Jun 19 2017

It’s Benign!

by at 2:09 pm

My surgeon Dr. Shawna Willey walked into the patient exam room where I waited nervously. I first saw her thumbs up before her beaming face. I could breathe again!

My friends and I who recently turned 40 and starting our baseline mammograms can’t help but wonder about the lack of consensus on optimal cancer screening strategies, target populations, its benefits and harms. My colleague Dr. Jeanne Mandleblatt and her team have studied breast screening strategies for decades and have shown that biennial screening from ages 50-74 achieves a median 25.8% breast cancer mortality reduction whereas annual screening from ages 40-74 reduces mortality an additional 12% but introduces very high false positive rates. Many women and their families are subject to extreme anxiety due to the sheer number of repeat mammograms, false-positives, benign biopsies and in 7% of the cases an over diagnosis.

My experience and of my friends with breast cancer screening are raising many questions. How can we better predict the target risk population who must undergo screening early and often? Would this decision-making process consider risk factors, lifestyle, and patient preferences? How often are patients with a diagnosis of a benign breast condition on a stereotactic core needle biopsy upgraded to a non-benign diagnosis on an excisional biopsy which requires full sedation and surgery? What was the care journey like for other patients like me – Asian female, healthy, no family history? How many in the US and globally have access to the excellent care and follow-up that I was privileged to receive from Dr. Willey and her expert team?

Touted as the fourth industrial revolution, Artificial Intelligence is poised to empower clinicians, patients and researchers in answering these questions. What is AI? The term was coined by Dartmouth professor Dr. John McCarthy in 1956 and defined as “the science and engineering of making intelligent machines, especially intelligent computer programs.” Applications of AI in medicine have been limited by the complexity of highly cognitive processes such as making a medical diagnosis or selecting a treatment which require integration of thousands of datasets with millions of variables and multiple interactions between these variables. It takes years to collect, organize and publish practice changing results such as Jeanne’s screening study. What if we could use data that we routinely collect during the care process and effectively use AI to assist clinicians in real-time to make informed treatment decisions?

Interested in learning more about AI in Biomedicine? Want to engage with expert scientists and product developers in AI? Register for Georgetown’s Big Data in Biomedicine symposium on October 27th!

Companies like Google and Amazon are betting big on this. Jeff Bezos wrote “..it is hard to overstate how big of an impact AI will have on society over the next 20 years”; Google’s Sundar Pichai, when asked recently about the next big thing at Google responded “I can’t quite tell exactly but advances in AI and machine learning, we are making a big bet on that and this will bring a difference in many many fields”.

We cannot have a conversation about AI in medicine without discussing IBM Watson, the supercomputer that sifted through 20 million cancer research papers, and conducted a differential diagnosis on a difficult to treat leukemia patient in 10 minutes by combining genomic data with the power of cognitive computing. One concern that informaticians including my informatics mentor Dr. Bill Hersh have raised is that the publicity around Watson has mostly been from news articles and press releases, primarily from researchers at IBM and call for a more scientific analysis, not n-of-one case reports, of its abilities in clinical decision making. Systems like Watson will benefit from systematic expert knowledge input to guide the cognitive computing processes in navigating the complex medical pathways.

While still early, AI is already starting to make important contributions to Medicine says AI professor at MIT and a recent breast cancer survivor, Dr. Regina Barzilay. She and her team are asking all the right questions of data – “can we apply the sophisticated algorithms we use to predict customer’s shoe-buying habits to adjust treatments for cancer patients?” “Can computers detect signs of breast cancer or even pre-malignancy earlier than humans are currently capable of?” And the Holy Grail – “Can we use the huge quantities of data from smart toothbrushes, wearables, genomic sequencing, medical records to get to the first and right treatment?”

What next?

In the last decade, big data in biomedicine has focused on collecting (e.g. through mobile and other IoT) and organizing (e.g. cloud computing) information but all signs point in one direction for the next decade – real world applications of AI. We will witness the development of expert systems, question-answering systems and deep learning methods that begin to address complex real world problems in medicine. These will augment, not replace, human expertise. Winners will find ways to rapidly and accurately integrate human input with computational output. Usability of these tools by end users and human factors will be key.

While a true tech automation enthusiast at heart and practice, I will never forget Dr. Willey’s kind and soft words as she clearly explained my pathology report. She also carefully noted in my medical record the rare chlorohexidine pre-op antiseptic agent hypersensitivity that I had developed post anesthetic induction.

One more data point!

              Let’s continue the conversation:

No responses yet | Categories: From the director's office,Newsletter,Subha Madhavan | Tags: , , , , ,

Feb 13 2015

Informaticians on the “Precision Medicine” Team

by at 8:34 am

My first recollection of the term “Precision Medicine” (PM) is from a talk by Harvard Business School’s Clayton Christensen on disruptive technologies in healthcare and personalized medicine in 2008. He contrasted precision medicine with intuitive medicine, saying, “the advent of molecular diagnostics enables precision medicine by allowing physicians to delineate conditions that are likely constellations of diseases presenting with a handful of symptoms.” The term became mainstay after NRC’s report, “Toward precision medicine: Building a knowledge network for biomedical research and a new taxonomy of disease.” Now, we converge on the NIH’s definition– PM is an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle.

