Journalist Jeff Howe defined crowdsourcing as “The act of taking a job traditionally performed by a designated agent (usually an employee) and outsourcing it to an undefined, generally large group of people in the form of an open call.” And what’s the best way to get people to willingly respond to that open call and perform a (at times grueling) task? Make it a game.
And, as unlikely as it sounds, that’s exactly what medical researchers have done. It turns out that: A. people are very good at finding patterns and coming up with different ways to solve puzzles and B. two minds are better than one, and there are a lot of minds on the internet. That’s why games like Phylo and Biogames’ Telpathology game are growing in popularity while helping researchers solve critical problems.
Phylo is a web-based game that “harnesses the computing power of mankind to solve a common problem: Multiple Sequence Alignments.” Here’s what Phylo has players doing and how it’s helping researchers identify potentially lifesaving DNA patterns:
What is a Multiple Sequence Alignment?
A sequence alignment is a way of arranging the sequences of D.N.A, R.N.A or protein to identify regions of similarity. These similarities may be consequences of functional, structural, or evolutionary relationships between the sequences.
From such an alignment, biologists may infer shared evolutionary origins, identify functionally important sites, and illustrate mutation events. More importantly, biologists can trace the source of certain genetic diseases.
Traditionally, multiple sequence alignment algorithms use computationally complex heuristics [a trial and error form of problem solving] to align the sequences.
Unfortunately, the use of heuristics do not guarantee global optimization as it would be prohibitively computationally expensive to achieve an optimal alignment. This is due in part to the sheer size of the genome, which consists of roughly three billion base pairs, and the increasing computational complexity resulting from each additional sequence in an alignment.
Humans have evolved to recognize patterns and solve visual problems efficiently.
By abstracting multiple sequence alignment to manipulating patterns consisting of coloured shapes, we have adapted the problem to benefit from human capabilities.
By taking data which has already been aligned by a heuristic algorithm, we allow the user to optimize where the algorithm may have failed.
All alignments were generously made available through UCSC Genome Browser.
In fact, all alignments contain sections of human DNA which have been speculated to be linked to various genetic disorders, such as breast cancer.
Every alignment is received, analyzed, and stored in a database, where it will eventually be re-introduced back into the global alignment as an optimization.
To recap, there are billions upon billions DNA/RNA/Protein sequences across all the species on earth. Some of the sequences can be tied to diseases or similar sequences in other species. Computers put together as many of these sequences as they can, but nothing is ever perfect. So Phylo takes these sequences and turns them into easy on the eyes, color coded blocks. Players are presented with a group of sequences with the goal of making as many of the colors match up as possible. By keeping track of how each player solves these puzzles researchers come across sequences and inter-species connections that the computer may have missed. Presto, crowdsourced research and quality assurance!
To give the project a game-like feel that will keep players coming back Phylo features different levels of difficulty, let’s you return to a specific level and gives players the option of working on puzzles related to a particular type of disease, such as cancer, brain and nervous system ailments, or blood and immune system issues. Plus, by making the “top scores” visible the game’s creators tap into people’s competitive nature and ensure that players will continuously try to shorten and refine the sequences.
And Phylo isn’t alone in harnessing the power of the crowd to solve critical, high-level problems. PCWorld recently featured an article about a project that turns diagnosing malaria into a free online game. From PCWorld:
Researchers at UCLA have created an online crowdsourcing game designed to let players help doctors in key areas of the world speed the lengthy process of distinguishing malaria-infected red blood cells from healthy ones.
Typically, malaria is diagnosed by a trained pathologist peering through a conventional light microscope. The time consuming process can overwhelm researchers in countries that have high numbers of cases and limited resources, UCLA researchers said. …
The crowdsourcing game, which is free to play, works off the assumption that large groups of non-experts can be trained to recognize microscopic images of infectious disease cells with the accuracy of trained pathologists.
So far mostly undergraduate UCLA volunteers have played the game, and have collectively been able to accurately diagnose malaria-infected red blood cells within 1.25 percent of the accuracy of a pathologist performing the same task, resesarchers said.
“The idea is, if you carefully combine the decisions of people — even non-experts — they become very competitive,” said Aydogan Ozcan, a UCLA associate professor of electrical engineering and bioengineering and an author of the crowd-sourcing research. “One person’s response may be OK, but if you combine 10 to 20, or maybe 50 non-expert gamers together, you improve your accuracy greatly.” …
By training hundreds, and perhaps thousands, of game players to identify malaria, the UCLA crowdsourcing app could lead to rapid and close to accurate diagnoses at virtually no cost, the researchers said.
Read the full article at PCWorld >>>
Essentially, this is a ‘one of these things is not like the others’ game. Players are given a tutorial when they start that shows what a healthy red blood cell looks like and what a Malaria infected cell looks like. The game consists of flagging each of the infected cells in a frame. Once players think they’re done they declare the remaining cells healthy and move on to the next frame. The success rate of each player can also be tracked.
Within each frame, there are a certain number of cells whose status (i.e., infected or not) is known by the game but not by the players. These control cell images allow Ozcan’s team to dynamically estimate the performance of gamers as they go through each frame and also helps the team assign a score for every frame the gamer passes through.
Biogames references the fact that this technology can be used “regardless of the source and imaging modality” but I’m not sure how they currently or will eventually integrate real time images into game play (which currently presents individual cell images in a neat grid). However it works, they do believe that a global crowd of everyday Joes can present a reliable alternative to professional diagnoses in some situations.
We have shown that by utilizing the innate visual recognition and learning capabilities of human crowds it is possible to conduct reliable microscopic analysis of biomedical samples and make diagnostics decisions based on crowd-sourcing of microscopic data through intelligently designed and entertaining games that are interfaced with artificial learning and processing back-ends. We demonstrated that in the case of binary diagnostics decisions (e.g., infected vs. uninfected), using crowd-sourced games it is possible to approach the accuracy of medical experts in making such diagnoses.
Specifically, using non-professional gamers we report diagnosis of malaria infected red-blood-cells with an accuracy that is within 1.25% of the diagnostic decisions made by a trained professional.
Crowdsourced music videos and movies are cool, but using the combine problem solving abilities of the crowd (and sustaining it by making the work into a game) to solve hugely complex issues and maybe even save lives puts the power of crowdsourcing into a whole new perspective.