Here’s an idea for an app: Wouldn’t it be great if there were a smartphone app that could scan my tummy & tell me why it’s hurting? Is it a stomach virus, gluten exposure, FODMAP related, lactose intolerance, corn intolerance, salmonella, too much guar gum or carrageenan, natamycin, not enough fiber, too much fiber, the frequency of meals, or the fact that my sister is coming to visit? I’m willing to do whatever it takes to make it stop hurting. The problem is, it’s often difficult to determine what that is.
I was recently part of a team that reviewed a pilot research grant for a study of the effectiveness of an app vs traditional treatment. While we ultimately chose not to fund that particular grant, I see great possibilities in the future of apps to assist healthcare professionals with diagnosis and treatment of disease and chronic conditions, the physical manifestations of trauma, and mental health issues in general.
While the human element offers insight that may be missed by technology, it also brings inherent bias and inconsistency. In the 1960s, the Oregon Research Institute set out to study how experts rendered judgments. Lew Goldberg, a psychologist, developed a case study in which researchers gathered a group of radiologists and asked them how they determined from a stomach X-ray whether a patient had cancer.
The doctors indicated that there were seven major cues that they looked for. In an effort to create an algorithm that would mimic the decision making of doctors, the researchers created a simple algorithm in which the likelihood of malignancy depended on the seven cues the doctors had mentioned, all equally weighted.
Researchers then presented the doctors with X-rays from 96 different individual stomach ulcers and asked them to rate each one on a seven-point scale from definitely malignant to definitely benign. Without disclosing what they were doing, the researchers showed the doctors each ulcer twice with the duplicates mixed in randomly so that the radiologists wouldn’t notice the duplication.
All data were transferred to punch cards and sent to UCLA for computer analysis. When UCLA sent back the analyzed data, it became clear that this simple computer model was surprisingly good at predicting the doctors’ diagnoses. Even more surprising, the data showed that the diagnoses were all over the place. In spite of the fact that they were trained experts, the radiologists didn’t agree with each other. In fact, they often didn’t agree with themselves. Every single doctor had sometimes contradicted his own diagnosis when given a duplicate X-ray.
The researchers also found that clinical psychologists and psychiatrists deciding whether it was safe to release a patient from a psychiatric hospital wildly differed from each other in their determinations. Further, those with the least training were just as accurate in their judgments as those with more training.
The Oregon researchers then tested the hypothesis that the simple computer model they had designed might be better than doctors at diagnosing cancer. Turns out, the algorithm outperformed even the single best doctor in the group of doctors being studied.
What Goldberg came to realize was that doctors had a good theory of the cues to look for in diagnosing cancer, but in practice they did not stick to their own ideas of how to best diagnose. They tended to weigh things differently. As a result, they were less accurate than a computer model.
Given our current reliance on experts to diagnose, this research isn’t particularly reassuring; however, it does bode well for the inclusion of artificial intelligence in diagnostic procedures. That’s right, just the sort of technology that could be deployed by an app on my handheld device.
Here’s an idea for an app
A decade ago, I attended the Game Developers Conference in Austin, Texas. There were sessions on massively multiplayer online (MMO) games, animation for video games, avatar development, and artificial intelligence in video games. I remember thinking as I sat through one of the sessions – the highest and best use of characters in a game who can learn from conflict would be to develop “games” or tools for families to learn to resolve conflict and improve communication.
If I create an avatar that behaves like me at first, but learns better ways to navigate specific situations, I can learn to improve my game, i.e., my life. With the distance of “playing” myself, I gain perspective. I still think there’s great potential for emotional and social growth applications.
Here’s an idea for an app
A couple of months ago, I spoke to the Trauma Institute & Child Trauma Institute in Northampton, MA about a new progressive counting method used to treat PTSD and the effects of childhood trauma. Essentially, the patient verbalizes their first chronological memory of trauma while the therapist counts. This process continues until the distress associated with that memory is resolved. Then the patient moves on to the next distressing memory and repeats the process.
I’m sure it’s a little more complicated than that, but my first thought was – I wonder if you can make a phone count out loud? If so, it seems like a lot of this could be done with a smartphone. Maybe you’d do a couple of sessions with a practitioner at the beginning and periodically thereafter, but the rest could be done in the safety and comfort of your own home. We need an app for that.
Well researched and well designed apps have the potential to propel us forward. Whether or not they include my ideas, I’m excited about their incorporation into medical and mental health practices.
http://www.ori.org/scientists/lewis_goldberg
http://psycnet.apa.org/record/1970-12828-001
http://michaellewiswrites.com/#top
http://www.cooking2thrive.com/blog/make-it-easier-to-stick-to-your-eating-plan/
