Lucas Glass, Gary Shorter, and Rajneesh Patil, IQVIA11.20.19
Artificial intelligence (AI) and machine learning (ML) tools are transforming how clinical development occurs by delivering significant time and cost efficiencies while providing better and faster insights to inform decision-making. Advances in analytics technology coupled with the availability and integration of vast amounts of healthcare data have already helped automate processes and improve data quality across dozens of clinical development efforts.
As these tools evolve, new opportunities will continue to emerge that drive further benefits to the clinical research landscape. Applications of AI and ML in healthcare are expected to grow to nearly $8 billion by 2022, up from $667.1 million in 2016, and almost half of global life science professionals say they are either using or interested in using AI in their research.1
Despite this growth, the industry continues to struggle with what these technologies are and how they work. And there is uncertainty on how to surmount the challenges required to leverage AI and ML.
When sponsors collaborate with partners that have the necessary technical and pharmaceutical expertise, they can achieve significant time and cost savings while reducing risks and improving the quality of their research. Here are five proven areas where AI and ML can directly impact operational efficiencies:
1. Study design
Poor study design has a catastrophic impact on the cost, efficiency and success potential of clinical trials. Leveraging vast healthcare data sets, AI, ML and natural language processing tools can be used to assess and select optimal primary and secondary endpoints during study design to ensure the most relevant protocols are defined for regulators, payers and patients. This helps to optimize the study design by informing ideal strategies for host countries and sites, enrollment models, patient recruitment and start-up plans.
Better study design leads to more predictable results, reduced cycle time for protocol development, fewer protocol amendments and higher efficiencies throughout a study. It also results in improved recruitment rates and fewer non-enrolling sites. These improvements facilitate realistic and accurate planning and increase chances of success.
2. Site identification and patient recruitment
Identifying trial sites that have access to enough patients who meet inclusion/exclusion criteria is an ongoing challenge. As studies target more specific populations, recruiting goals become even harder to achieve, which drives costs up, increases timelines and raises the risk of failure. According to Tufts Center for the Study of Drug Development, nearly half of all sites miss enrollment targets.
AI and ML can mitigate these risks by identifying and suggesting the sites with the highest recruitment potential and suggesting appropriate recruitment strategies. This involves mapping patient populations and proactively targeting sites with high predicted potential to deliver the most patients—before a single site is opened—and identifying the best avenues to recruit them. This means sponsors can open fewer sites, accelerate recruiting and reduce the risk of under-enrollment.
3. Pharmacovigilance
To ensure drug safety, massive amounts of structured and unstructured data must be integrated and reviewed. PV units that seek to harness the power of this data find that AI and ML technologies address many of the challenges they face by providing new levels of insight and predictive analytics. These tools can automate manual processing tasks and translate and digitize case safety reporting and adverse drug reaction documents. They can also monitor digital conversations on social media and other platforms to ensure that adverse events are promptly identified.
Natural language processing, Optical character recognition, and deep neural networks are being used to analyze and format structured and unstructured data for faster and more efficient safety reviews.
The insights derived from using AI for PV tasks lead to faster assessment of subject, site and study risks and overall study performance by domain experts. This allows project leads to increase efficiencies as well as patient safety.
4. Clinical monitoring
Tremendous manual effort is spent analyzing site risks and generating “action items” to mitigate those risks. AL and ML concepts can alleviate these pressures by assessing the risk environment and delivering predictive analytics to generate more effective clinical monitoring insights.
Advanced analytics provide composite site rankings for holistic risk assessments, allowing for more specific identification of risks and removal of false positives. Using the composite evaluation of site risks across the study quickly shows high-risk sites, key risk indicators and site risk rank. Evaluations can also be used to proactively identify which sites are more likely to have recruitment and performance issues, or which patients are at higher risk for potential AEs. These insights facilitate faster action and avoid potential problems.
5. Patient care
Disease detection algorithms are now being designed to leverage medical information, such as symptoms and procedures that typically precede a diagnosis, to identify patients who are very likely to develop diseases. This allows for proactive care, as well as new recruiting insights for prodromal or early-stage disease studies and those that require treatment-naive patients.
