Tyler O'Neal, Staff Editor BIOLOGY

Owkin, Cleveland Clinic develop artificial intelligence model that predicts liver cancer prognosis

Findings presented at European Society of Medical Oncology

Owkin, a startup that deploys artificial intelligence (AI) and Federated Learning technologies to augment medical research and enable scientific discoveries, presented findings in Hepatocellular Carcinoma (HCC) with Cleveland Clinic at the 2021 European Society of Medical Oncology (ESMO) conference. The collaborative research leverages Cleveland Clinic’s high-quality datasets and world-class medical research and Owkin’s pioneering technologies and research platform.

Cleveland Clinic researchers, led by Federico Aucejo, MD, Director of the Liver Cancer Program and Surgical Director of the Liver Tumor Clinic, have collaborated with Owkin to develop and validate a deep learning model that predicts survival and outcome of HCC patients following liver transplantation. This was performed on 298 whole-slide images stained with hematoxylin/eosin and clinical data from HCC patients who had a liver transplant at Cleveland Clinic. 

The findings showed the deep learning model trained on histopathology data predicted recurrence among transplant patients both in the whole cohort and in subgroups of patients treated with or without loco-regional therapy prior to transplantation. These results were comparable to a separate model that incorporated clinical, biological, and pathological data. Most significantly, combinations of both histological and clinical models outperformed scoring systems currently used in the literature. Taken together, this study demonstrates the prognostic power of deep learning applied to histology slides to predict the recurrence of HCC patients following liver transplantation.

“Machine learning technology is emerging to revolutionize the world we live in. Its application in patient populations, risk stratification, and personalized medicine is expected to enhance safety and allow for a more cost-effective healthcare environment. In line with this, partnerships and alliances among healthcare networks and the tech industry will be instrumental to paving the way towards this paradigm change,” Dr. Aucejo said. “This collaboration resulted in the development of an algorithm to predict outcome in patients undergoing liver transplantation with HCC by scrutinizing histopathology digital slides. This approach proved to be superior to predict tumor recurrence than conventional metrics.”

“Our collaborative research aims to advance the prediction of HCC patient outcomes and identify prognostic markers following treatment. The richness and uniqueness of Cleveland Clinic’s research cohorts, together with Owkin’s extensive expertise in developing predictive AI models, can pave the way for a breakthrough, forward-thinking science and will allow the opportunity to further develop our collaboration in future research areas,” Meriem Sefta, Ph.D., Chief Data and Clinical Solutions said.

HCC is amongst the leading causes of cancer-related deaths in the world, representing ~90% of primary liver cancers. Currently, liver transplantation remains the best treatment for cirrhotic patients with early-stage HCC, however, tumor recurrence following liver transplant is observed in 15-20% of cases, which correlates with poor survivorship. Moreover, there are currently no reliable histological markers of relapse-free survival in HCC patients following a liver transplant, which is critical in predicting patient prognosis.

Building on these results, additional deep-learning models and multimodal models developed on medical imaging, molecular, and genomics data, in addition to clinical and histopathological data, will shed further insights into diagnostic and biomarkers that may predict HCC prognosis and survivorship following treatment to improve patient care and long-term outcomes.

BU Medicine wins $2.1 million NIH grant to further study arterial disease

Tyler O'Neal, Staff Editor BIOLOGY

The study will focus on next-generation devices to prevent limb loss

Peripheral artery disease (PAD) is a major cause of limb loss. It is estimated that PAD affects between 8.5 and 12 million Americans, with a prevalence that has increased by about 25 percent over the preceding decade.

Vijaya B. Kolachalama, Ph.D., FAHA, assistant professor of medicine at Boston University School of Medicine (BUSM), and his colleagues aim to address this issue. He has been awarded a $2.1 million R01 grant from the National Institutes of Health’s (NIH) National Heart, Lung, and Blood Institute. Using machine learning, image processing, physics-driven modeling as well as experimental and animal studies, Kolachalama and his colleagues will establish algorithms and models to understand the mechanisms of drug delivery and optimize device design for treating PAD.

While interventional devices such as drug-coated balloons (DCBs) are effective at treating PAD, recent studies suggest the potential for DCBs to cause harm. This has prompted the FDA to issue a warning that ultimately led to a marked reduction of the clinical use of DCBs.

According to Kolachalama, this response by the clinical and regulatory communities underscores a need to develop next-generation DCBs that could show improved efficacy and safety profiles. “Drawing from our previous experience related to studies on drug-eluting stents and more recently on DCBs, we will engineer the next generation of DCBs with improved efficacy and safety profiles to help restore DCB treatment options for patients with PAD,” he explains.

The researchers will seek to predict optimal DCB designs for both acute and sustained drug delivery using a model that computes mechanical interactions during DCB use.

The proposed project builds upon a history of collaboration between BUSM, Massachusetts General Hospital, and the University of South Carolina in the targeted area of translational research in cardiovascular disease. “It constitutes a novel and timely study on an emergent therapeutic approach and includes an aspect of mechanistic discovery that holds relevance in the broader domain of endovascular technologies,” adds Kolachalama.

A part of the U.S. Department of Health and Human Services, the NIH is the largest biomedical research agency in the world.

Liverpool researchers develop AI tool that accelerates discovery of new materials

Tyler O'Neal, Staff Editor BIOLOGY

Researchers at the University of Liverpool have created a collaborative artificial intelligence tool that reduces the time and effort required to discover truly new materials.

The new tool has already led to the discovery of four new materials including a new family of solid-state materials that conduct lithium. Such solid electrolytes will be key to the development of solid-state batteries offering a longer range and increased safety for electric vehicles. Further promising materials are in development. New materials

The tool brings together artificial intelligence with human knowledge to prioritize those parts of unexplored chemical space where new functional materials are most likely to be found.

Discovering new functional materials is a high-risk, complex, and often long journey as there is an infinite space of possible materials accessible by combining all of the elements in the periodic table, and it is not known where new materials exist.

The new AI tool was developed by a team of researchers from the University of Liverpool’s Department of Chemistry and Materials Innovation Factory, led by Professor Matt Rosseinsky, to address this challenge.

The tool examines the relationships between known materials at a scale unachievable by humans. These relationships are used to identify and numerically rank combinations of elements that are likely to form new materials. The rankings are used by scientists to guide the exploration of the large unknown chemical space in a targeted way, making experimental investigation far more efficient. Those scientists make the final decisions, informed by the different perspectives offered by AI.

The lead scientist Professor Matt Rosseinsky said: “To date, a common and powerful approach has been to design new materials by close analogy with existing ones, but this often leads to materials that are similar to ones we already have.

“We, therefore, need new tools that reduce the time and effort required to discover truly new materials, such as the one developed here that combines artificial intelligence and human intelligence to get the best of both.

“This collaborative approach combines the ability of computers to look at the relationships between several hundred thousand known materials, a scale unattainable for humans, and the expert knowledge and critical thinking of human researchers that leads to creative advances.

“This tool is an example of one of many collaborative artificial intelligence approaches likely to benefit scientists in the future.”

Society’s capacity to solve global challenges such as energy and sustainability is constrained by our capability to design and make materials with targeted functions, such as better solar absorbers making better solar panels or superior battery materials making longer-range electric cars, or replacing existing materials by using less toxic or scarce elements.

These new materials create societal benefits by driving new technologies to tackle global challenges, and they also reveal new scientific phenomena and understanding. All modern portable electronics are enabled by the materials in lithium-ion batteries, which were developed in the 1980s, which emphasizes how just one materials class can transform how we live: defining accelerated routes to new materials will open currently unimaginable technological possibilities for our future.