“Cures for major diseases including cancer are within our reach if only we have the will to work together and find them.  Precision medicine will be the way forward,” says Dr. John Marshall, head of GI Oncology at MedStar Georgetown University Hospital.

The main question in my mind is: How can we apply PM to improve health and lower cost? Many sectors/organizations are buzzing with activity around PM to help answer this question.

NIH is developing focused efforts in cancer to explain drug resistance, genomic heterogeneity of tumors, monitoring outcomes and recurrence and applying that knowledge in the development of more effective approaches to cancer treatment. In a recent NEJM article, Drs. Collins and Varmus describe NIH’s near-term plan for PM in cancer and a longer-term goal to generate knowledge that is broadly applicable to other diseases (e.g., inherited genetic disorders and infectious diseases). These plans include an extensive characterization and integration of health records, behavioral, protein, metabolite, DNA, and RNA data from a longitudinal cohort of 1 million participants. The cost for the longitudinal cohort is roughly $200M to expand trials of genetically tailored treatments, explore cancer biology, and set up a “cancer knowledge network” for sharing this information with researchers and oncologists.

FDA is working with the scientific community to ensure that the public can be confident that genomic testing technology is safe and effective while preserving innovation among developers. The FDA recently issued draft guidance for a framework to regulate laboratory-developed tests (LDTs). Until now, most genomic testing is done through internal custom developed assays or commercially available LDTs. The comment period just ended on Feb 2.

Pharma/Biotech companies are working to discover and develop medicines and vaccines to deliver superior outcomes for their customers (patients) by integrating “Big Data” (clinical, molecular, multi-omics including epigenetics, environmental, and behavioral information).

Providers, health systems, and Academic Medical Centers are incorporating appropriate molecular testing in the care continuum and actively participating in clinical guideline development for PM testing and use.

Public and private Payors are working to appropriately determine clinical utility, value and efficacy of testing to determine reimbursement levels for molecular diagnostic tests – a big impediment for PM testing right now. Payors recognize that collecting outcomes data is key to determining clinical utility and developing appropriate coding and payment schedule.

Diagnostic companies are developing and validating new diagnostics to enable PM, especially capitalizing on the new value-based reimbursement policies for drugs. They are also addressing joint DX/RX approval processes with the FDA.

Professional organizations are setting standards and guidelines for proper use of “omics” tests in a clinical setting – examples include AMA’s CPT codes, ASCO’s QOPI guidelines, or NCCN’s compendium.

Many technology startups are disrupting current models in targeted drug development and individualized patient care to deliver on the promise of PM. mHealth domain is rapidly expanding with innovative mobile sensors and wearable technologies for personal medical data collection and intervention.

As informaticians and data scientists, we have atremendous opportunity to collaborate with these stakeholders to contribute in unique ways to PM:

  1. Develop improved decision support to assist physicians in taking action based on genomic tests.
  2. Develop common data standards for molecular testing and interpretation
  3. Develop methods and systems to protecting patient privacy and prevent genetic discrimination
  4. Develop new technologies for measurement, analysis, and visualization
  5. Gather evidence for clinical utility of PM tests to guide decisions on utility
  6. Develop reference databases on the molecular status in health and disease
  7. Develop new paradigms for clinical trials (N of one trials, basket trials, adaptive designs, other)
  8. Develop methods to bin patients by mutations and pathway activation rather than by tissue site alone.
  9. Create value from Big Data

What are your ideas? What else belongs on this list?

Jessie Tenenbaum, Chair, AMIA Genomics and Translational Bioinformatics shares: “It’s an exciting time for informatics, and translational bioinformatics in particular. New methods and approaches are needed to support precision medicine across the translational spectrum, from the discovery of actionable molecular biomarkers, to the efficient and effective storage and exchange of that information, to user-friendly decision support at the point of care.”

A PricewaterhouseCoopers analysis predicts the total market size of PM to hit between $344B-$452B in 2015. This includes products and services in molecular diagnostics, nutrition and wellness, decision support systems, targeted therapeutics and many others. For our part, at ICBI, we continue to develop tools and systems to accurately capture, process, analyze, and visualize data at patient, study, and population levels within the Georgetown Database of Cancer (G-DOC). “Precision medicine has been a focus at Lombardi for years, as evidenced by our development of the G-DOC, which has now evolved into G-DOC Plus. By creating integrated clinical and molecular databases we aim to incorporate all relevant data that will inform the care of patients,” commented Dr. Lou Weiner, Director, Lombardi Comprehensive Cancer Center who was invited to the White House precision medicine rollout event on January 30.

Other ICBI efforts go beyond our work with Lombardi. With health policy experts at theMcCourt School of Public Policy, we are working to identify barriers to implementation of precision medicine for various stakeholders including providers, LDT developers, and carriers. Through our collaboration with PRSM, the regulatory science program at Georgetown, and the FDA, we are cataloging SNP population frequencies in world populations for various drug targets to determine broad usefulness of new drugs. And through theClinGen effort, we are adding standardized, clinically actionable information to variant databases.

