Georgia Tech develops multi-algorithm approach that helps deliver personalized medicine for cancer patients

Today, machine learning, artificial intelligence, and algorithmic advancements made by research scientists and engineers are driving more targeted medical therapies through the power of prediction. The ability to rapidly analyze large amounts of complex data has clinicians closer to providing individualized treatments for patients, intending to create better outcomes through more proactive, personalized medicine and care.  Ovarian Cancer Cells

“In medicine, we need to be able to make predictions,” said John F. McDonald, professor in the School of Biological Sciences and director of the Integrated Cancer Research Center in the Petit Institute for Bioengineering and Bioscience at the Georgia Institute of Technology. One way is through understanding cause and reflecting relationships, as a cancer patient’s response to drugs, he explained. The other way is through correlation. 

“In analyzing complex datasets in cancer biology, we can use machine learning, which is simply a sophisticated way to look for correlations. The advantage is that computers can look for these correlations in extremely large and complex data sets.”

Now, McDonald’s team and the Ovarian Cancer Institute are using ensemble-based machine learning algorithms to predict how patients will respond to cancer-fighting drugs with high accuracy rates. The results of their most recent work have been published in the Journal of Oncology Research.  

For the study, McDonald and his colleagues developed predictive machine learning-based models for 15 distinct cancer types, using data from 499 independent cell lines provided by the National Cancer Institute. Those models were then validated against a clinical dataset containing seven chemotherapeutic drugs, administered either singularly or in combination, to 23 ovarian cancer patients. The researchers found an overall predictive accuracy of 91%.

“While additional validation will need to be carried out using larger numbers of patients with multiple types of cancer,” McDonald noted, “our preliminary finding of 90% accuracy in the prediction of drug responses in ovarian cancer patients is extremely promising and gives me hope that the days of being able to accurately predict optimal cancer drug therapies for individual patients is in sight."

The study was conducted in collaboration with the Ovarian Cancer Institute (OCI) in Atlanta, where McDonald serves as a chief research officer. Other authors are Benedict Benigno, MD (OCI founder and chief executive officer, as well as an obstetrician-gynecologist, surgeon, and oncologist); Nick Housley, a postdoctoral researcher in McDonald’s Georgia Tech lab; and the paper’s lead author, Jai Lanka, an intern with OCI. 

The challenges in predicting cancer treatments

The complex nature of cancer makes it a challenging problem when it comes to predicting drug responses, McDonald said. Patients with the same type of cancer will often respond differently to the same treatment. 

“Part of the problem is that the cancer cell is a highly integrated network of pathways and patient tumors that display the same characteristics clinically may be quite different on the molecular level,” he explained. 

A major goal of personalized cancer medicine is to accurately predict likely responses to drug treatments based upon genomic profiles of individual patient tumors. 

“In our approach, we utilize an ensemble of machine learning methods to build predictive algorithms — based on correlations between gene expression profiles of cancer cell lines or patient tumors with previously observed responses — to a variety of cancer drugs. The future goal is that gene expression profiles of tumor biopsies can be fed into the algorithms, and likely patient responses to different drug therapies can be predicted with high accuracy,” said McDonald.   

Machine learning is already being applied to the data coming from the genomic profiles of tumor biopsies, but before the researchers’ work, these methods have typically involved a single algorithmic approach. 

McDonald and his team decided to combine several algorithm approaches that use multiple ways to analyze complex data; one even uses a three-dimensional approach. They found using this ensemble-based approach significantly boosted predictive accuracy.

The algorithms the team used have names like Support Vector Machines (SVM), Random Forest classifier (RF), K-Nearest Neighbor classifier (KNN), and Logistic Regression classifier (LR). 

“They’re all fairly technical, and they’re all different computational mathematical approaches, and all of them are looking for correlations,” said McDonald. “It’s just a question of which one to use, and for different data sets, we find that one model might work better than another.”

However, more patient datasets that combine genomic profiles with responses to cancer drugs are needed to advance the research.  

