Big Data Fuels Innovation

How Do We Define Big Data?

Most industries are collecting terabytes, petabytes, and even larger units of data. Compared to 10 or 20 years ago, the amount of data being collected is exploding. The data itself is nothing but a collection of binary numbers, so it’s up to those amassing the data to use it intelligently. The ability to extract useful information from data is crucial, regardless of the size of the data set.

In the area of clinical trial research, the data flow starts with collection of data at the clinical site. The data is then analyzed according to the protocol, and then shared with the sponsor and the regulatory agency such as the FDA. Clinical trials must focus on a few key measurements, but big data represents untapped potential with ancillary data points.

Big Data as the New Oil

A useful analogy when considering big data is one suggested by Kiran Bhageshpur in a Forbes Technology Council blog post in 2019. Bhageshpur quoted an Economist story titled “The world’s most valuable resource is no longer oil, but data.”¹ Bhageshpur concludes by stating, “. . . . the main impact on humanity isn’t how tech giants are monetizing our attention units, but about how data can improve our lives. . . . Oil causes pollution, yet it was also responsible for lifting a large majority of the population of the world out of dire poverty. We must manage the dark side of data, but the advances in data fuels are worth the effort.”¹

Examining the analogy more closely provides some further insights into big data. When you look at converting crude oil into gasoline, there are multiple steps that need to take place from drilling wells, to shipping the crude oil via trucks or pipelines, to refining the oil into gasoline. Data is similar in that there are several steps needed to prepare the data for analysis to make it useful and actionable.

This holds true for clinical trial data. Drug developers can say a particular drug works to cure or effectively treat a particular disease based on certain evidence, namely data exists that shows a particular effect of the drug on the disease. For example, in ophthalmology, there could be a wet AMD drug that is applied to the eye that would need to demonstrate that a subject’s visual acuity has been enhanced following the treatment. Thus, the visual acuity must be accurately measured. Another example of structural biomarker to evaluate the treatment effectiveness in wet AMD would be the central subfield retinal thickness measurement obtained from OCT. This structural endpoint is required to be measured accurately, across different data formats and characteristics from various OCT manufacturers, to prove effectiveness of the drug and treatment.

This is the type of data that can form the basis for big data analyses, but as with getting from crude oil to gasoline, there are some crucial steps that come before the final product is available. Before the data can be analyzed it must be standardized and cleaned, which is equally important as the actual analysis of the data. High quality data is the foundation of actionable analysis.

Big Data and AI

AI has become an intrinsic part of big data and is one of its greatest potential benefits. The prevalence of AI has expanded in recent years due to the large amounts of data collected and as the deep learning model is fueled by exponentially expanding computing power.

In the world of ophthalmology clinical practice, AI can assist with diagnosis, and in the field of image analysis for clinical trials it offers several possible benefits as well:

1. AI can increase the accuracy, efficiency, and effectiveness of measurements. In terms of image analysis, computers can now provide segmentation of different layers, measurement of lesions, and automatic detection of lesions.
2. AI can be used to discover additional, non-intuitive biomarkers. An example of an intuitive biomarker for diabetic patients would be that they are known to have certain indications in the eye (specific lesion types in the retina such as hemorrhage) that are indicative of the progression of the patient’s diabetes.
   • Alzheimer’s: Something less evident might be Alzheimer’s patients who might show certain parameters in the retina in the retinal nerves or ganglion cells that are correlated with the disease progression of Alzheimer’s.
   • Hypertension: Another example would be examining the images from 5,000 patients with recorded hypertension to detect if there is a correlated biomarker in the eye.
3. AI is often used for genomic analysis or what is sometimes called personalized medicine. This is very common in cancer treatments. Certain therapies are only effective for patients with a specific genetic mutation. Image analysis coupled with genomic analysis may help to find subjects that will respond better to certain treatments.

Turn Your Big Data into Actionable Results WITH MERIT

MERIT’s suite of services and technology empowers sponsors to turn big data into actionable results. Our proprietary cloud-based software, EXCELSIOR™, offers the following differentiators:

• Effective and instantaneous transfer of data from one location to another. Image data can be directly and automatically moved to the sponsor’s data warehouse for further analysis, including AI.
• Access to the data in real-time.
• No geographic limitations. With cloud-based data infrastructure, data can be viewed and analyzed anywhere. Given the global nature of clinical trials, data is collected from multiple geographic locations.
• Latest web and imaging technologies for data cleaning, standardization, analysis, grading and reporting.
• EXCELSIOR is HIPAA and 21 CFR part 11 compliant and is cleared with the FDA as a class II medical device (K130453), with a specific indication for use in managing ophthalmic clinical trials.

Connect with us to learn more about how our expertise and approach can support bringing your product to market on-time and on-budget. Your success is our priority.