Tipping Points in Digital Pathology
“Social consensus through the influence of committed minorities” (DOI: 10.1103/PhysRevE.84.011130) is a fascinating read. In it, Xie et al. of the Rensselaer Polytechnic Institute take the well-known phenomenon of an inflection point in thinking – a tipping point past which a minority opinion will become the majority opinion – and use computational and analytic simulations to try to determine what that point might be. Their conclusion? That when the number of “true believers” reaches 10% of a given population, conversion of the rest of the population becomes inevitable. Below 10%, it would take far longer for such a conversion to take place.
Historically, there is some evidence to back this up: the article cites how the civil rights movement, for instance, only came into full fruition “shortly after the size of the African-American population crossed the 10% mark”. From our own history, we know that Joseph Lister and Ignaz Semmelweis, though they were correct about the important role of hand-washing in patient care, were not able to convince the majority of their peers on this matter. It took the work of Louis Pasteur in confirming the germ theory to sway the majority professional opinion, years later. More recently, when Barry Marshall and Robin Warren first discovered that gastritis and peptic ulcers can be caused by Helicobacter pylori, they were initially met with a great deal of skepticism; it was not until later that their discovery would be generally accepted as medical fact.
This article does have a few limitations – first of all, it utilizes a computational sociology model known as the Naming Game . In Xie et al.’s implementation of this model, “at each simulation time step, a randomly chosen speaker voices a random opinion from his list to a randomly chosen neighbor, designated the listener. If the listener has the spoken opinion in his list, both speaker and listener retain only that opinion, else the listener adds the spoken opinion to his list”. While this model has its advantages – computational simplicity among them – it has some drawbacks:
- It assumes that all interaction is strictly held in pairs.
- It does not attempt to place a hierarchy of trustworthiness on the population.
- It does not take into account the feasibility of an idea being expressed.
Compare this to the real world, in which:
- A large amount of interaction is not pair-based.
- Opinions carry different weight depending on who is articulating them.
- A given population generally has an idea of what is feasible and what is not.
Even with its flaws, this paper does make me ponder on our own field of pathology informatics. As we all know, ask ten different practicing pathologists about how technologies like whole-slide imaging will affect our profession, and you’ll get ten different answers. I just did a totally unscientific survey of residents (n = 7) and attending pathologists (n = 9); 5 of 7 residents believed that WSI is the way pathology will be practiced in the near future, as opposed to 1 in 9 pathologists. I wonder what a survey of a larger population of both trainee and attending pathologists would show.
How close are we to our digital pathology tipping point?
