Big data can be described as the 7-Vs: Volume, Velocity, Variety, Variability, Veracity, Value, and Visualization. At Quantori, we continually grapple with many of these challenges — but the one that can be a particular challenge is data volume. Below, we will discuss our strategy for acquiring more data to unlock valuable scientific insights. Spoiler alert: It all comes down to knowing where to look.

Recently, we encountered a business case that initially seemed quite common. One of Quantori's clients, a biopharma company, approached us for assistance with data analysis as they delved into the field of Neurology. They inquired whether we could apply a unique AI/ML approach to analyze patient data before the launch of a new pre-clinical program. Since multimodal data analysis is one of our core specialties, we entered the meeting with high hopes of a quick and easy project. That’s when they revealed they had data for at most 20 patients. 

That was a problem. For meaningful insights, datasets often require hundreds, if not thousands, of patient data points — even if the data is genomic or proteomic, we need representation from larger patient groups to find meaningful clusters and trends.

While this posed a challenge for traditional analysis, it paved the way for the introduction of Quantori’s Data Landscaping practice. Further, we will explain what is unique about this approach and how it could handle the challenge of a limited dataset. 

To resolve this issue, our team started to search for global datasets that tend to be open-source or available at easy terms. Some of these datasets included multimodal data — data on the same patient but in different modalities — such as EHR, imaging, and NGS data. The trick was knowing where they reside, getting permission to download, and seamlessly integrating them into a client’s internal organization to combine them with client data. 

For our biopharma client, we located more than 50 diverse datasets representing over 34,000 patients, of which 10% were diagnosed with the disease stage targeted for clinical research.

While there were some data sets behind paywalls, more than 30 were in the public domain and freely available once the right data use agreements were signed. In summary, this approach expanded the client's initial dataset of 20 cases rather dramatically. They now have more than enough cases for their preliminary research — as well as specialty data sets that can help them in additional hypothesis generation. 

The work on that project didn’t stop there. After incorporating the enriched dataset, Quantori’s Data Analytics team deployed the QFLOW, our advanced machine learning system, to prepare the data for analysis and discover new insights across the data sets the company sought. 

Takeover:

Consider Quantori Data Landscaping Practice to maximize the potential of your AI efforts when researching new disease states. It could be the difference between a dead end and progress on your pre-clinical or clinical studies.

For more information about Quantori’s Data Landscaping Team, please contact Dr. Steven Labkoff at steven.labkoff@quantori.com