AI for Life Sciences

Advancing Heart Health Through Machine Learning and Open-Source Discovery

This year, the MLSC awarded $3.7 million across five projects through the Bits to Bytes program, which provides grants for data-driven, cross-sector projects focused on imaging, cancer, neuroscience, drug discovery, and clinical trial design that develop capital assets, data resources, and novel algorithms to address a myriad of health issues.

The MLSC is proud to highlight the Broad Institute and Dr. Anthony Philippakis’ 2019 project involving the utilization of data-driven approaches to further research and identify trends in cardiovascular diseases.

When Dr. Philippakis, then Chief Data Officer at the Broad Institute and now a General Partner at Google Ventures, teamed up with Bayer Pharmaceuticals, they approached the MLSC with the idea to leverage large-scale health data and machine learning tools to develop new ways to identify people at risk of cardiac events.

Recognizing the significant potential of this research, the MLSC awarded the team with a grant through the Bits to Bytes program. This funding allowed the Broad Institute and Bayer to use computational biology and machine learning to dive into large-scale UK Biobank data and develop new methods for identifying clinically relevant cardiovascular disease subsets.

The project team set out to refine the way cardiovascular diseases are classified, while developing open-source tools that would be made available for other researchers to use in the creation of new products and therapies worldwide. By utilizing both machine learning and cloud computing, they found previously unreported features that are associated with cardiac structure and function, and found correlations between specific genes and unique cardiac diseases.

The team released a variety of open-source tools and methods that the global research community has been able to use to run analyses in other impactful research projects. These have already resulted in several scientific publications related to innovative research in cardiovascular disease.

According to Mahnaz Maddah, Director of Machine Learning for Health at the Broad Institute, the funding from this MLSC program has enabled research that has led to 26 scientific publications to date in academic journals, with more under review, all driving innovation across cardiology, neurology, and beyond.

Unlocking Innovation: The MLSC’s R.E.D. and D.A.T.A. Platforms for Shared Scientific Progress

The foundation of Massachusetts’ thriving life sciences ecosystem is dependent on access to state-of-the-art equipment, cutting-edge data, and collaborative spaces that bring ideas to fruition. Since its inception, the MLSC has made strategic investments in infrastructure and equipment that enable multidisciplinary research, enhance shared core facilities, and increase access to state-of-the-art technologies for Massachusetts-based investigators and industry scientists.

Through its capital grant programs, the MLSC supports the acquisition of advanced research equipment and infrastructure which enables collaboration between industry and academia. These shared assets empower researchers across sectors at any stage to accelerate their work and expand the impact of their projects on patients’ lives around the globe.

Explore R.E.D.

Explore D.A.T.A.

To ensure these resources are accessible and leveraged throughout the ecosystem, the MLSC maintains two public databases. The Research Equipment Database (R.E.D.) is available for companies, research institutions, and any other researchers in Massachusetts to search, find, and engage with our various partners across the Commonwealth to further utilize MLSC-funded equipment.

Examples of recent additions to the R.E.D. database are a $4M advanced Nuclear Magnetic Resonance (NMR) spectrometer at UMass Amherst, a $3.2M whole-body PET/CT imaging machine at MGH, and a $2.6M Cryo-electron microscope at MIT. These cutting-edge resources are now available for all researchers in MA to use because of MLSC investment.

The Database, Algorithms, Tools, and Analyses (D.A.T.A.) Repository serves as a hub for MLSC-funded digital resources, providing access to shared datasets and computational tools which advance scientific collaboration and innovation. These digital resources include neurovascular imaging datasets, metabolomics analysis platforms, and cancer classification neural networks that are now open source and available for any researcher to utilize.

Since its inception, R.E.D. has provided a platform for engagement of almost 500 pieces of high-end scientific research equipment across 36 different institutions ranging from imaging systems to benchtop automation, to sequencers. Over the past six years, D.A.T.A. has compiled information on 51 MLSC-funded datasets, algorithms, software tools, and data-analysis toolkits produced across 13 institutions.