Tyler O'Neal, Staff Editor GOVERNMENT

UCI researchers' deep learning algorithm solves Rubik's Cube faster than any human

Work is step toward advanced AI systems that can think, reason, plan and make decisions

Since its invention by a Hungarian architect in 1974, the Rubik's Cube has furrowed the brows of many who have tried to solve it, but the 3D logic puzzle is no match for an artificial intelligence system created by researchers at the University of California, Irvine.

DeepCubeA, a deep reinforcement learning algorithm programmed by UCI computer scientists and mathematicians, can find the solution in a fraction of a second, without any specific domain knowledge or in-game coaching from humans. This is no simple task considering that the cube has completion paths numbering in the billions but only one goal state - each of six sides displaying a solid color - which apparently can't be found through random moves. {module In-article}

For a study published today in Nature Machine Intelligence, the researchers demonstrated that DeepCubeA solved 100 percent of all test configurations, finding the shortest path to the goal state about 60 percent of the time. The algorithm also works on other combinatorial games such as the sliding tile puzzle, Lights Out and Sokoban.

"Artificial intelligence can defeat the world's best human chess and Go players, but some of the more difficult puzzles, such as the Rubik's Cube, had not been solved by computers, so we thought they were open for AI approaches," said senior author Pierre Baldi, UCI Distinguished Professor of computer science. "The solution to the Rubik's Cube involves more symbolic, mathematical and abstract thinking, so a deep learning machine that can crack such a puzzle is getting closer to becoming a system that can think, reason, plan and make decisions."

The researchers were interested in understanding how and why the AI made its moves and how long it took to perfect its method. They started with a computer simulation of a completed puzzle and then scrambled the cube. Once the code was in place and running, DeepCubeA trained in isolation for two days, solving an increasingly difficult series of combinations.

"It learned on its own," Baldi noted.

There are some people, particularly teenagers, who can solve the Rubik's Cube in a hurry, but even they take about 50 moves.

"Our AI takes about 20 moves, most of the time solving it in the minimum number of steps," Baldi said. "Right there, you can see the strategy is different, so my best guess is that the AI's form of reasoning is completely different from a human's."

The veteran computer scientist said the ultimate goal of projects such as this one is to build the next generation of AI systems. Whether they know it or not, people are touched by artificial intelligence every day through apps such as Siri and Alexa and recommendation engines working behind the scenes of their favorite online services.

"But these systems are not really intelligent; they're brittle, and you can easily break or fool them," Baldi said. "How do we create advanced AI that is smarter, more robust and capable of reasoning, understanding and planning? This work is a step toward this hefty goal."

Rush unveils quality composite rank

Tyler O'Neal, Staff Editor GOVERNMENT

New composite rank aggregates hospital rating system scores into single, consumer-friendly score

Rush University Medical Center researchers have proposed a rating system that standardizes and combines data from five leading hospital rating systems into an easy-to-understand composite score of one to 10 that will help guide consumer's hospitals choice.

In a paper published July 2 in the American Journal of Medical Quality, the authors first cited research showing that despite almost two decades of public reporting of quality metrics, consumers have found hospital rating systems "to be limited and lacking in personalization or relevance for individual consumers." This lack of consumer engagement, the authors suggest, is likely driven by the substantial variability that exists between the ranking of top performing hospitals in different ranking systems: The U.S. News & World Report Best Hospitals List, the Vizient Quality and Accountability Study, the Centers for Medicare & Medicaid Services (CMS) Star Rating, the Leapfrog Hospital Safety Grade, and the Truven Top 100 Hospitals list. Rush University Medical Center's chief analytics officer Dr. Bala Hota is the lead author of 'Disagreement Between Hospital Rating Systems: Measuring the Correlation of Multiple Benchmarks and Developing a Quality Composite Rank'.{module In-article}

Lead author Dr. Bala Hota, the Medical Center's chief analytics officer, noted that while each of the rating organizations provides valuable data and insight that help drive hospital quality improvement efforts, their complexity and variability have made them difficult for consumers to use.

"The science behind each rating systems is very complex and measures different outcomes, domains and even time periods," Hota said. "And while this wealth of data supporting the ratings is vital to hospitals, consumers are confused when the ratings disagree."

Thus nearly two years ago, Hota and his Rush colleagues began gathering the data and assembling an objective framework to needed assess the overall similarity of rating systems to one another. The paper, "Disagreement Between Hospital Rating Systems: Measuring the Correlation of Multiple Benchmarks and Developing a Quality Composite Rank" details how they aggregated scoring data from multiple hospital ranking systems to generate a single measure, the Quality Composite Rank (QCR).

For the study, the scores for 70 high-performing hospitals ranked by the various raking systems were combined into a core data set of ten performance measures. Using a series of statistical correlation approaches that accounted for differences and similarities in what each rating organization measured, researchers were able to better identify variations and ultimately generate a single digit composite score that rewards hospitals for consistency across ratings systems.

"Standardizing what is measured more objectively identifies hospitals that do well in multiple measurement systems. Hospitals with the best QCR scores had higher quality scores across more areas and measured by more scoring systems. We believe that suggests a more sustained and institutional commitment to quality care," Hota said.

More importantly, the authors believe a single-digit QCR composite score built from the various ratings systems will benefit patients.

"The most important metrics are those that help patients navigate the health system. But publicly-reported quality measures that the public does not understand defeats their purpose," said Omar Lateef, DO, Rush University Medical Center chief executive officer and paper co-author.

"When patients see conflicting ratings, they must then reconcile that information in their mind. What we've done is to develop a measure that quantitatively does that reconciliation."

NYU’s Jennifer Crodelle develops model exploring daily rhythms in pain sensitivity

Tyler O'Neal, Staff Editor GOVERNMENT

Findings could open new paths toward better pain management strategies

A new computational model successfully predicts how daily pain sensitivity rhythms affect pain processing, both in healthy adults and in people with neuropathic pain. Jennifer Crodelle of New York University and colleagues present these findings in PLOS Computational Biology. {module In-article}g

Just as processes like metabolism and alertness exhibit a daily rhythm, pain sensitivity changes over the course of the day. Sensitivity is usually highest in the middle of the night and lowest in late afternoon. However, this rhythm is flipped for people with neuropathic pain, who feel severe pain in response to a typically non-painful stimulus. For these patients, the lowest pain sensitivity occurs at night.

The mechanisms underlying both normal and neuropathic pain rhythms have been unclear. To gain new insights, Crodelle and colleagues built a mathematical model that simulates how pain is transmitted from a nerve to the spinal cord's dorsal horn, where pain is initially processed.

The researchers found that their model successfully reproduces experimental results on pain sensitivity and predicts how these results are affected by time of day. For instance, it predicts the time-of-day effects on pain inhibition, the phenomenon in which one feels a lessening of pain from applying light pressure, such as by grabbing a stubbed toe.

The model also suggests a potential mechanism for the flipped sensitivity rhythm in people with neuropathic pain: a change from inhibition to excitation in the synaptic connections between nerve cells. This finding points to targets for further experimental study and potential treatment.

"Our modeling results provide a first step in understanding how the daily rhythm in pain sensitivity affects normal pain processing across the day and potentially how the daily rhythm can benefit pain management strategies in clinical settings," Crodelle says. "For example, pain relief medication could be titrated appropriately across the day, thus reducing the total amount of medication needed."

Potential next steps are to incorporate factors that may influence the daily pain sensitivity rhythm, such as sleep deprivation and jet lag. The model could also aid investigations into how pain sensitivity is reduced by a chronic pain treatment known as spinal cord stimulation.