29 Sep PMWC Interview with Slava Akmaev, Ph.D., SVP & Chief Analytics Officer, Berg – Speaker at PMWC 2018 Silicon Valley
We had a chance to sit down with Dr. Slava Akmaev – who will be speaking at our upcoming PMWC 2018 Silicon Valley conference this coming January – to discuss his vision of “why data-driven research is important, and how machine learning and AI will advance precision medicine in the years to come”.
To learn more about the different AI and machine learning sessions planned for PMWC 2018 SV, see the full agenda here. Furthermore, we have scheduled an AI Company Showcase which will take place on Monday January 22 and will provide a 15-minute time slot for selected AI companies. Presenting companies display their latest developments and innovations in the critical areas of data-driven patient care, population health, healthcare process automation, or AI-enabled drug discovery and development. Confirmed Presenting Companies include AI Cure, Altoida, BayLabs, CureMetrix, doc.ai, Good AI Lab, Insilico, Lunit, Medimsight, Mendel.ai, Numerate, NuMedii, Owkin, PathAI, PhenoMX, Physiq, quibim, SkinVision, Suggestic, twoXAR, UCSF, Viome, and Zebra Medical Vision.
Q: Berg provides the Berg Interrogative Biology platform. Can you tell us more about this platform and its AI Analytics? What need is it addressing? What kind of programs are you running using the Berg platform?
A: The Interrogative Biology® platform offers a new direction in drug discovery by juxtaposing disease biology and normal cellular (homeostatic) biology, which is comprised of characteristic signatures including genes, proteins, lipids, metabolites and their network of expression. Interrogative Biology® is a novel automated approach to biopharma research. It is a top-down artificial intelligence driven workflow for conducting studies in molecular mechanisms. The platform has also been designed to pinpoint key regulatory mechanisms and potential intervention points in an automated fashion. BERG is the pioneer in using artificial intelligence in pharmaceutical research. Over a decade, BERG has developed an extensive therapeutics and diagnostics pipeline by applying Interrogative Biology® in disease areas such as cancer, diabetes, and neurology. In parallel, BERG has validated this method through rigorous and expert guided experimental work in its in vitro laboratory. The platform’s advantage is in rapid discovery of new therapeutic targets and biomarkers. Additionally, BERG utilizes the platform for effective patient stratification in its clinical programs. Interrogative Biology® is designed for discovery of efficacy and safety biomarkers that help BERG succeed in clinical trials by selecting the right patient populations and shortening the approval process.
Q: Berg addresses big data challenges. What type of data do you focus on and what solutions do you offer to derive actionable insights from these data?
A: BERG is engaged in projects that span the entire spectrum of healthcare data sources. In drug development and diagnostics, BERG uses the artificial intelligence platform bAIcis® for integration and analysis of high throughput molecular data, phenotypic data and clinical trial records. In digital health, the company works with patient information such as claims data, physician notes and records, pharmacy records, and hospital data. With its partners, BERG analyzes extensive phase III clinical trial data and real world outcomes. The bAIcis® platform integrates data across the continuum and derives insight in a data source unbiased way. bAIcis® is a Bayesian artificial intelligence inference software capable of operating with thousands of features across millions of observations. It learns feature relationships directly from the variability in the data and provides concise and straightforward insight from network topology analyses.
Q: Why is data-driven research important, and how will machine learning and AI advance precision medicine in the years to come? What types of trends should we be on the look-out for?
A: Data-driven research brings another dimension to pharma R&D. By initiating de-novo discovery and research programs, BERG is able to chart largely novel mechanisms of disease progression and on-set. Whether it’s molecular data or data that is more outcomes centered, generating therapeutic and patient management hypotheses outside of common, community thinking is of tremendous benefit. It takes the company away from costly in-licensing agreements and positions our R&D programs in a context of novel intellectual property unencumbered by other participants in the space. From a scientific and clinical perspective, BERG has done distinguished work in oncology and neurodegeneration. Our research programs go well beyond the typical therapeutic framework, they target cancer metabolism and unfamiliar molecular mechanisms in Parkinson’s and Alzheimer’s diseases.
Q: How is Berg addressing the industry need to provide better (or the right) treatment for the right person?
A: BERG is committed to Precision Medicine. We use Interrogative Biology® extensively in clinical development with the aim of the discovery of patient stratification molecular biomarkers. The biomarker discovery process starts early on in the Phase I clinical trials where BERG collects multiple time points of biological fluids from the patients in the trial. The findings are validated in Phase II studies leading to a set of validated complementary molecular and digital diagnostics. These signatures are used for patient stratification in Phase III. This strategy allows BERG to pinpoint the right patient population for the therapy before the drug is on the market.
Q: Where should industry and healthcare resources be invested in for better disease understanding?
A: From my perspective, there are three areas that I hope will get more attention from the investors and pharma executives. One is data diversity. To better our disease understanding, the industry R&D needs to look at molecular data more comprehensively. It is important to remember that cellular biochemical reactions at large define normalcy. When we study disease phenotype and its molecular inner workings, careful assessment of changes in the levels of the proteins, metabolites, and lipids provides critical hypotheses and intervention points for drug development. This leads us to the second point of data quality. Extending scientific and technology resources to enhance the quality of public and private data repositories would go a long way in solidifying the baseline foundation of disease understanding at the molecular level. Thirdly are the analytics tools and methods. BERG has invested in development of a Bayesian AI platform bAIcis®. Others in the industry are actively engaged in applying neural network algorithms to imaging and clinical data. There is room for other approaches that will engage complex mathematics in life sciences applications.
Q: When it comes to working with data in healthcare, what are some of the challenges on our way to positively affect real-world outcomes, and how can we overcome those?
A: Achieving proof points is a labor-intensive and lengthy task in healthcare. When BERG started on this path almost a decade ago, typical feedback and questions we received were about the power of the BERG approach and substantive use cases for Interrogative Biology®. Developing a successful drug from a de novo discovery takes several years if not more; validating a novel diagnostic might take several years as well. We are excited to have a number of advanced programs in the clinic and other stages of development. However getting to this stage will remain a challenge for new technologies entering the healthcare industry today.
Q: Can you talk about an example of the Berg Interrogative Biology platform used in a data-driven approach in drug discovery or biomarker identification? What are the results and lessons learned?
A: Our flagship drug development program in oncology is currently in phase II studies in pancreatic cancer and phase I/II in glioblastoma. The compound BPM31510 is a cancer metabolism therapeutic. Its development is guided by artificial intelligence. BERG applied Interrogative Biology® platform to elucidate the compound mechanism of action, guide clinical development toward the right tumor types, and BERG currently uses the platform to develop complementary diagnostics for potential patient stratification in registrational trials. In addition, BERG has late stage pre-clinical programs in diabetes and Parkinson’s disease, a validated diagnostic panels in prostate cancer, and validation ready predictive diagnostic programs in neurodegeneration.
