Tyler O'Neal, Staff Editor ASTRONOMY

Sex-specific adverse drug effects identified by Columbia University algorithm

There is a paucity of real-world clinical data that evaluates adverse drug effects in women, among other underserved populations, due to a long history of trials done on relatively homogenous patient populations (healthy white males).

Without heterogeneous data availability, biased results leave women in the dangerous position of not having accurate information on adverse drug effects, currently the fourth-leading cause of death in the U.S.

An example of this issue is Ambien, an insomnia drug that previously had the same dosage prescribed for both men and women. When evidence appeared that women were having a significantly greater rate of adverse reactions the following morning, the FDA reduced the recommended dosage by half in 2013.

"Rather than take the stance that we wait for evidence to become so overwhelming that we have to do something about, we wanted to be more proactive," said Nicholas Tatonetti, Ph.D., a Columbia University researcher who co-led a study in Patterns that uses machine learning to identify these adverse effects in women. "We want to use databases like the Adverse Event Reporting System (FAERS) from the FDA or the electronic health records to get a jump on identifying sex-specific adverse events before it's too late." {module INSIDE STORY}

Tatonetti, an associate professor in the Department of Biomedical Informatics, collaborated with Payal Chandak, an undergraduate student in Columbia's Department of Computer Science, to develop AwareDX (Analysing Women At Risk for Experiencing Drug toxicity), an algorithm that leverages advances in machine learning to predict sex risks.

"We developed a machine-learning framework to data-mine for sex-specific adverse events," Tatonetti said. "We went through hundreds of thousands of hypotheses and evaluated them. Payal designed a system that addresses confounding biases because it's very difficult to study these effects because some drugs or effects are more common in either women or men. We invented a technique that mitigates the confounding biases, develops a statistical basis to identify sex difference in adverse effects, and rank them by the strength of that evidence."

An example of findings from this algorithm (see graphic, right), believed to be the first validated approach for predicting sex risks, is the confirmation that a single gene (ABCB1) can pose different risks for men (from simvastatin) and women (from risperidone). Overall, this resource includes 20,817 adverse drug effects posing sex-specific risks, and it presents an opportunity to minimize adverse events by tailoring drug prescription and dosage to sex.

"We were motivated by the lack of information across different communities on the efficacy and safety of drug effects," Tatonetti said. "We addressed that issue in this study specifically for women, who hold greater risk for these adverse effects than men due to differences in pharmacokinetics and pharmacodynamics. This specific knowledge can impact guidelines about drug prescription and dosage and creating safer, healthier conditions for women."

University of Tokyo uses supercomputer simulations for a clearer view of what makes glass rigid

Tyler O'Neal, Staff Editor ASTRONOMY

Researchers led by The University of Tokyo employed a new supercomputer model to simulate the networks of force-carrying particles that give amorphous solids their strength even though they lack long-range order. This work may lead to new advances in high-strength glass, which can be used for cooking, industrial, and smartphone applications.

Amorphous solids such as glass--despite being brittle and having constituent particles that do not form ordered lattices--can possess surprising strength and rigidity. This is even more unexpected because amorphous systems also suffer from large anharmonic fluctuations. The secret is an internal network of force-bearing particles that span the entire solid which lends strength to the system. This branching, dynamic network acts like a skeleton that prevents the material from yielding to stress even though it makes up only a small fraction of the total particles. However, this network only forms after a "percolation transition" when the number of force-bearing particles exceeds a critical threshold. As the density of these particles increases, the probability that a percolating network that goes from one end to the other increases from zero to almost certain. A team of scientists led by the University of Tokyo uses supercomputer simulations to study the rigidity of amorphous solids like glass {module INSIDE STORY}

Now, scientists from the Institute of Industrial Science at The University of Tokyo have used computer simulations to carefully show the formation of these percolating networks as an amorphous material is cooled below its glass transition temperature. In these calculations, binary particle mixtures were modelled with finite-range repulsive potentials. The team found that the strength of amorphous materials is an emergent property caused by the self-organization of the disordered mechanical architecture.

"At zero temperature, a jammed system will show long-range correlations in stress due to its internal percolating network. This simulation showed that the same is true for glass even before it has completely cooled," first author Hua Tong says.

The force-bearing backbone can be identified by recognizing that particles in this network are must be connected by at least two strong force bonds. Upon cooling, the number of force-bearing particles increases, until a system-spanning network links together.

"Our findings may open up a way towards a better understanding of amorphous solids from a mechanical perspective," senior author Hajime Tanaka says. Since rigid, durable glass is highly prized for smartphones, tablets, and cookware, the work can find many practical uses.

US Army grant helps UH researchers develop techniques to stop cyber attackers in their tracks

Tyler O'Neal, Staff Editor ASTRONOMY

The COVID-19 pandemic and the presidential election have led to a significant increase in cyberattacks via email, text message and social media aimed at stealing personal information and destroying vital data. To help stop these hackers before they strike, computer science researchers at the University of Houston have been awarded a three-year, $660,000 grant from the U.S. Army Research Office.

Rakesh Verma, a computer science professor at the UH College of Natural Sciences and Mathematics and co-principal investigator, said his research team will go beyond commonly-used cyber defense techniques such as honeypots or moving target defense, both focused on fooling the hacker by mimicking likely targets of attacks or increasing uncertainty and complexity.

"Instead, we will generate new attacks of our own. We want to be proactive rather than reactive," said Verma. "Cybercriminals are getting more creative at each turn, so the idea is to be one step ahead of any type of attack." Arjun Mukherjee is principal investigator of the U.S. Army Research Office grant and associate professor of computer science at the UH College of Natural Sciences and Mathematics.{module INSIDE STORY}

The team of postdoctoral, graduate, and undergraduate researchers will design machine learning and natural language processing techniques -- a computer system's ability to read and understand spoken or written language -- that can produce unlimited, open-ended attacks. The goal is to generate novel attacks on a daily basis using adversarial machine learning to help develop new, ingenious filters to ward off those attacks.

"We want to close the loop by subjecting our detectors to these new attacks, so the detectors are continuously learning and improving themselves rather than passively waiting for attacks," said Arjun Mukherjee, principal investigator and associate professor of computer science. "We will compare our techniques against state-of-the-art baselines on diverse datasets in realistic scenarios."

A recent assessment by the International Criminal Police Organization (INTERPOL) on the impact of COVID-19 on cybercrime showed a significant target shift from individuals and small businesses to major corporations, governments and critical infrastructure. The agency projects a further increase in cybercrime in the future.

"Vulnerabilities related to working from home and the potential for increased financial benefit will see cybercriminals continue to ramp up their activities and develop more advanced and sophisticated modi operandi," according to the report.

Verma believes at some point online criminals will start to use machine learning and natural language processing to produce cyberattacks.

"We want to have that cycle of automatic improvement, so we create our own attacks and subject our filtering methods to the new attack and see if the new attacks are successful," Verma explained. "If they are unsuccessful, then we will try to generate better attacks and if they are successful, we will work to improve our filters."