The President’s recent announcements on precision medicine have raised awareness and prompted smart minds to think deeply about how PM will improve health and lower cost. We are one step closer to realizing the vision laid out by Christensen’s talk in 2008. ICBI is ready for what’s next.

Let’s continue the conversation – find me on e-mail at subha.madhavan@georgetown.edu or on twitter at @subhamadhavan

No responses yet | Categories: From the director's office,Newsletter,Subha Madhavan | Tags: , , ,

Jan 12 2014

Genomes on Cloud 9

by at 4:51 pm

Genome sequencing is no longer a luxury available only to large genome centers. Recent advancements in next generation sequencing (NGS) technologies and the reduction in cost per genome have democratized access to these technologies to highly diverse research groups. However, limited access to computational infrastructure, high quality bioinformatics software, and personnel skilled to operate the tools remain a challenge. A reasonable solution to this challenge includes user-friendly software-as-a-service running on a cloud infrastructure. There are numerous articles and blogs on advantages and disadvantages of scientific cloud computing. Without repeating the messages from those articles, here I want to capture the lessons learned from our own experience as a small bioinformatics team supporting the genome analysis needs of a medical center using cloud-based resources.

 Why should a scientist care about the cloud?

Reason 1: On-demand computing (such as that offered by cloud resources) can accelerate scientific discovery at low costs. According to Ian Foster, Director of the Computation Institute at the University of Chicago, 42 percent of a federally funded PI’s time is spent on the administrative burden of research including data management. This involves collecting, storing, annotating, indexing, analyzing, sharing and archiving data relevant to their project. At ICBI, we strive to relieve investigators of this data management burden so they can focus on “doing science.” The elastic nature of the cloud allows us to invest as much or as little up front for data storage. We work with sequencing vendors to directly move data to the cloud avoiding damaged hard drives and manual backups. We have taken advantage of Amazon’s Glacier data storage that enables storage of less-frequently used data at ~10 percent of the cost of regular storage. We have optimized our analysis pipelines to convert raw sequence reads from fastq files to BAM files to VCF in 30 minutes for exome sequences using a single large compute instance on AWS with benchmarks at 12 hrs and 5 hrs per sample for whole genome sequencing and RNA sequencing, respectively.

Reason 2: Most of us are not the Broad, BGI or Sanger, says Chris Dagdigian of BioTeam, who is also the co-founder of the BioPerl project. These large genome centers operate multiple megawatt data centers and have dozens of petabytes of scientific data under their management. The rest of the 99 percent of us thankfully deal in much smaller scales of a few thousand terabytes, and thus manage to muddle through using cloud-based or local enterprise IT resources. This model puts datasets such as 1000 genomes, TCGA, UK 10K, etc. in the fingertips (literally a click away) of a lone scientist sitting in front of his/her computer with a web browser.  At ICBI we see the cloud as a powerful shared computing environment, especially when groups are geographically dispersed.  The cloud environment offers readily available reference genomes, datasets and tools.   To our research collaborators, we make available public datasets such as TCGA, dbGAP studies, and NCBI annotations among others. Scientists no longer need to download, transfer, and organize other useful reference datasets to help generate hypotheses specific to their research.

Reason 3: Nothing inspires innovation in the scientific community more than large federal funding opportunities. NIH’s Big Data to Knowledge (BD2K), NCI’s Cancer Cloud Pilot and NSF’s BIG Data Science and Engineering programs are just a few of many programs that support the research community’s innovative and economical uses for the cloud to accelerate scientific discovery. These opportunities will enhance access to data from federally funded projects, innovate to increase compute efficiency and scalability, accelerate bioinformatics tool development, and above all, serve researchers with limited or no high performance computing access.

So, what’s the flip side? We have found that scientists must be cautious while selecting the right cloud (or other IT) solution for their needs, and several key factors must be considered.  Access to large datasets from the cloud will require adequate network bandwidth to transfer data. Tools that run well on local computing resources may have to be re-engineered for the cloud.  For example, in our own work involving exome and RNAseq data, we configured Galaxy NGS tools to take advantage of Amazon cloud resources. While economy of scale is touted as an advantage of cloud-based data management solutions, it can actually turn out to be very expensive to pull data out of the cloud. Appropriate security policies need to be put in place, especially when handling patient data on the cloud. Above all, if the larger scientific community is to fully embrace cloud-based tools, cloud projects must be engineered for end users, hiding all the complexities of the operations of data storage and computes.

My prediction for 2014 is that we will definitely see an increase in biomedical applications of the cloud. This will include usage expansions on both public (e.g. Amazon cloud) and private (e.g. U. Chicago’s Bionimbus) clouds. On that note, I wish you all a very happy new year and happy computing!

Let’s continue the conversation – find me on e-mail at sm696@georgetown.edu or on twitter at @subhamadhavan

No responses yet | Categories: From the director's office,Subha Madhavan | Tags: , , , ,