The best approach to maximise RWE technology platform output through the use of advanced computing
Senior Director, Technology, Real World Evidence Strategy & Analytics
Technology industry buzzwords - AI, ML, NLP, to name a few - have been making their way into the public vernacular more and more in recent years. For the pharma and biotech industries, the applications of these buzzwords are far reaching, particularly when it comes to processing large data sets to generate RWE.
As the availability of big data continues to grow exponentially, it only makes sense that new ways of applying these advanced computing skills to process these data sets would become available as well. But what specifically do they mean, what are their differences, what operational processes do they perform and what are the advantages of implementing these innovative technologies in a centralized and secure RWE technology platform?
Let’s start with high level definitions:
Artificial Intelligence (AI): In its simplest form, AI is the development of computer systems to perform tasks that normally require human intelligence. There are many computing functions today that fall in to the AI bucket.
Machine Learning (ML): A subset of AI that enables computers to discover patterns in large data sets, make predictions and improve these predictions over time with repeated exposure to the data. There are three subcategories of ML:
- Supervised Learning: Can discover patterns when examples of those patterns are provided; usually called upon for predictive analytics
- Unsupervised Learning: Can discover patterns without specific examples; most useful when grouping data
- Reinforcement Learning: Based on trial and error; can learn from failures to understand which process works the best; most useful with robotized functions.
The next evolution of ML, often times referred to as Deep Learning, is advanced computing that most closely resembles the way a human brain would process information. Deep Learning involves learning by example and making informed decisions automatically from what it has learned.
Natural Language Processing (NLP): A component of AI that enables computers to understand and process unstructured text and extract meaning from it. A prime research example of NLP would be mining text-based physician notes for pertinent research study data.
Now that we’ve reviewed these terms, let’s discuss five ways that these innovative technologies can be utilized to improve efficiencies in real world data ingestion, normalization and outcomes research.
- When ingesting massive amounts of secondary data from dissimilar sources, there is a need to link and normalize the data into a common data model to generate meaningful insights. In the past, this normalization could only be done manually, and could take a prohibitive amount of time. Now with NLP and ML, the entire end-to-end process can be automated and insights can be generated in a fraction of the time.
- In patient treatment pathways, ML can be used to analyse and compare treatment effectiveness of a particular drug or therapy and ensure, from a payer’s perspective, the prescribed treatment is meeting value expectations.
- Current processes with uncovering adverse events have been quite laborious, with many hours spent mining through data inputs to track and log actual events. With innovative approaches driven by NLP, these large data sets can now be processed in record time. This approach opens up new social repositories of data for mining.
- In the foreseeable future, physicians will be able to rely on AI to better determine if a patient is at high risk for developing certain diseases. Based on available secondary data sources and learned predictive modelling, a patient’s profile can be aligned with these models to identify markers that would determine risk levels.
- With data being collected into electronic health records (EHRs), we now have better access to RWD and ability to track patient behaviour and treatment adherence. ML can be applied to these data sets and used to predict which patients are more likely to discontinue a drug or therapy for a chronic disease. With this analysis, pharma and biotech companies can then take action to focus on improving patient adherence and in turn, increase revenue.
The technology advancements of AI, ML and NLP, as seen in these examples, offer pharma and biotech companies the power to increase meaningful RWE output, decrease time to insights, and make the most of the vast data sources now available. An RWE technology platform that delivers smart data processing, analysis and outcomes offers an unparalleled opportunity to capitalize on these computing advancements. When used as part of an overall comprehensive RWE strategy, AI, ML, and NLP innovations can enhance drug development, improve patient treatment and access, and drive valuable new business opportunities.
To learn more or to discuss your real world data needs, contact our Real World Evidence (RWE) team.
About the Author
Bruce Capobianco has over 25 years’ experience in the architecture, development and implementation of complex big data solutions. He leads a team to develop, enhance, and maintain Real World Evidence (RWE) technology solutions for ICON clients. He has a proven track record of identifying and implementing secure, usable and enduring technologies that augment business processes and optimize productivity. At Syneos he led a global team of architects, developers, PMs and SQA staff in the development of a HIPAA-compliant, trial patient recruitment system, and established and drove disruptive technology trends for competitive advantage.