|
Penn Nursing Study: Is There a Digital Hood? Disadvantaged Youth Can’t Get Away from Negative Interactions, Whether on the Street or Online
Penn Study: Machine Learning at Arraignments Can Cut Repeat Domestic Violence
Penn Engineers Use Network Science to Predict How Ligaments Fail
Penn Nursing Study: Is There a Digital Hood? Disadvantaged Youth Can’t Get Away from Negative Interactions, Whether on the Street or Online
A new, novel study from the University of Pennsylvania School of Nursing (Penn Nursing) shows that there is an alarming connection between the negative social interactions disadvantaged youth experience in the neighborhoods they live in and on social media. The study, led by Robin Stevens, assistant professor in the department of family & community health and director of the Health Equity & Media Lab, is set for publication in New Media & Society.
The team conducted semi-structured interviews with 30 females and 30 males, ranging in age from 18-24 years old, about their social worlds and neighborhoods, both online and offline. The study took place in predominantly African-American and Hispanic neighborhoods. Forty-three percent of the participants were African American; 43 percent were Latino; and 13 percent were of mixed African-American and Latino ethnicity. All of the interviewees were in either high school or community college at the time.
“It is estimated that more than 75 percent of youth across the country are on some sort of social media,” said Dr. Stevens. “Teens and young adults who are at the margins of society may have experiences in dealing with social media that others don’t. Unfortunately, what we found was that not only do they have to deal with negative social interactions in their neighborhoods, those interactions also seep into their online lives, sometimes in a larger, more problematic, way.” The study’s findings reveal a dynamic and somewhat concerning interplay between a physical neighborhood and a digital neighborhood, where negative interactions are reproduced and amplified online.
The participants told interviewers of the drama that takes place on social media, which is a byproduct of living in a disadvantaged neighborhood. Researchers not only discovered that the physical negativity that these young people experience in their neighborhoods spills over to their lives on social media, but that the opposite is also true. “Participants told us that drama that starts out on social media can also manifest itself in serious, physical altercations on the streets,” added Dr. Stevens. “Social media is an amplified reflection of the real and digital neighborhoods in which they live.” The investigators suggest that more research is needed in the use and side effects of social media in diverse populations (i.e. cultural, financial and geographical).
The study team consisted of Stacia Gilliard-Matthews, assistant professor at Rutgers University; Jamie Dunaev, doctoral student at Rutgers University-Camden; Marcus Woods, elementary school teacher in New Orleans, Louisiana; and Bridgette Brawner, assistant professor at Penn Nursing.
Penn Study: Machine Learning at Arraignments Can Cut Repeat Domestic Violence
In one large metropolitan area, arraignment decisions made with the assistance of machine learning cut new domestic violence incidents by half, leading to more than 1,000 fewer such post-arraignment arrests annually, according to new findings from Penn. In the US, the typical pre-trial process proceeds from arrest to preliminary arraignment to a mandatory court appearance, when appropriate. During the preliminary arraignment, a judge or magistrate chooses whether to release or detain the suspect, a decision intended to account for the likelihood that the person will return to court or commit new crimes. This is especially important in domestic violence, which is often a serial offense and directed at a particular individual.
Richard Berk, a criminology and statistics professor in Penn’s School of Arts & Sciences and Wharton School, and Susan B. Sorenson, a professor of social policy in Penn’s School of Social Policy & Practice, found that using machine-learning forecasts at these proceedings can dramatically reduce subsequent domestic violence arrests. “A large number of criminal justice decisions by law require projections of the risk to society. These threats are called ‘future dangerousness,’” Dr. Berk said. “Many decisions, like arraignments, are kind of seat of the pants. The question is whether we can do better than that, and the answer is yes we can. It’s a very low bar.”
For domestic violence crimes between intimate partners, parents and children or even siblings, there’s typically a threat to one particular person, said Dr. Sorenson, who directs Penn’s Evelyn Jacobs Ortner Center on Family Violence. “It’s not a general public safety issue,” she said. “With a domestic violence charge, let’s say a guy—and it usually is a guy—is arrested for this and is awaiting trial. He’s not going to go assault some random woman. The risk is for a re-assault of the same victim.”
To understand how machine learning could help in domestic violence cases, Drs. Berk and Sorenson obtained data from more than 28,000 domestic violence arraignments between January 2007 and October 2011. They also looked at a two-year follow-up period after release that ended in October 2013. They published their work in the March issue of The Journal of Empirical Legal Studies.
