7-8 October 2015 | Ronald Reagan Building and International Trade Center, Washington DC
MDCPartners and the ta-Scan clinical business intelligence keynote presentation by David Cocker on the third Big Data in Clinical Development conference in Washington.
2. My Structure
• Quick heads-up on the oncology space
• Evolving complexity, big science and big data
• What has changed in available data?
• Challenges for pharma (old versus new)
• Enlightenment versus disentanglement
• Merging new technologies with old business models
• Voilà: what is possible!
10. Leveraging technologies for a more complete view
New data
Technologies
· Ontologies
· Smart algorithms
· Robotics
Private data
· CTMS
· Impact
Tools & applications
Connected Information
· Modeling (Monte Carlo sim)
· Communication platforms
· Custom visualizations
· Aggregation
· Data ranking
· Summary statics
Application is any activity independent of the data
source. It can be part of a connected process platform
or independent
18. BUSINESS QUESTIONS
Design & specifications
· Current disease management
· Standards of care seeking advice
Global feasibility
Recruitment assumptions
Operational choices
· Country allocation
· Site selection
· Investigator selection
Supplementing investigator signature with claims and ethnicity data
19. BUSINESS QUESTIONS
Design & specifications
· Current disease management
· Standards of care seeking advice
Global feasibility
Recruitment assumptions
Operational choices
· Country allocation
· Site selection
· Investigator selection
Mapping site activity by competing drugs
20. BUSINESS QUESTIONS
Design & specifications
· Current disease management
· Standards of care seeking advice
Global feasibility
Recruitment assumptions
Operational choices
· Country allocation
· Site selection
· Investigator selection
Scenario modeling
Modeling (Monte Carlo simulation)
21. BUSINESS QUESTIONS
Design & specifications
· Current disease management
· Standards of care seeking advice
Global feasibility
Recruitment assumptions
Operational choices
· Country allocation
· Site selection
· Investigator selection
Epidemiology
Site Properties Mapping out the big picture (Reporting)
22. BUSINESS QUESTIONS
Design & specifications
· Current disease management
· Standards of care seeking advice
Global feasibility
Recruitment assumptions
Operational choices
· Country allocation
· Site selection
· Investigator selection
Epidemiology
Aggregated view of a high-profile clinical investigator
23. Would you let your kid fly a A380
without any training?
Editor's Notes
Thank you, Jack, for the kind introduction.
Good morning, everybody. First, I’d like to thank Jack for his kind introduction, as well as Teresa and Trent for making a great case for innovation and the opportunities to be gleaned from big data in a clear and engaging presentation.
As my title slide indicates, I’m here to talk about data-driven decision making in oncology immunotherapy. As you can see, data can take you in many directions, all around the world. But if you have the right navigator you’ll never get lost. And you might even arrive at your destination sooner than expected.
Today, we hear about big biology, big pharma, big data, and big money expecting big things from big cancer. That’s a lot of BIGs.
It’s a given that the Internet is BIG, as it generates tons of data. So our focus is on categorizing these data and making use of evolving data sources, so all stockholders can start making better decisions in drug development.
I’ll be looking at the current state of the oncology space, and addressing challenges associated with data availability and complexity, new pharmaceutical practices and regulations, multiple data interpretations, as well as technical challenges and opportunities.
And speaking of challenges, 20 minutes may not be enough time to fully elaborate on all the fantastic science unraveling in the oncology space today. But I’ll start with immunotherapy, as the results of many ongoing clinical trials have been so promising, and in some cases spectacular.
We’ve analyzed the abstracts of all major cancer societies by searching keywords, and found a remarkable amount of interest generated by checkpoint inhibitors, and in particular PD-1 inhibitors vs other disease modifying drugs. We’ve also seen spectacular results in studies of therapies for melanoma, and investigators are quite confident they’ll find similarly impressive outcomes for the treatment of lung and head and neck cancers.
The race is on for pharmaceutical companies to establish and maintain their position in this emerging therapeutic space.
There are leaders and many are rushing to catch up. And understandably, starting new trials and optimizing clinical strategy is high on their agendas.
So, to gain some perspective on how far we’ve come with data, I did what anyone over the age of six does these days. I went straight to YouTube. And as soon as I typed in Memorial Sloan-Kettering, up popped a clip of famed oncologist Jose Baselga, reminiscing about the tools he had as an intern 20 years ago. They included a written patient chart, a few lab tests, foggy X-rays, a meager selection of drugs, and a few books to consult.
Today, of course, we can sequence thousands of tumors to unravel their genomic character, utilize state-of-the-art imaging technologies, measure numerous biomarkers, and choose from hundreds of drugs to treat specific cancers. And we can expect another 800 in the pipeline. Yes, 800! And physicians are no longer burdened with reading text books, when all is available online.
