Featured Product
This Week in Quality Digest Live
FDA Compliance Features
Kari Miller
An effective strategy requires recruiting qualified personnel familiar with the process and technology
Dario Lirio
Modernization is critical to enhance patient experience and boost clinical trial productivity
Alexander Khomich
Healthcare software opens up opportunities for clinics in both management and patient care
Gary Shorter
Pharma needs to adapt and evolve with the changing environment of life science data
Etienne Nichols
Quality management system regulation explained

More Features

FDA Compliance News
Delivers time, cost, and efficiency savings while streamlining compliance activity
First trial module of learning tool focuses on ISO 9001 and is available now
Free education source for global medical device community
Good quality is adding an average of 11 percent to organizations’ revenue growth
Further enhances change management capabilities
Creates adaptive system for managing product development and post-market quality for devices with software elements
VQIP allows for expedited review and importation for approved applicants that demonstrate safe supply chains
An invite from Alcon Laboratories
Intended to harmonize domestic and international requirements

More News

Gary Shorter

FDA Compliance

Four Steps to Accelerating Discoveries in Life Sciences

Pharma needs to adapt and evolve with the changing environment of life science data

Published: Monday, May 9, 2022 - 12:03

Predictive and prescriptive insights driven by data analytics have risen to prominence as tools that can help research teams cut the time, complexity, and cost of clinical trials. At the same time, these insights can enhance the quality of a study and accelerate new drugs to market. But to uncover these insights, we need to rethink how data management and research get done.

While clinical trials used to center around the electronic capture of data, the introduction of decentralized clinical trials (DCTs) allowed researchers to gather data from more sources in real time. Approximately 1,300 DCTs will kick off in 2022, a 28 percent increase from last year, and a 93 percent increase over 2020. Data management used to be enough. But with more data coming from more sources, faster, in more formats, and with varying levels of quality, clinicians now face an imperative to use data science to parse and organize the information for better insights and outcomes.

To successfully manage this evolution, clinical trial providers need to leverage data science expertise as well as technology that supports more rapid data collection and is equipped with artificial intelligence (AI) and machine learning (ML).

Here are four key steps researchers should take to successfully run DCTs and embrace the new normal of clinical research.

1. Acquire and understand your data

Nowadays, healthcare data come from watches, smartphones, connected medical devices, and web applications as well as traditional sources such as medical records and electronic data capture for clinical trials. With more variability in the data across the globe, and without a central data lake repository, clinical researchers need new AI and ML tools to help them automate and understand the varying structures while complying with regulatory standards. These technologies should enable fast support and efficient data gathering while maintaining good audit trails that can be tracked from anywhere, anytime, around the world.

For example, if a DCT was held in multiple countries, some of them might have different regulatory requirements for handling patient data. Therefore, it’s important to be able to collect data within the required country, identify them automatically, and push them through the relevant workflow for collecting and storing locally. In addition, it’s vital that key anonymized data be extracted for pooled analysis with the appropriate regulatory rigor.

Maintaining standards centrally is key to integrating data from all sources and supporting regional regulatory requirements. AI and ML support these efforts by providing good data throughput while retaining a high level of efficiency despite the ever-increasing variability in the data sources and regulatory rigor required today.

2. Ensure data standardization and quality

With data flowing into a centralized repository, AI models can automatically standardize the data and ensure their quality. These algorithms can also automatically map data into industry-standard data models, making it easier to query the data and improve the quality without much human intervention.

During this process, clinical researchers need to have audit trails available and be able to explain the AI algorithms to ensure they’re getting quality results from trials. Taking the time to do this can help guarantee that the AI is providing accurate results and validate that the data weren’t changed somewhere during the process. Audit trails ensure your process is repeatable and accurate, verifying the results of the trial.

3. Add artificial intelligence and machine learning

Once a helpful level of automation has been achieved, researchers can then introduce machine learning models that will analyze the collected data for predictive and prescriptive insights. These insights can be pushed to other areas of the business with reusable “plug-and-play” models. The models can classify how patients are responding during the trials, highlight any outliers, and identify variables that may correlate with differing results.

The data being analyzed shouldn’t be limited just to what’s collected during the trials. Researchers should also gather population data prior to the trials to help them determine the best locations to hold them.

For example, if you’re a researcher testing a new hepatitis B drug, you’ll want to gather population health data to identify outbreaks of hepatitis B and then use transportation data to determine how easy it will be for prospective patients to get to your site, either by taking public transportation, driving themselves, or hiring an Uber driver. AI and ML can help you think through all of these aspects to pick locations that are likely to provide you with the highest level of success.

4. Integrate human feedback

Rather than aiming for complete automation, clinical trials should use AI and ML to augment and optimize their trials. As helpful as AI and ML algorithms can be, without human feedback and oversight they can potentially provide false positives or biased answers. To get the best use of artificial intelligence, researchers must include oversight from data scientists to validate the algorithms and their outcomes.

AI and ML make clinical trials more efficient, but they can’t replace the need for human data scientists. Instead, they enable researchers to focus on insights and results rather than the repetitive and manual tasks associated with data collection and management.

Data science is critical to successful DCT

The pharmaceutical industry needs to adapt and evolve with the ever-changing environment of life science data. While data management used to be the norm, data science is key to running successful DCTs. Data science provides deeper insights before, during, and after trials to help researchers get the most accurate results from their data. By adding AI and ML algorithms to their data management tools, clinicians can automate the repetitive, manual tasks that take time away from analyzing results.

As DCTs gain popularity, researchers will have to use artificial and augmented intelligence to make better decisions about their trials, organize and standardize the large volumes of data they gather, and provide predictive insights about their results. By selecting platforms that already include AI and ML, clinicians can lower their time to value and gain better insights faster.

Discuss

About The Author

Gary Shorter’s picture

Gary Shorter

Gary Shorter is head of artificial intelligence and data science at IQVIA. He has been pursuing the use of emerging technologies to enhance and transform clinical trial management at IQVIA and Quintiles for more than 20 years. He has built out a global team of more than 50 technologists researching and developing AI/ML capabilities and creating products and services that can plug into industry SaaS solutions to make them more efficient. Gary is a subject matter expert on ethical AI and participates regularly in C-Suite discussions on the topic.