Faster decision making and better trial oversight at all levels requires truly next generation data integration and analytics. Source and format agnostic data aggregation aided by modern data lake architecture and advanced analytics deliver interactive visualizations to identify new signals, discover hidden insights, minimize risks to trial success enabling data-driven faster decisions and successful outcomes.
Traditionally, controlled trials have dominated drug development. Increasing focus on rare diseases and access to relevant patients have led to challenges where no single dataset can fulfil all research and development requirements. Biopharmaceutical companies need to prioritise their business questions and map them to the appropriate source. Constructing a portfolio allows maximum value to be derived from data sources and enables deep analyses. Historically, this has meant large trials and a focus on data gathering, collection and analysis. This approach increases costs but does not consistently improve outcomes.
As regulatory requirements continue to tighten and research and development costs continue to rise, there is an evergrowing need to:
• ‘Connect the dots’ across studies / compounds / patients - search for association across a large spectrum of data such as genomic, transcriptomic, proteomic, metabolomic, cellular and clinical
• Enable faster preclinical candidate decisions and provide better regulatory responses –to increase productivity of R&D pipeline
• Move from historical analysis towards prediction – hidden relationships between compounds and target genes, molecular pathways, diseases and side effects form a combination of structural, biochemical and cellular data to inform molecule design.
In a recent case of a cancer immunotherapy drug trial failure1 was attributed to the fact that the study examined a patient sample that was too broad while it should have selected only those patients that have a specific immune response. Regulators today demand personalised immunotherapy to identify who will benefit from drug A as against who will benefit from drug B.
Such demand for greater focus on value and patient outcomes highlights the need to leverage an unmatched dataset to address critical issues in clinical trials, thereby escalating the demand for real world data. There is an imperative for accelerating pharmaceutical R&D through targeted patient recruitment for clinical trials by:
• Using claims data, EMR databases, and genomic data to quantify the number of patients with a specific profile as defined by inclusion and exclusion criteria to develop new personalised medicines and support the study design and participant selection for clinical trials
• Analysing physician prescribing behaviour, influence networks, patient mix, payer reimbursement, and estimated retention and lifetime value to determine the most profitable physician targeting strategy
• Seeking of medical scientific information and advice from healthcare professionals and patients, while avoiding the risk of antibribery / anti-corruption regulations
• Leveraging molecular biomarkers to select clinical trial participants more effectively.
An instance of a targeted patient R&D success story2 is the approval received for a lung cancer drug for a 5 - 7 per cent patient subset in 2015. The trials were conducted on only 255 patients and took a total time of three years from discovery to approval, less than half of the typical timeframe!
With the majority of clinical trial cost spent performing on-site monitoring, a new method is needed for an efficient, targeted approach. As TransCelerate BioPharma describes, ‘The principles of centralised analytics can be effectively applied to pharmaceutical development and clinical trial monitoring through RBM, impacting the earlier detection of data quality issues, making possible a string of actions such as:
• Controlling and ensuring that clinical data quality is accurate, complete and verifiable
• Enhancing the safety of patients in a clinical trial, ensuring the rights and well-being of human subjects are protected’
RBM brings data collection and tracking in real-time, and trends and analytic capabilities to the forefront with a direct result in reduction in on-site monitoring cost. For many years, the gold standard in clinical research to achieve these goals has been 100 per cent Source Document Verification (SDV), where Clinical Research Associates (CRAs) check every data point on the information reported by investigators against source records to ensure the information is complete, accurate and valid. However, SDV has shown a negligible effect on data quality. TransCelerate indicated that only 2.4 per cent of all queries generated were SDV related on critical data and thus stated that “SDV has a negligible effect on data quality.”
As risk-based approaches focus on demonstrating actual outcomes, R&D organisations have an opportunity to improve productivity using data analytical capabilities. By combining real-world outcomes data with clinical data, genetic data, and more broadly understanding regional and population data, analytics-savvy organisations can gain insights to recognise research failures faster, design more efficient streamlined clinical trials, and speed the discovery and approval of new medicines while reducing the cost burden.
The following are some of the major reasons for clinical trials becoming increasingly complex:
• Regional regulatory authority requires trials to be conducted on the local patient population (upto a certain percentage of total patients in the trials) for marketing approval in the region
• Trials spanning multiple countries require oversight and ability to identify key risk areas
• Disparate data capture systems store data differently
• Variety of data formats from lab vendors across the countries
• Trials require the processing of data from images which necessitates special software packages.