Choice, of course, adds complexity. But I was still a bit taken aback by Baselga’s conclusion that “the complexity of cancer treatment decisions today cannot be left in the hands of a physician only.”
In my big data world, I’m looking for data sources to help my clients see the full scope of a therapeutic area. I’m striving to augment their existing intelligence so they can make more informed decisions. So the true focus of this talk is on risk reduction through the organization of public information into analyzable objects.
The increased availability of public data has already drastically improved the process of clinical research. A great deal of dependable information already resides on online, but just needs to be properly accessed to produce optimal insights. Easier said than done, right?
I somewhat agree, as we’ve taken up to 5 years to get some of our ontologies ready for prime time.
Heuristically speaking, we want to programming software and create the most efficient algorithms to accomplish various tasks. Using all data available data private domain, such as CTMS, and the public domain. More power can be brought to improve methods for predicting successes or failures. So a good feasibility plan is the phantom target all of us chase.
One Challenge will such enhanced solutions be adopted?
Do they fit with the current business model?
As an industry, we don’t have good KPIs to measure return on investment of innovation.
Just imagine how long we have been talking about electronic data capture and the paperless evolution.
How do you innovate when many companies are running five-year-old versions of Internet Explorer rush to mind)?
Cancer has been one of the biggest public health challenges facing medicine: According to the WHO, lung cancer alone killed 1.6 million people in 2012.
. We have a huge unmet medical need. Ok, we have promising drugs I calculate an major increase in the number of trials in the coming years. Not only in lung cancer.
We will need to add 50,000 or more patients to lung cancer studies. But how are we going to do this?
Where shall I go? Which country should I pick? To decide, I’ll need to consult organizations with the right resources, culture, disease savvy, and awareness of trends.
Which phase which investigators,
Do there communicate well. And most important, deliver my patients cohort on schedule.
Slide 16
Big data is very good at getting facts and statistics together but maybe fails a bit on the human side
And who’s that best site to pick If there is an adequate patient pool?
Are diagnostics and screening facilities in place?
Is the organization able to do clinical research?
Are there any other clinical programs run at the organization that could be in conflict
Unfortunately, there are only a handful of healthcare organizations with the necessary resources and clinical experience to handle very early phase oncology drugs. Protocols forused in other therapeutic areas are not as complex. For example, studies of a lipid lowering agent have the simple surrogate endpoint of cholesterol levels.
A few weeks ago I asked the team to analyze the impact on top of enrolling sites and lung cancer
I was quite surprised to see how dominant the top 50 US Institute were. They participate in all trials but the contribution patients in Phase 3 is only a few percent
This slide clearly shows where companies in the space populated trials when moving into phase 3
But conducting feasibility outside the US ads complexity reliance on affiliate companies and CRO’s and simply it was just list data, us is a highly sophisticated healthcare system with the availability of public data. This is the case in Brazil Belgium or Bellies
Picking the right countries a pivotal part of the feasibility process, regulations and economics compel companies to use the US and Europe often they ony have about 30% of freedom to choose other locations.
They're about 65,000 sites in the world that have a signature performing lung cancer trials about 30% seem to be discarded
Even adding a quite benign Data like the US census can add tremendous value if you have already analyzed other data sources, layering claims data may reveal significant areas we doctors could referred patients to your clinical site.
Assembling a helicopter view when everybody is going is not trivial to do manually
Analyzing the registry and inferring clinical sites by their academic footprint can reveal trial saturation, internal cannibalism or new opportunities
10 years ago you might be lucky if you could simulate your own internal historical data to garner some view on expected recruitment rates,
today we have over 200,000 trials that can render quite robust recruitment assumptions, plus it allows you to quickly adjust scenarios. Most analysts did not have time to adjust and revalidate their assumptions, you had one shot.
Time does not permit me to explain the depth and richness such technologies can compile about a large institute,
especially today. it's a very large checklist of what you need to verify
possibly this is one of the reasons why oh industry is reluctant to expand its investigative pool
Anything which gets you half way there must be a good thing.
Site selection is not analogous to investigative selection
Investigators have to coexist with the bureaucracies of their own organizations. Whether a professional US site or a UK hospital trust hospital, your investigator holds it all together. And expanding your investigator pool especially if you're moving into new therapeutic area is it timely activity an often based on hearsay and recommendations.
In the global clinical theatre sometimes it is useful to have metrics that can remove the emotion from the equation, without assessing a KOL For an advisory board or a clinical trial, aggregating the key data saves endless debates about who is best for the job.
Big data will usher in a flood of groovy dashboards graphics tools and applications, corporations will only have to invest in the algorithms but seriously investment in the training and adoption of these new techniques
I look forward to meeting you in the next few days and haering your stories how we can mobilize data for betterment of patient care.