KDnuggets Home » News » 2015 » Mar » Opinions, Interviews, Reports » Interview: Slava Akmaev, Berg on Healthcare Transparency & Effectiveness using Big Data ( 15:n08 )

Interview: Slava Akmaev, Berg on Healthcare Transparency & Effectiveness using Big Data


We discuss Big Data Analytics at Berg, making Healthcare effective through Big Data, impact of falling cost of DNA sequencing, Berg AI-Analytics Suite and more.



Slava AkmaevSlava Akmaev, Ph.D. is Senior Vice President and Chief Analytics Officer at Berg. Dr. Akmaev leads innovation in Big Data analytics applied to fundamental patient management problems in healthcare, drug development, and diagnostics. During his tenure at Berg, Dr. Akmaev has developed and launched Berg Analytics Suite of data science applications that allow the life scientists and clinicians to harvest the power of Big Data and affect real world patient outcomes. He works closely with the drug and biomarker development teams and directs research informatics, healthcare analytics, and personalized medicine programs within Berg and its subsidiary companies.

Dr. Akmaev continues to innovate in the application of Bayesian artificial intelligence in healthcare IT. Prior to joining Berg, he was Vice President of Scientific Affairs at a Big Data analytics company, Scientific Associate Director at Genzyme Genetics and a Bioinformatics lead at Genzyme. Dr. Akmaev holds a Ph.D. in Applied Mathematics from the University of Colorado at Boulder.

Here is my interview with him:

Anmol Rajpurohit: Q1. What does Berg do? How important is Big Data to Berg's strategy and operations?

Berg LogoSlava Akmaev: Berg is a widely diversified company in healthcare. We have four business divisions that independently focus on areas such as drug discovery and development, detection of novel biomarkers, development and commercialization of diagnostic tests, data analytical work in healthcare, and predictive algorithms for prevalent human diseases. Collectively, Berg’s business divisions worked with Big Data years before many in the industry knew what it was. In fact, our initial research and development of the Berg Interrogative Biology™ platform centered around high volume molecular data. More recently, we transitioned to digital data such as clinical records, patient EMRs, billing data, etc. The ability to operate and derive actionable insight in these data modalities is required for the realization of true precision medicine in clinical practice.

AR: Q2. One of the biggest concerns in healthcare is the rising costs. What potential solutions does Big Data offer for this problem?

SA: According to the Institute of Medicine, approximately 30 percent of health expenditures in the U.S. is wasted - costs that are not related to medical procedures or drug pricing. Those expenses are directly linked to malpractice, drug side effects, unnecessary treatment, re-admissions, repeated Health Spendingoffice visits etc. Many of these issues arise because the medical and scientific communities do not have a clear understanding on how a drug or procedure will affect a specific individual. For instance, will it have a different outcome for a 40-year old African American male versus an 85-year old Asian female? Most of today’s treatment strategies for chronic and acute disorders are population-based, which may work for 70 percent of patients, and then it becomes a matter of luck whether it works for you, or are you in the selected 70 percent or the unlucky 30 percent. Berg believes that in the 21st century we should be able to do better and provide the right treatment for the right person.

AR: Q3. The cost of DNA sequencing has been falling faster than Moore's Law. Has this led to a proportionate increase in the amount of data generated and consequently, analytical insights?

DNA SequencingSA: The genomics revolution has led to tremendous insight into human disease. However, it is only one piece of the puzzle. If we think about clinical practice and medicine, we could probably count on both hands the number of clinically actionable genomics outcomes that we, as a research community, have discovered.

If one compares the number of actionable insights with the public and private money spent on genomics in the last 15 years, it is apparent the clinically useful productivity of genomic research is not that great. It’s time to look beyond DNA and invest more public resources into research functionally linking the variety of molecular data to phenotypes.


AR: Q4. What are the key goals of Berg Interrogative Biology platform? What progress has been made so far?

SA: Berg Interrogative Biology™ is a discovery platform. Our philosophy at Berg is to start every development project from the basics in biology, generate quality data, analyze data by complex, artificial intelligence-based mathematical methods and let the analysis guide our triage process. Over the last 8 years, we have been able to look at diseases such as cancer, diabetes, Alzheimer’s and Parkinson’s and others.
Interrogative Biology
On the diagnostics side, Berg has two programs that are at the late stages of clinical development, a panel of three biomarkers in prostate cancer and a panel of biomarkers in heart failure. Additionally, we have a number of diagnostic programs in the development phase.

AR: Q5. What are the major components and capabilities of Berg AI-Analytics suite?

SA: Berg AI-Analytics suite is an assembly of data analysis workflows implemented in various programming and statistical languages such as C/C++, R, Perl. The idea behind this suite of tools is to provide to the end user the capabilities of performing data ETL, analysis and visualization in one package. We have developed a number of ETL and normalization workflows that cover the extent of the data type diversity we work with on a daily basis. In collaboration with Eric Schadt’s team at the Icahn Institute for Genomics and Multiscale Biology at Mount Sinai Hospital in New York, we produced a Bayesian network inference package that was originally developed at IIGMB. That software and our internally developed Bayesian network learner allow us to infer cause-and-effect relationships in large molecular and clinical data sets. More importantly, it allows us to integrate disparate data modalities into a single mathematical model.
Berg AI
Finally, one of the most important features of the suite is visualization. We have been using Cytoscape and Cytoscape Web for many years. Our most common visualization templates are standardized and created automatically from the resultant networks. In addition to two dimensional network presentations, more recently we implemented a 3D visualization workflow that gives the expert user the ability to create stunning animations and images.
3D Visualization Berg
Second and last part of the interview

Related: