Jay Ferro is the Chief Information, Technology and Product Officer at Clario, he has over 25 years of experience leading Information Technology and Product teams, with a strong focus on data protection and a passion for creating technologies and products that make a meaningful impact.
Before joining Clario, Jay held senior leadership roles, including CIO, CTO, and CPO, at global organizations such as the Quikrete Companies and the American Cancer Society. He is also a member of the Board of Directors at Allata, LLC. His professional accomplishments have been recognized multiple times, including awards from Atlanta Technology Professionals as Executive Leader of the Year and HMG Strategy as Mid-Cap CIO of the Year.
Clario is a leader in clinical trial management, offering comprehensive endpoint technologies to transform lives through reliable and precise evidence generation. Specializing in oncology trials, Clario emphasizes patient-reported outcomes (PROs) to enhance efficacy, ensure safety, and improve quality of life, advocating for electronic PROs as a more cost-effective alternative to paper. With expertise spanning therapeutic areas and global regulatory compliance, Clario supports decentralized, hybrid, and site-based trials in over 100 countries, leveraging advanced technologies like artificial intelligence and connected devices. Their solutions streamline trial processes, ensuring compliance and retention through integrated support and training for patients and sponsors alike.
Clario has integrated over 30 AI models across various stages of clinical trials. Could you provide examples of how these models enhance specific aspects of trials, such as oncology or cardiology?
We use our AI models to deliver speed, quality, precision and privacy to our customers in more than 800 clinical trials. I’m proud that our tools aren’t just part of the AI hype cycle – they’re delivering real value to our customers in those trials.
Today, our AI models largely fall into four categories: data privacy, quality control assistance, read assistance and read analysis. For example, we have tools in medical imaging that can automatically redact Personally Identifiable Information (PII) in static images, videos or PDFs. We also employ AI tools that deliver data with rapid quality assessments at the time of upload — so there’s a lot of confidence in that data. We’ve developed a tool that monitors ECG data continuously for signal quality, and another that confirms correct patient identifiers. We’ve developed a read-assist tool that enables slice prediction, lesion propagation and disease detection. Additionally, we’ve improved read analysis by automating and standardizing data interpretation with tools like AI-supported quantitative ulcerative colitis Mayo scoring.
Those are just a few examples of the types of AI models we’ve been developing since 2018, and while we’ve made lots of progress, we’re just getting started.
How does Clario ensure that AI-driven insights maintain high accuracy and consistency across diverse trial environments?
We are constantly training our AI models on vast amounts of data to understand the difference between good data and data that is not good or relevant. As a result, our AI-driven data analysis detects, pre-analyzes rich data histories, and ultimately leads to higher quality results for our customers.
Our spirometry solutions nicely illustrate why we do that. Clinicians use spirometry to help diagnose and monitor certain lung conditions by measuring how much air a patient can breathe out in one forced breath. There are a variety of errors that can occur when a patient uses a spirometer. They might perform the test too slowly, cough during testing, or not be able to make a complete seal around the spirometer’s mouthpiece. Any of those variabilities can cause an error that might not be discovered until a human can analyze the results. We’ve trained deep learning models on more than 50,000 examples to learn the difference between a good reading and a bad reading. With our devices and algorithms, clinicians can see the value of the data in near real-time rather than having to wait for human analysis. That matters in part because some patients might have to drive several hours to participate in a clinical trial. Imagine driving that distance home from the site only to learn you’re going to have to take another spirometry test the following week because the first one showed an error. Our AI models are delivering accurate overreads while the patient is still at the site. If there’s an error, it can be rectified on the spot. It’s just one of the ways we’re working to reduce the burden on sites and patients.
Could you elaborate on how Clario’s AI models reduce data collection times without compromising data quality?
Generating the highest quality data for clinical trials is always our focus, but the nature of our AI algorithms means the capture and analysis is sped up dramatically. As I mentioned, our algorithms allow us to conduct quality control analysis faster and at a higher level of precision than human interpretation. They also allow us to conduct quality checks as data are entered. That means we can identify missing, erroneous or poor-quality patient data while the patient is still at the trial site, rather than letting them know days or weeks later.
How does Clario address the challenges of decentralized and hybrid trials, especially in terms of data privacy, patient engagement, and data quality?
These days, a decentralized trial is really just a trial with a hybrid component. I think the concept of letting participants use their own devices or connected devices at home really opens the door to greater possibilities in trials, especially in terms of accessibility. Making trials easier to participate in is a key focus of our technology roadmap, which aims to develop solutions that improve patient diversity, streamline recruitment and retention, increase convenience for participants, and expand opportunities for more inclusive clinical trials. We offer at-home spirometry, home blood pressure, eCOA, and other solutions that deliver the same data integrity as more traditional solutions, and we do it in concert with oversight from our endpoint and therapeutic area experts. The result is a better patient experience for better endpoint data.
What unique advantages does Clario’s AI-driven approach offer to reduce trial timelines and costs for pharmaceutical, biotech, and medical device companies?
We’ve been developing AI tools since 2018, and they’ve permeated everything we’re doing internally and certainly across our product mix. And what has never left us is making sure that we’re doing it in a responsible way: keeping humans in the loop, partnering with regulators, partnering with our customers, and including our legal, privacy, and science teams to make sure we’re doing everything the right way.