“If we want to have a clinical impact, we must validate our models using data from a large number of patients,” said McDonald, who added that many datasets are held by pharmaceutical companies who use them in drug development. That data is typically considered proprietary, private information. And although a significant amount of genomic data of cancer patients is generally available, it’s not typically correlated with patient responses to drugs.

McDonald is currently talking with medical insurance companies about access to relevant datasets, as well. “It costs insurance companies a significant amount of money to pay for drug treatments that don’t work,” he noted. Time, medical fees, and ultimately, many lives could be saved by providing researchers with these types of information. 

“Right now, a percentage of patients will not respond to a drug, but we don’t know that until after six weeks of chemotherapy,” said McDonald. “What we hope is that we will soon have tools that can accurately predict the probability of a patient responding to first-line therapies — and if they don’t respond, to be able to make accurate predictions as to the next drug to be tried.”

Manchester scientists take a step towards detecting nanohertz gravitational-wave background

The European Pulsar Timing Array (EPTA) is a scientific collaboration bringing together teams of astronomers around the largest European radio telescopes, as well as groups specialized in data analysis and supercomputer modeling of gravitational wave (GW) signals. 1920 eptagravitationalwaveback

The international research team has a detailed analysis of a candidate signal for the since-long sought gravitational wave background (GWB) due to in-spiraling supermassive black-hole binaries. Although a detection cannot be claimed yet, this represents another significant step in the effort to finally unveil GWs at very low frequencies, of order one billionth of a Hertz.

The candidate signal has emerged from an unprecedented detailed analysis and using two independent methodologies. Moreover, the signal shares strong similarities with those found from the analyses of other teams.

Dr. Michael Keith of The University of Manchester said: “For the last 20 years or so we have been trying to detect the gravitational waves produced by supermassive black holes in the centers of distant galaxies. Although these waves are very tiny - nanosecond fluctuations over tens of years, the detection of these waves have implications for the formation of all galaxies, including our own Milky Way.

“So far nobody has detected these waves, but we have found an intriguing signal in the data that matches some, but not all, of the properties of the gravitational wave signal, we are looking for. The paper presents the data and some of the extensive range of tests we have done to support the hypothesis that the observed signal is from ultra-low frequency gravitational waves passing over the earth.”

The results were made possible thanks to the data collected over 24 years with five large-aperture radio telescopes in Europe. They include; the world-renowned Lovell Telescope at The University of Manchester’s Jodrell Bank, MPIfR’s 100-m Radio Telescope near Effelsberg in Germany, the 94-m Nançay Decimetric Radio Telescope in France, the 64-m Sardinia Radio Telescope at Pranu Sanguni, Italy, and the 16 antennas of the Westerbork Synthesis Radio Telescope in the Netherlands. In the observing mode of the Large European Array for Pulsars (LEAP), the EPTA telescopes are tied together to synthesize a fully steerable 200-m dish to greatly enhance the sensitivity of the EPTA towards gravitational waves.

Radiation beams from the pulsars’ magnetic poles circle their rotational axes, and we observe them as pulses when they pass our line of sight, like the light of a distant lighthouse. Pulsar timing arrays (PTAs) are networks of very stably rotating pulsars, used as galactic-scale GW detectors. In particular, they are sensitive to very low-frequency GWs in the billionth-of-a-Hertz regime. This will extend the GW observing window from the high frequencies (hundreds of Hertz) currently observed by the ground-based detectors LIGO/Virgo/KAGRA.

While those detectors probe short-lasting collisions of stellar-mass black holes and neutron stars, PTAs can probe GWs such as those emitted by systems of slowly in-spiraling supermassive black-hole binaries hosted at the centers of galaxies. The addition of the GWs released from a cosmic population of these binaries forms a GWB.

The small fluctuations in the arrival times of the pulsars’ radio signal at Earth can be measured, caused by the spacetime deformation due to passing-by very-low-frequency gravitational waves. In practice, these deformations manifest as sources of very low-frequency noise in the series of the observed times of arrival of the pulses, a noise which is shared by all the pulsars of a pulsar timing array.