A computer can “learn” from training data which kinds of individuals are likely to re-offend. For this research, the 35 initial inputs included age, gender, prior warrants and sentences and even residential location. These data points help the computer understand appropriate associations for projected risk, offering extra information to a court official deciding whether to release an offender. “In all kinds of settings, having the computer figure this out is better than having us figure it out,” Dr. Berk said.
That’s not to say there aren’t obstacles to its use. The number of mistaken predictions can be unacceptably high, and some people object in principle to using data and computers in this manner. To both of these points, the researchers respond that machine learning is simply a tool. “It doesn’t make the decisions for people by any stretch,” Dr. Sorenson said. These choices “might be informed by the wisdom that accrues over years of experience, but it’s also wisdom that has accrued only in that courtroom. Machine learning goes beyond one courtroom to a wider community.”
“The algorithms are not perfect. They have flaws, but there are increasing data to show that they have fewer flaws than existing ways we make these decisions,” Dr. Berk said. “You can criticize them—and you should because we can always make them better—but, as we say, you can’t let the perfect be the enemy of the good.”
Penn Engineers Use Network Science to Predict How Ligaments Fail
When doctors diagnose a torn ligament, it’s usually because they can see ruptures in the ligament’s collagen fibers, visible on a variety of different scans. However, they also often treat patients with many of the symptoms of a tear, but whose ligaments don’t show this kind of damage. Researchers from Penn’s School of Engineering & Applied Science are using network science to gain new insights into these “subfailure” injuries, which can lead to pain and dysfunction despite the lack of obvious physical evidence. The mechanisms that lead to these symptoms happen on a microscopic level and can’t be detected by existing clinical tools.
In a study recently published in the Journal of the Royal Society Interface, the Penn team has put human ligament samples to the test, stretching them until they tear, while looking at these microscopic features. Using a polarized-light-based system that can reveal the angles of collagen fibers in the tissue, the researchers have shown how groups of neighboring fibers changing their orientations in tandem prefigures the spots where failure eventually occurs. These insights could help identify regions of ligaments that are prone to tearing, and could eventually be incorporated into new diagnostic techniques or therapies. They could also help explain the painful symptoms patients experience in sub-failure injuries.
The study was conducted by Beth Winkelstein, professor in the departments of bioengineering in Penn Engineering and neurosurgery in Penn’s Perelman School of Medicine; Sijia Zhang, a graduate student in the Winkelstein lab; and Danielle Bassett, the Skirkanich Assistant Professor of Innovation, with appointments in the departments of bioengineering and electrical & systems engineering.
Earlier work in the Winkelstein lab showed microscopic evidence of the first tears appearing in a ligament as it was put under strain. These visible ruptures are often initiated by disorganization of ligament fibers amounted to a few pixels on an optical scan, so Ms. Zhang was interested in adding more context to the picture. “We’re investigating the mechanisms of how injury, even if it isn’t visible, induces pain,” she said. “Our hypothesis was that the cells embedded in the collagen matrix are being stretched during ligament loading, affecting cell behaviors, so we set out to see how the matrix is being reorganized under strain.”
Ms. Zhang was enrolled in Dr. Bassett’s introductory class on network science, in which students are encouraged to bring their own datasets to serve as illustrative examples. Her data were obtained from experiments in the Winkelstein lab during which ligament samples were stretched until failure while they were observed with an imaging system using polarized light. Much like how polarized sunglasses work by blocking all light aligned at a particular angle, this system can show the orientations of collagen fibers in the ligament by measuring how much light they allow through. “With this method,” Ms. Zhang said, “you can track changes in collagen fiber orientation during loading, and you can measure the angles of the collagen fibers and tell how strongly they are aligned.”
Network science investigates how individual elements of complex systems interact to determine the system’s overall behavior. By using analysis techniques derived from this discipline, Ms. Zhang was able to show the degree to which concerted reorientation prefigured the spots where failure first occurred.
“Network science offers a fundamental explanatory mechanism for subfailure damage, a process that we think may lead to pain,” Dr. Bassett said. “If a single fiber is turning, a tear is unlikely, as is the activation of pain fibers; but when there is a coordinated change in many fibers, pain and tears may be more likely.” |