Responsibly developing and deploying AI should affect our customers in a variety of positive ways. The foundation of our AI program is built on what we believe to be the industry’s first Responsible Use Principles. Anyone at Clario who touches AI follows those five principles. Among them, we take every measure to ensure we’re using the most diverse data available to train our algorithms. We monitor and test to detect and mitigate risks, and we only use anonymized data to train models and algorithms. When we apply those kinds of guidelines when developing a new AI tool, we’re able to rapidly deliver precise data – at scale – that reduces bias, increases diversity and protects patient privacy. The faster we can get sponsors accurate data, the more impact it has on their bottom line and, ultimately, patient outcomes.
AI models can sometimes reflect biases inherent in the data. What measures does Clario take to ensure fair and unbiased data analysis in trials?
We know bias occurs when the training data set is too limited for its intended use. Initially, the data set might seem sufficient, but when the end user starts using the tool and pushes the AI beyond what it was trained to respond to, it can lead to errors. Clario’s Chief Medical Officer, Dr. Todd Rudo, sometimes uses this example: We can train a model to determine proper lead placement in electrocardiograms (ECGs) so clinicians can tell if technicians have put the leads in the proper places on the patient’s body. We’ve got tons of great data so we can train that model on 100,000 ECGs. But what happens if we only train our AI model using data from adult tests? How will the model react if an ECG is done on a 2-year-old patient? Clearly it could potentially miss errors that have an impact on treatment.
That’s why at Clario, our product, data, R&D, and science teams all work closely together to ensure that we’re using the most comprehensive training data to ensure accuracy and reliability in real-world applications. We use the most diverse data available to train the algorithms incorporated into our products. It’s also why we insist on using human oversight to mitigate risks during the development and use of AI.
How does Clario’s human oversight and monitoring process integrate with AI outputs to ensure regulatory compliance and ethical standards?
Human oversight means we have teams of humans who know exactly how our models are developed, trained and validated. Both in development and after we’ve integrated a model into a technology, our experts monitor outputs to detect potential bias and ensure the outputs are fair and reliable. I believe AI is about augmenting science and human brilliance. AI gives humans the ability to focus on a higher level of challenge. We are remarkably good at solving problems and still much better at intuition and nuance than machines. At Clario, we use AI to remove the burden on repeatable things. We use it to analyze broad data sets, whether it’s patient images or prior trials or any other thing that we want to analyze. Generally, machines can do that faster, and in some cases, better than humans can. But they can’t replace human intuition and the science and real-world experience that the wonderful people in our industry have.
How do you foresee AI impacting clinical trials over the next few years, particularly in fields like oncology, cardiology, and respiratory studies?
In oncology, I’m excited about advancing the use of applied AI in radiomics, which extracts quantitative metrics from medical images. Radiomics involves several steps, including image acquisition of tumors, image preprocessing, feature extraction, and model development, followed by validation and clinical application. Using increasingly advanced AI, we will be able to predict tumor behavior, tailor treatment response, and foresee patient outcomes based non-invasive imaging of tumors. We’ll be able to use it to detect early signs of disease and early detection of disease recurrence. As more advanced AI tools become more integrated into radiomics and clinical workflows, we’re going to see huge strides in oncology and patient care.
I’m equally excited about the future of respiratory studies. This past year, we acquired ArtiQ, a Belgian company that built AI models to improve the collection of respiratory data in clinical trials. Their founder is now my Chief AI Officer, and we’re expecting big things in respiratory solutions. Our approach to algorithm application has become a game-changer, not least because it’s helping reduce patient and site burden. When exhalation data isn’t analyzed in real time, and an anomaly is detected later, it forces the patient to come back to the clinic for another test. This not only adds stress for the patient, but it can also create delays and additional costs for the trial sponsor, and that leads to various operational challenges. Our new spirometry devices leverage the ArtiQ models to address that burden by offering near real-time overreads. That means if any issues occur, they are identified and resolved immediately while the patient is still at the clinic.
Finally, we’re developing tools that will have an impact across therapeutic areas. Soon, for example, we’ll see AI deliver increasingly more value in electronic clinical outcomes assessments (eCOA). We’ll see AI models that capture and measure subtle changes experienced by the patient. This technology will help a multitude of researchers, but for example, Alzheimer’s researchers will be able to understand where the patient is in the stage of the disease. With that kind of knowledge, drug efficacy can be better gauged while patients and their caretakers can be better prepared for managing the disease.
What role do you believe AI will play in expanding diversity within clinical trials and improving health equity across patient populations?
If you only look at AI through a tech lens, I think you get into trouble. AI needs to be approached from all angles: tech, science, regulatory and so on. In our industry, true excellence is achieved only through human collaboration, which expands the ability to ask the right questions, such as: “Are we training models that take into consideration age, gender, sex, race and ethnicity?” If everyone else in our industry asks these types of questions before developing tools, AI won’t just accelerate drug development, it will accelerate it for all patient populations.
Could you share Clario’s plans or predictions for the evolution of AI in the clinical trials sector in 2025 and beyond?
In 2025, we’re set to see biopharma leverage AI and real-time analytics like never before. These advancements will streamline clinical trials and enhance decision-making. By speeding up study builds and implementing risk-based monitoring, we’ll be able to accelerate timelines, ease the burden on patients, and enable sponsors to deliver life-saving treatments with greater precision and efficiency. This is an exciting time for all of us, as we work together to transform healthcare.
Thank you for the great interview, readers who wish to learn more should visit Clario.
Credit: Source link
