From artificial intelligence (A.I.) to materials science, this year’s Xconomy Awards finalists in the Innovation at the Intersection category are bringing a variety of disciplines outside of biology to bear on tough problems in life science. The hope is that advanced algorithms, novel biomaterials, and digital technologies will make drug discovery more efficient, cancer immunotherapies more effective, and wearable devices more beneficial for health and well-being. Here’s more on the finalists. (Click here to learn more about the CEO finalists and here for the Startup finalists.)
Angela Belcher, MIT
Angela Belcher’s molecular tool of choice has long been a virus that infects bacteria, called the M13 phage. The MIT materials scientist has engineered the virus so that it can grab onto nanomaterials to build better batteries and solar cells. In recent years, as a member of MIT’s Koch Institute for Integrative Cancer Research, she has turned her attention to cancer. She’s using the versatile virus to create new imaging agents that could one day help cancer surgeons find and remove small tumors that they currently can’t see during surgery.
Her lab has tweaked the M13 virus to bind to nanoparticles called carbon nanotubes that fluoresce in near-infrared light, which safely shines through human tissue. The virus also attaches to ovarian cancer cells, bringing with it the fluorescent nano-beacons that light up tumors when hit with near-infrared light, allowing surgeons to spot tumors and take them out.
Belcher has presented data showing that her technology has allowed surgeons to find tumors less than 1 millimeter in size in mice, and that lab animals that had their tumors removed with the help of the imaging technique lived 40 percent longer. Her lab is continuing to build better imaging equipment, has come up with other kinds of cancer-imaging nanoparticles, and is moving towards non-invasive imaging and screening for the early detection of ovarian cancer.
Iya Khalil, GNS Healthcare
Trained as a theoretical physicist, Iya Khalil co-founded GNS Healthcare in 2000 to apply mathematics and A.I. technologies to problems in the life sciences. Khalil, who is GNS’s chief commercial officer, co-developed the company’s machine learning algorithms that she and her team are unleashing on a variety of large data sets—genomic, molecular, clinical, medical record, and insurance claims data—from its pharmaceutical and health insurance partners. The goal is to tease out cause-and-effect relationships from the data. For drug makers, that means, for example, mathematically modeling how a drug interacts with a molecular pathway and how that ultimately affects patients, and then using that information to figure out which patients are most likely to benefit from a drug. The result is a biomarker that doctors can use to identify those patients.
GNS recently presented results from a project, in partnership with the Alliance for Clinical Trials in Oncology, that analyzed phase 3 clinical trial data from patients with advanced colorectal cancer who received one of two different treatments. GNS’s models showed that the location of the primary tumor can independently predict patient survival outcomes. They also identified potential biomarkers that might be used to categorize patients at the beginning of treatment according to how likely their cancers will grow.
David Mooney, Harvard University
Fourteen years ago, David Mooney made a fortuitous decision to focus the research in his Harvard University lab on developing biomaterials that rev up the immune system. At that time, clinical trials testing some of the first checkpoint inhibitors for cancer (which release the brakes on tumor-fighting T cells) were getting underway. Now that several of these drugs are on the market, it’s clear that they work for only a subset of cancer patients. Mooney thinks that the biomaterials that his lab has designed, when injected or implanted into the body, could help make these drugs more effective, by carrying key immune-stimulating molecules that further boost immune responses against cancer.
Mooney’s lab (also part of Harvard’s Wyss Institute for Biologically Inspired Engineering) has come up with a variety of materials—including a pill-sized implant that goes under the skin and injectable microparticles—using compounds that have already been used and proven safe in other drugs and medical devices. The biomaterials are all designed to work in a similar way: they form a three-dimensional, porous structure that can carry immune-stimulating molecules. These molecules call in immune cells, which crawl through the structures and are exposed to key antigens from tumors that are also embedded in the 3D structure. These antigens essentially provide the “instructions” to immune cells for which tumors they’re supposed to target. The biomaterials act as a cancer vaccine and are designed to work either on their own or in conjunction with checkpoint inhibitors.
One of Mooney’s materials is already being tested in humans. A phase 1 clinical trial, in 20 patients with late-stage melanoma, is being run by the Dana-Farber Cancer Institute and should wrap up this year. And earlier this year, Novartis (NYSE: NVS) agreed to license technology from Mooney’s lab, giving the pharma company rights to commercially develop the biomaterials for cancer immunotherapy.
When Nimbus Therapeutics was founded in 2009, it set out to use key principles of physics and computational chemistry to create software that designs small-molecule drugs. The idea is to build molecules on a computer first before making them in the lab—akin to the way engineers use software to design new buildings or machines before actually building them. This ‘in silico’ approach upends traditional methods of drug discovery, which involves screening large libraries of actual molecules first to see which ones might bind to the drug target.
What has allowed Nimbus to do this is a long-standing partnership with Schrodinger, a chemical software company that was a Nimbus co-founder. Schrodinger coders have come up with algorithms for Nimbus that allow the drug hunters to simulate how molecules interact with protein drug targets and predict which molecules will most likely fit best with the target. Nimbus says these tools allow its researchers to zero in on a shorter list of promising drug candidates than traditional methods, so that they can make fewer molecules in the lab for further testing, and home in on their top candidate faster.
The method has already paid off for Nimbus. Gilead Sciences (NASDAQ: GILD) bought Nimbus’s lead compound, GS-0976, for non-alcoholic steatohepatitis (NASH) in 2016 in a deal worth up to $1.2 billion, and the drug is in phase 2 testing. Another Nimbus program, a partnership with Celgene (NASDAQ: CELG) to develop a Tyk2 inhibitor for autoimmune disease, will soon enter clinical testing.
Now Nimbus is working with Schrodinger to do more of the design and refinement of molecules in silico, with the goal of further reducing the amount of molecular tweaking in the lab.
Rosalind Picard, MIT
MIT’s Rosalind Picard wants to build technology that improves people’s well-being, and to do this, she and her lab are designing devices to sense and recognize human emotions. She founded the field of affective computing, after writing a key paper and then a book in the mid-1990s. She argued that computers need to be able to sense and express emotion if they are to interact naturally with and benefit humans in health and many other areas. She predicted that wearable computers would have a major role in digital health long before wearables and digital health became the hot trends of today.
Those ideas have translated into commercial products, based on Picard’s research that has brought together computer science, A.I., psychology, neurology, and cognitive science. Picard co-founded two companies: Affectiva’s technology can detect emotion from people’s facial expressions and voice. And Empatica (an Xconomy Award finalist in the Digital Trailblazer category) got FDA approval earlier this year for its wristband device that detects signs of impending seizures in epilepsy patients and sends alerts to caregivers.
Some of Picard’s more recent research centers on autism. In a paper published last month, Picard and her collaborators developed machine learning software that was personalized to individual autistic children and could estimate their emotional state and level of engagement during therapy. Other research projects from her lab include finding ways to forecast depression before symptoms arrive and finding more objective measures of pain.