of it myself,” Steinberg says. So he made a cold call to the Department of Computer Science at Binghamton University.

He met with computer science faculty including Associate Professor Madhusudhan Govindaraju. While iPhone programming is not part of the Watson School curriculum, they decided to give the partnership a shot.

“We teach students the foundations of computer science, programming, software and hardware,” Govindaraju says. “They are proficient in some programming languages, best practices in software development and how to apply the concepts they have learned to developing applications for a smart phone.”

But just as Steinberg had learned, the students couldn’t rely on picking up a book to learn exactly what to do. The iPhone environment evolves quickly, with Apple releasing new operating systems and new versions of the parameter framework every few months to a year.

The first student to take on the challenge was Jonathan Alter ’10.

Warfarin app

Computer Science students worked with Joshua Steinberg, MD, to develop the Warfarin Guide app that's available for purchase in the Apple store.

“Dr. Steinberg explained what he wanted, how the program should function and its goals. I was to find a way to make it happen,” Alter explains. He — like those students who have followed — read blogs, scoured websites and interacted with other programmers to learn the latest recommended API and methodology to use with the Objective C language for developing interesting features for the iPhone application. “Once I amassed enough knowledge, I wrote some simple code, did testing and played around,” he says. “After a month or so of learning, I broke the project into model–view–controller (a common design pattern for user-interface development — think model = HTML, view = CSS and controller = browser). Then I broke down the code I would need to write, and after a few tries it started coming together.”

To create a truly useful app, Alter also had to address the intricacies of health-care human-computer interaction.

“The target users of these tools are people with very little time who need to see data instantaneously,” Govindaraju says. “We need to figure out their attention span, how IT affects them, their time and their way of doing things. It needs to be an aid to them at the point of care without requiring that they follow it exactly.”

What are all the questions that should be asked? What does the decision-tree look like? Does the patient have X or does the patient have Y and, if so, then what should be asked or what procedure should be performed?

Done correctly, the applications are a decision-support system at the point of care to prevent errors. Providing quick answers, easily and effectively, in 10 to 30 seconds, gets the doctor back to the patient.

Alter, now a second-year medical student at Ross University School of Medicine on the island of Dominica, West Indies, sees firsthand the benefits of his undergraduate work. “Patients get the right care, at the right time, with lower costs, fewer tests and a more thorough understanding of their ailment. The doctors have a better grasp of the ailment, too, so they can treat the patient better, more quickly and more effectively,” he explains. However, there are potential drawbacks to having a phone in-hand at all times. “If you are checking your iPhone, you’re not fully listening to the patient, and it could make things more difficult,” Alter says.

Steinberg and the Watson School have created nine apps thus far that meet three criteria: topics that are recurrent, complex and high-stakes.

A topic that meets all the elements is the blood thinner Warfarin. “This is probably one of the most dangerous medicines in common use besides chemotherapy, and I went the first decade of my career not being particularly good or confident in the management of this medicine,” Steinberg says. “Thin the blood so that a patient doesn’t get blood clots, but don’t make it too thin or else they’ll start bleeding, never stop, and die.”

The Warafin Guide app, written for all types of healthcare providers, answers common questions such as, how do I treat with Warfarin after a first deep-vein thrombosis (a blood clot, most commonly in the legs)? Or, when do I give vitamin K to reverse Warfarin anticoagulation and at what dose?

“For us, it’s a great educational experience for our students,” Govindaraju says. “Students are working on real projects and seeing what it means to interact with a real client. They’re able to put this on their résumé, and some are getting interview calls and subsequently jobs because of it.”

A valuable link between computer science and medicine


Jonathan Alter

Jonathan Alter '10

Medical Student at Ross University School of Medicine, Dominica, West Indies

My two best friends as a freshman convinced me to join Harpur’s Ferry and become an EMT. I enjoyed my time as an EMT but I knew that I could, and should, do more.

I decided to study computer science. But medicine is my calling — I would spend my days at the bedside in lieu of a cubicle. I like the idea of learning every day for my entire life. I like that I get to interact with people every day. And, that I work as part of a team and as an individual.

My CS classes and lab work taught me how to effectively and methodically analyze a problem. In my medical studies, some elements of my education are simple memorization of facts and numbers. But generally, each disease or symptom can also be broken down into smaller pieces to show a grander picture.

Granted, I won’t be using what I learned in Theory of Computation anytime soon. But those skills to view or find a problem, understand the problem at its core, break it down and solve each piece so I can build it back up with a better solution — that is what CS at Binghamton taught me. And I can apply that to almost anything.”


Watson Review Masthead