Low patient enrollment and under-performing sites can bring costly delays to clinical trials, lengthening the time it takes for treatments to reach patients.
The data represented here shows an evaluation of 100 sites across three key dimensions: performance, quality, and patient density (based on inclusion/exclusion criteria).
The resulting analysis, enabled at scale by proprietary IQVIA machine learning algorithms, allows IQVIA to more accurately predict optimal sites, increase enrollment speed, and reduce trial timelines for sponsors.
Exhibit 1: CORE™-powered clinical trial sites (part 1)
A new level of confidence in site selection
Dive deeper. Look closer. Each dimension of IQVIA’s algorithm for evaluating sites carries its own complexity. Here, the data shows the foundation of dynamic tiering – an algorithm that predicts the performance of an investigator on their next study and predicts operational risk.
This model informs the performance dimension in site assessment. Using Human Data Science, past experience with sites and investigators is transformed into more objective, more informed decision making.
Exhibit 2: CORE™-powered clinical trial sites (part 2)
Finding the best investigators, from the beginning
Why do treatments only work for some patients? Are there signs of a rare disease before symptoms appear – signs that would make diagnosis faster?
The data used in this piece represents what we know about genetic mutations – 100 randomly selected genetic mutations from the publicly available 1,000 Genomes Project. The data set identifies how often each mutation appears and pairs that with other information such as the position within the chromosome.
Information like this is critical to expanding our understanding of genetic diseases. But to elevate this information into clinically valuable insights, the data needs a transformation…
(to be continued)
Exhibit 3: Genomic data (before)
How rare is a rare disease? What we know…
Transforming the genomic information we have into the clinical insights our customers need takes Human Data Science. And a relentless pursuit of privacy-preserving technologies that allow us to link genomic and clinical data and limit risk.
This piece takes the same information from the publicly available 1,000 Genomes Project and runs it through IQVIA’s proprietary privacy-preserving technology.
Through data metamorphosis, the data — cleaned and non-identified — is now ready for richer analysis, ready to be connected to other data sets and used to see new trends and new predictors of clinical outcomes.
Exhibit 4: Genomic data (after)
Data metamorphosis at work
Novel methods for preserving patient privacy can place us on the brink of new discoveries in human health. But what if we could do even more? What if we could use synthetic data for research purposes — data which is not based on the healthcare treatment of any real patients, but is designed to be statistically valid and suitable to meet certain research needs? In other words, we could have access to granular data without any of the privacy risk.
Here you see a well-known diabetes data set that IQVIA is using to build and test synthetic data, a cutting edge use of AI and machine learning.
Using Human Data Science, researchers can develop analytic and predictive models that answer critical questions about human health, without any risk to personal information.
Exhibit 5: Training data for artificial intelligence
Synthetic data powers AI for healthcare
Much of what impacts human health happens outside a single healthcare environment. Or takes years to develop. It is the result of a series of decisions, successes, and failures. Understanding that journey is key to improving it.
This data set is an example of IQVIA’s longitudinal analyses that looks at a variety of medical sources and builds a 360 degree view of the patient journey; an unprecedented look at the human experience of healthcare.
Human Data Science can reveal the complexity of health, and the patient story – across doctor’s offices, EMR records, lab results, hospital stays, etc. – comes to life.
Exhibit 6: Non-identified longitudinal patient data
Seeing the human side of healthcare
To improve outcomes, we need to know which medicines work, which don’t, and why. But to get there requires a collaborative approach to data, technology, and even care delivery.
This piece is generated from a random sample of non-identified data from IQVIA’s Oncology Data Network. Specifically, anti-cancer medications, and doses, administered by cancer stage between 2016 and 2019.
This convergence of real world data and insights builds a common source of knowledge for clinicians to make better decisions for their patients.
Exhibit 7: Real world data for oncology
Collaboration accelerates the fight against cancer
How can we help sales people talk to doctors at the most relevant time?
Here, a machine learning algorithm predicts the non-identified cancer patients who are likely to move to a second-line treatment in the next three months, based on a series of triggers. The algorithm also includes information on whether the assessment was validated; the more predictions it makes, the more accurate they are.
Human Data Science is creating unprecedented insight into the reality of patient care, physician behavior, and clinical decision making. So sales teams have the information they need to have better interactions with physicians, optimize their products, and even impact patient outcomes.
Exhibit 8: Disease progression data
Sales teams find the right moment to support physician decisions
The Art of Human Data Science
This gallery brings Human Data Science to life using Algorithmic Art. It was curated using data sets from around IQVIA.
Each piece of art in this gallery represents how we use data to answer questions and push healthcare forward. How to pick the right site for a clinical trial. How we can protect patient privacy to continue genomic research. How to better understand what works, and what doesn’t, in cancer care.
The Art of Human Data Science reminds us that there can be abstract beauty in discrete data, and objective evidence behind striking art.
Flip through to see this visual representation of Human Data Science using Algorithmic Art.
CONTINUE
CONTINUE