However, the amplitude of this noise is incredibly tiny (estimated to be tens to a couple of hundreds of a billionth of a second) and in principle, many other effects could impart that to any given pulsar in the PTA.

To validate the results, multiple independent codes with different statistical frameworks were then used to mitigate alternate sources of noise and search for the GWB. Importantly, two independent end-to-end procedures were used in the analysis for cross-consistency. Additionally, three independent methods were used to account for possible systematics in the Solar-system planetary parameters used in the models predicting the pulse arrival times, a prime candidate for false-positive GW signals.

The EPTA analysis with both procedures found a clear candidate signal for a GWB and its spectral properties (i.e. how the amplitude of the observed noise varies with its frequency) remain within theoretical expectations for the noise attributable to a GWB.

Dr. Nicolas Caballero, a researcher at the Kavli Institute for Astronomy and Astrophysics in Beijing and co-lead author explains: “The EPTA first found indications for this signal in their previously published data set in 2015, but as the results had larger statistical uncertainties, they were only strictly discussed as upper limits. Our new data now clearly confirm the presence of this signal, making it a candidate for a GWB."

The Ripple Factor: Economic losses from weather extremes can amplify each other across the world

Weather extremes can cause economic ripples along our supply chains. If they occur at roughly the same time the ripples start interacting and can amplify even if they occur at completely different places around the world, a new study shows. The resulting economic losses are greater than the sum of the initial events, the researchers find in supercomputer simulations of the global economic network. Rich economies are affected much stronger than poor ones, according to the calculations. Currently, weather extremes around the world are increasing due to greenhouse gas emissions from burning fossil fuels. If they happen simultaneously or in quick succession even at different places on the planet, their economic repercussions can become much bigger than previously thought. Photo by Ian Taylor on Unsplash

“Ripple resonance, as we call it, might become key in assessing economic climate impacts especially in the future,” says Kilian Kuhla from the Potsdam Institute for Climate Impact Research, the first author of the study. “The effect of weather extremes in our globalized economy yield losses in some regions that face supply shortages and gains in others that see increased demand and thereby higher prices. But when extremes overlap economic losses in the entire global supply network are on average  20 percent higher. This is what we see in our simulations of heat stress events, river floodings, and tropical cyclones; and it is a most worrying insight.”

Generally, extreme weather leading to, for example, the flooding of a factory does lead not only to direct local output losses. It is known that the economic shocks also propagate in the global trade network. Now the researchers find that these propagated effects do not just add up but can amplify each other. The researchers modeled the response of the global network, calculating 1.8 Million economic relations between more than 7000 regional economic sectors.

Richer economies are hit harder

While not all countries suffer from the ripple resonance effect, most countries that are economically relevant do. Specifically China, due to its prominent position in the world economy, shows an above-average effect of more than 27% of extra losses when extreme events overlap compared to when they hit independently from each other.

“The phenomenon of economic ripple resonance means that two separate incidents send shock waves through the world economy, and those waves build up – like a tidal wave,” says Anders Levermann department head at Potsdam Institute and scientist at Columbia University in New York, who led the author team. “Supply shortages increase the demand and that increases the prices. Firms have to pay more for their production goods.  In most cases, this will get passed down to the consumer. Since weather extremes happen abruptly, there’s no smooth adaptation of capacities and prices at least for a short period of time. If other suppliers fail, due to economic repercussions of another weather extreme elsewhere, the interfering price shocks are intensified.”

The overlap makes total losses larger than the sum of two events’ damages

“If something gets rare, it gets expensive, and if it gets rare worldwide it gets very expensive – clearly, that’s not new,” says Levermann. “The new thing is the overlap. So far, people mostly looked at the local damage or at most the economic repercussions of one disaster at a time. Now we find that a second disaster happening at about the same time, even if it’s in a different corner of the world, can lead to higher worldwide economic losses.”

This holds not just for simultaneous but also for consecutive disasters if the economic effects of the different disasters overlap. “By allowing climate change to run wild, we add climate-induced economic losses on top of everything else. If we do not rapidly reduce greenhouse gases, this will cost us – even more than we’ve expected so far.”