Brainmatters | Computer beats human yet doesn’t win


When they want to generate research revenue, scientists can get quite creative. The first sponsoring requests for experiments have appeared on Kickstarter and Indiegogo already. Sometimes, scientists make up their own data. Chris Snijders aims for the big bucks via the App Store and Google Play.

It all started as a research assignment with colleague Ad Kleingeld of Industrial Engineering: the number of patients that met with dermatologists for suspect lesions that turned out to be harmless had to decrease drastically. Dermatologists blamed the family doctors, as they don't know how to diagnose properly. Enter two graduate students and their analyses. Soon, they had developed a working model that could pretty accurately predict both actinic keratosis and basal cell carcinoma, based on the answers to twelve questions concerning the lesion and the patient in question.

The fact these predictions were rather convincing is unsurprising to most engineers, who tend to believe in the power of data over that of intuition. And literature on the subject will tell you that computer models, provided with the same input data, hardly ever lose to human experts.

But we forgot about reality. Although we proved the use, need, and benefits of our project, it also proved impossible to implement it anywhere. Family doctors are too busy, dermatologists don’t believe the model could possibly do a better job (“that model doesn’t even get to see the patient”), heads of dermatology fear liability issues, and nurses are afraid to take responsibility for the program’s decisions. We should have known: even a landslide victory by the computer won’t have us break free from the dogged belief in our own abilities.

Yet, being the innovative researchers we are, we saw an opportunity. OddSpot is now available for iOS and Android through the App Store and Google Play. For 80 cents, which is about the price for half a cup of cappuccino in the cafeteria, it will provide you with the latest information - it’s better than your doctor. We’ve dubbed it fourth-party revenue research! Bring in the big bucks.

Chris Snijders is professor of Sociology of Technology and Innovation

Share this article