News Intern Rachel Fritts talks with host Sarah Crespi about a new way to think about endometriosis—a painful condition found in one in 10 women in which tissue that normally lines the uterus grows on the outside of the uterus and can bind to other organs.
Next, Raphael Townshend, founder and CEO of Atomic AI, talks about predicting RNA folding using deep learning—a machine learning approach that relies on very few examples and limited data.
Finally, in this month’s edition of our limited series on race and science, guest host and journalist Angela Saini is joined by author Lundy Braun, professor of pathology and laboratory medicine and Africana studies at Brown University, to discuss her book: Breathing Race into the Machine: The Surprising Career of the Spirometer from Plantation to Genetics.
“It may be good for you,” says Dariush Mozaffarian, dean of the Friedman School of Nutrition Science and Policy at Tufts University. “I think we can say with good certainty it’s not bad for you.” (Additives are another story.)
After the link appeared between coffee intake and a reduced risk of heart failure in the Framingham data, Kao confirmed the result by using the algorithm to correctly predict the relationship between coffee intake and heart failure in two other respected data sets. Kosorok describes the approach as “thoughtful” and says that it “seems like pretty good evidence.”
Should you drink coffee? If so, how much? These seem like questions that a society able to create vaccines for a new respiratory virus within a year should have no trouble answering. And yet the scientific literature on coffee illustrates a frustration that readers, not to mention plenty of researchers, have with nutrition studies: The conclusions are always changing, and they frequently contradict one another.
Finding medicines that can kill cancer cells while leaving normal tissue unscathed is a Holy Grail of oncology research. In two new papers, scientists at UC San Francisco and Princeton University present complementary strategies to crack this problem with “smart” cell therapies—living medicines that remain inert unless triggered by combinations of proteins that only ever appear together in cancer cells.
Biological aspects of this general approach have been explored for several years in the laboratory of Wendell Lim, PhD, and colleagues in the UCSF Cell Design Initiative and National Cancer Institute– sponsored Center for Synthetic Immunology. But the new work adds a powerful new dimension to this work by combining cutting-edge therapeutic cell engineering with advanced computational methods.
For one paper, published September 23, 2020 in Cell Systems, members of Lim’s lab joined forces with the research group of computer scientist Olga G. Troyanskaya, PhD, of Princeton’s Lewis-Sigler Institute for Integrative Genomics and the Simons Foundation’s Flatiron Institute.
Using a machine learning approach, the team analyzed massive databases of thousands of proteins found in both cancer and normal cells. They then combed through millions of possible protein combinations to assemble a catalog of combinations that could be used to precisely target only cancer cells while leaving normal ones alone. In another paper, published in Science on November 27, 2020, Lim and colleagues then showed how this computationally derived protein data could be put to use to drive the design of effective and highly selective cell therapies for cancer.
“Currently, most cancer treatments, including CAR T cells, are told ‘block this,’ or ‘kill this,’” said Lim, also professor and chair of cellular and molecular pharmacology and a member of the UCSF Helen Diller Family Comprehensive Cancer Center. “We want to increase the nuance and sophistication of the decisions that a therapeutic cell makes.”
Over the past decade, chimeric antigen receptor (CAR) T cells have been in the spotlight as a powerful way to treat cancer. In CAR T cell therapy, immune system cells are taken from a patient’s blood, and manipulated in the laboratory to express a specific receptor that will recognize a very particular marker, or antigen, on cancer cells. While scientists have shown that CAR T cells can be quite effective, and sometimes curative, in blood cancers such as leukemia and lymphoma, so far the method hasn’t worked well in solid tumors, such as cancers of the breast, lung, or liver.
Cells in these solid cancers often share antigens with normal cells found in other tissues, which poses the risk that CAR T cells could have off-target effects by targeting healthy organs. Also, solid tumors also often create suppressive microenvironments that limit the efficacy of CAR T cells. For Lim, cells are akin to molecular computers that can sense their environment and then integrate that information to make decisions. Since solid tumors are more complex than blood cancers, “you have to make a more complex product” to fight them, he said.
From a “Circulation: Heart Failure” Journal study (Feb 25, 2020):
The study shows that wearable sensors coupled with machine learning analytics have predictive accuracy comparable to implanted devices.
We demonstrate that machine learning analytics using data from a wearable sensor can accurately predict hospitalization for heart failure exacerbation…at a median time of 6.5 days before the admission.
Heart failure (HF) is a major public health problem affecting >23 million patients worldwide. Hospitalization costs for HF represent 80% of costs attributed to HF care. Thus, accurate and timely detection of worsening HF could allow for interventions aimed at reducing the risk of HF admission.
Several such approaches have been tested. Tracking of daily weight, as recommended by current HF guidelines, did not lead to reduction of the risk of HF hospitalization, most likely because the weight gain is a contemporaneous or lagging indicator rather than a leading event. Interventions based on intrathoracic impedance monitoring also did not result in reduction of readmission risk. The results suggest that physiological parameters other than weight or intrathoracic impedance in isolation may be needed to detect HF decompensation in a timely manner. In fact, 28% reduction of rehospitalization rates has been shown with interventions based on pulmonary artery hemodynamic monitoring. More recently, in the MultiSENSE study (Multisensor Chronic Evaluation in Ambulatory HF Patients), an algorithm based on physiological data from sensors in the implantable cardiac resynchronization therapy defibrillators, was shown to have 70% sensitivity in predicting the risk of HF hospitalization or outpatient visit with intravenous therapies for worsening of HF.
This week, machine learning helps batteries charge faster, and using bacterial nanowires to generate electricity from thin air.
In this episode:
00:46 Better battery charging
A machine learning algorithm reveals how to quickly charge batteries without damaging them. Research Article: Attia et al.
07:12 Research Highlights
09:21 Harnessing humidity
A new device produces electricity using water in the air. Research Article: Liu et al.
16:30 News Chat
Coronavirus outbreak updates, the global push to conserve biodiversity, and radar reveals secrets in an ancient Egyptian tomb. News: Coronavirus: latest news on spreading infection; News: China takes centre stage in global biodiversity push
Researchers are using artificial intelligence techniques to invent medicines and materials—but in the process are they upending the scientific method itself? The AI approach is a form of trial-and-error at scale, or “radical empiricism”. But does AI-driven science uncover new answers that humans cannot understand? Host Kenneth Cukier finds out with James Field of LabGenius…