Is it a good time to be a data scientist in the pharmaceutical industry?
We discuss how the role of data analytics is changing in the industry and what the real "value add" is
Data analytics is surrounded by a lot of hype, but where is the real ‘value add’ for the Pharmaceutical industry? How is the industry preparing for the future of drug development and patient engagement?
With the Data Analytics for Pharma Development Forum coming up, we spoke to Nigel Hughes, Scientific Director at Janssen, to find out how the role of data analytics is changing in the pharmaceutical industry.
An Evolving Model Triggers The Need For Change
The drug development process has changed in the past few years, in part due to the growth of new technologies. This process change will continue, beginning to involve the incorporation of technology shifts, methodological developments, cultural changes and regulatory and post-authorization requirements such as HTA. “We are now moving from the former blockbuster ‘one size fits all’ model to the precision, translational and outcome-driven model” says Hughes. As a consequence, roles such as Scientific Directors have to work increasingly around external collaboration and the field requires a greater flexibility than in the past, for anyone involved in the process. Hughes believes that a model for the future could be around an involvement in IMI programs for example. In this model, there would be pre-competitive collaboration between industry partners, academia and SMEs, to name a few. The aim of this new model would be to have a wider and more holistic approach to generating insights into biology and healthcare.
Data: Hype Or Real Potential For Clinical Development Processes?
The potential behind leveraging data analytics as a means to create smaller, short and cost-effective clinical development processes, for example through data mining, data modelling and predictive analytics all comes down to two things. Firstly, the question companies are trying to answer and secondly, what is the best method and data to answer it. There is growing interest in adaptive, pragmatic and platform studies, incorporating the traditional randomized controlled trials with real world data or evidence by regulators, HTA bodies, academics and the industry. Despite this Hughes argues that “there remains a balance between the old and the new that needs to be addressed.” At the moment, he recognizes there is considerable hype surrounding Data Analytics and ‘smart’ data, but there is also perhaps a lack of appreciation of the complexity of working with real world data. There are a few challenges hindering the access to real world evidence, such as the lack of well captured data – necessarily of variable quality –, governance constraints, harmonization challenges, contextual inhibitors, and methodological and technological hurdles, amongst others. “No one data source is the whole truth” adds Hughes.
Formative work around study feasibility, to machine learning, and then towards more advanced analytics all remain nascent areas, and still belong to the realm of methodological research. It is true that the industry is incorporating these into how they conduct Discovery, Development and Deployment/Commercialization, but currently this is predominantly done within pilots and experimental approaches.
The Challenges Associated With Data In The Drug Development Process
The Ethical Challenges
“We have a clear commitment to safeguard the individual, while maximizing the opportunity to gain better insights into our biology, management and treatment of disease and outcomes” says Hughes. As technology is a driver rather than a facilitator, there will be an increased opportunity to link data from multiple data sources to provide an almost 360 degree insight into patients’ journeys through an illness. The diverse phenotypic, genotypic - and therefore phenomic - understandings available will help the industry truly understand illness beyond the episodic periods of evaluations by healthcare providers. This level of understanding will require informed and functional consent by individuals , with an in-depth understanding of the requisite security and technology risks that this poses. This will then facilitate federated, and consensual, data exchange – either by trusted third parties, nascent technologies or novel approaches.
Examples such as Estonia - which has an ‘X road’ blockchain backbone to data management for citizens - are helpful in illustrating what is feasible in this space. Although, even if it is a technological approach to ensuring absolute transparency to data access, it does require governance and policies in place to manage all the actors involved with safeguards and penalties. “Informatics is in the majority about people, not data” argues Hughes
The Legal and Regulatory Challenges
The acceptance of regulators on the use of real world data and evidence, with the continuing issues of privacy, confidentiality, security and consent is critical, and they all remain vexing challenges. Understanding what issues the regulators have and whether real world data or evidence is the right medium to answer is important. Constraints on data use - at an individual, citizen or patient level, at an institutional or platform level, at a country and then EU level - is complex, fragmented and challenging. “We need a society-wide discussion on how to take this forward , to define the next steps.” “This is for the betterment of healthcare, outcomes and people” adds Hughes. Unfortunately, much of the debate has been mired in post-Snowden security concerns, and the main points of discussion keep focusing on worries and concerns about the risks. These risks - although legitimate - do not allow the positives to come into the discussion
The Company Culture Challenges
One of the most significant challenges currently is the clash of cultures between the ‘traditional’ randomized and controlled trial mind-set and the use of real world data. The former sees data proprietary to the industry - for regulatory requirements, or for approval. The latter sees the use of real world data to support say the feasibility study and clinical trial enrolment. This is not proprietary to the industry and can be generated by, for example, the patients and healthcare providers. This approach is much more ‘messy’ and complex. Real world data and the derived real world evidence can be utilized from Discovery, through Development, to Deployment or Commercialization. All require deepening insights into our biology and the real world, whilst also needing new skills, capabilities, functions and platforms. “It is a good time to be a data scientist in the industry” says Hughes.
The Future Of Predictive Analytics In The Drug Development Process
Hughes believes that it is hard to say exactly what will happen in this space over the next five years. Now, the industry is mostly focused on how it can augment the randomized controlled trials approach that has existed for decades with enhanced feasibility studies and targetted, immediate, enrolment. However, an increasing use of predictive analytics from Discovery onwards is to be expected, depending on the veracity of the real world data available, as well as the methodologies and receptiveness of, for example, regulators. “What we want to move to is being able to extrapolate from real world phenomic cohorts to modelling” says Hughes, further adding that this could provide the industry with important insights into everything from disease, natural history, therapeutics and outcomes, through to service planning and provision and population health. “We are in the progenitor phase today and it will take more than five years to reach this goal.”