Consequently, the success of clinical trials is based on retrieving trial data of uncertain quality from disparate electronic systems. This data is critical for understanding the value of pharmaceuticals and their place in an efficient healthcare delivery system, based on real-world evidence.
Biopharmaceutical companies continue to seek access to a variety of healthcare data sources to help improve development programs and to provide better evidence of the value of treatments to patients once a drug is in the market. Companies are unable to leverage their own internal data as the information resides in structured and unstructured sources across a multitude of systems/databases.
Companies must not only cope with this flood of information but also access and harness it to improve the efficiency and perceived value of the innovation effort. A key component of the healthcare value equation is data mining; either to address scientific endpoints or to determine the comparative cost and quality outcomes of a drug candidate.
While in the past, integrating data from multiple sources into an electronic data capture platform or building a data warehouse solution might have worked, it is very restrictive, cost prohibitive and not sustainable any more due to the varying and dynamic nature of the data sources.
Overall, there is a need for moving from a world characterised by fragmented knowledge and processes driven by silos of data sets, to a world where one central data repository holds data derived from scientific discovery, clinical trials, commercial research, enabling data analytics and generating insights across the entire spectrum of the drug development process. This fundamental change is no longer an option, but a necessity for the industry.
Surveys exploring performance of the R&D function and its ability to generate returns have shown that ‘Strategic choices during the drug development can have a significant impact on long term commercial value’. These choices can only be made by aligning end-to-end decision making across the organisation.
The imperative is capitalising on modern data architectures and analytics frameworks.
To access the vast amount of data available, there must be collaboration among data providers, laboratories, hospitals, pharmacies, research groups, and universities, and numerous other sources.
The quality of the data and analysis will ultimately determine the additional value that it can add. Over time, providers with quality data and robust analytic capabilities will be able to differentiate themselves from their peers. The new paradigm is about unlocking the value of the data that is available and combining it with increasing volumes of new data.
Large organisations have the technology, processes and data analytics capabilities in place, which gives them a competitive edge. Many of them also have integrated technology platforms that provide real-time monitoring during clinical studies. These companies can potential adapt these platforms to monitor the market in real-time once a treatment is launched. espite their competitive advantage, the ongoing challenge of data integration and contextualisation remains.
While most organisations remain focused on applying data for reporting, market leaders use analytics to evaluate risks and tradeoffs, understand cost and revenue drivers, and predict trends to help drive performance and innovation. Analytics pushes the innovation agenda forward to create value and provide a comprehensive and informed view of factors that impact R&D outcomes.
A platform with access to real world evidence, clinical trials data, regulations, health information exchanges and electronic health records etc enables companies to leverage advanced analytics. This would have a significant impact on drug research and development, clinical trials, patient care and safety monitoring.
A refreshed analytics strategy that can make information accessible across business units, departments and geographies is required. This solution must bring in a modern approach to address the data aggregation challenge – one that is not relational database oriented. A schema must be defined upfront outlining a flexible solution that can handle any data from any data source, shifting from reactive to realtime decision making.
The proven ‘data lake’ architecture fulfils the need of clinical data integration and aggregation, using AI and machine learning techniques.
This alleviates the need for pre-defined structures and stringent rules-definition employed by legacy software. The platform should:
• Have the ability to process data in a ‘source agnostic and format agnostic’ manner into a central data repository
• Have a short set-up time with low fixed costs, while providing global accessibility and near real-time data analytics and visualisations. This enables effective trial oversight at all levels, enhanced patient safety and faster decision making.
• Enable interactive visualisations and data discovery
• Offer a revolutionary risk analytical platform encompassing study/site/country level dashboards with actionable insights, review and workflow for holistic trial oversight
• Deliver a scalable solution to accommodate the ever-changing business and regulatory needs of clinical research including, ICH E6 R2
Once broken down from their silos, these data from myriad sources serve multiple functions as they traverse the clinical R&D landscape. This enables an ‘analytics ecosystem’ with easy access to variety of data sources, tools and capabilities to perform both exploratory and defined analytics. Data analytical teams appreciate the ability to source data from a centralised repository, and gain access to dynamic real-time visualisation, medical reviews and riskbased monitoring.
Thus, an analytics platform will help the sponsor:
• visually get a handle on multiple studies in near-real time
• help retain control while working with multiple outsourced providers across different geographies
• help monitor the safety aspects of the studies
• enable comprehensive risk-based analysis and
• gain a comparative edge.