One area where this type of model is having a profound impact is Alzheimer’s disease research. Due to the exponential effects of a delayed AD diagnosis, much clinical research activity in AD is focused on the prodromal stage. Yet, traditional screenings for prodromal AD patients deliver only a 20 percent precision rate.
AI and ML: Augmenting the Future of Clinical Development
The clinical development landscape is changing quickly, and sponsors need to have the best insights to make the right decisions to increase predictability, reduce time to market and improve efficiencies. By using these methods in clinical development, we can introduce change to prove—and improve—how AI-derived insights can be applied to many aspects of development.
These technologies aren’t going to replace human expertise. But they are going to accelerate the ability to analyze the data and take meaningful action in response to create a safer, more streamlined research environment. As the volume and complexity of clinical data continue to increase and more real-world evidence is introduced, AI applications have great potential value for the clinical trial process and the pharmaceutical drug discovery industry.
Talent, Technology and Expertise
These technologies hold huge promise for clinical development, but that promise can only be achieved when organizations have the tools, talent and partners to leverage these technologies in relevant ways. To achieve the full potential of AI and ML in clinical development, pharma companies need partners who can provide:
The proven promise and huge potential of using AI and ML to accelerate drug discovery while cutting costs and risks is a tremendous catalyst for innovation. By continuing to leverage data, intelligence, analytics and domain expertise, there is an enormous opportunity for the industry to fundamentally transform the clinical development landscape in ways that will greatly benefit patients, sponsors, payers and physicians alike.
References
Lucas Glass is the Global Head of IQVIA’s Analytics Center of Excellence; responsible for researching, developing and operationalizing machine learning and data science solutions in the R&D business.
Gary Shorter, Head of Artificial Intelligence, Research & Development Solutions, IQVIA. Gary Shorter holds an MSc and has served as global biostatistics lead for multiple compounds in clinical trials.
Rajneesh Patil, Head of Process Design & Analytics, Centralized Monitoring Services, IQVIA. Rajneesh Patil’s expertise spans process design, project portfolio management, risk-based monitoring and advanced analytics for clinical trial applications.
As these tools evolve, new opportunities will continue to emerge that drive further benefits to the clinical research landscape. Applications of AI and ML in healthcare are expected to grow to nearly $8 billion by 2022, up from $667.1 million in 2016, and almost half of global life science professionals say they are either using or interested in using AI in their research.1
Despite this growth, the industry continues to struggle with what these technologies are and how they work. And there is uncertainty on how to surmount the challenges required to leverage AI and ML.
When sponsors collaborate with partners that have the necessary technical and pharmaceutical expertise, they can achieve significant time and cost savings while reducing risks and improving the quality of their research. Here are five proven areas where AI and ML can directly impact operational efficiencies:
1. Study design
Poor study design has a catastrophic impact on the cost, efficiency and success potential of clinical trials. Leveraging vast healthcare data sets, AI, ML and natural language processing tools can be used to assess and select optimal primary and secondary endpoints during study design to ensure the most relevant protocols are defined for regulators, payers and patients. This helps to optimize the study design by informing ideal strategies for host countries and sites, enrollment models, patient recruitment and start-up plans.
Better study design leads to more predictable results, reduced cycle time for protocol development, fewer protocol amendments and higher efficiencies throughout a study. It also results in improved recruitment rates and fewer non-enrolling sites. These improvements facilitate realistic and accurate planning and increase chances of success.
2. Site identification and patient recruitment
Identifying trial sites that have access to enough patients who meet inclusion/exclusion criteria is an ongoing challenge. As studies target more specific populations, recruiting goals become even harder to achieve, which drives costs up, increases timelines and raises the risk of failure. According to Tufts Center for the Study of Drug Development, nearly half of all sites miss enrollment targets.
AI and ML can mitigate these risks by identifying and suggesting the sites with the highest recruitment potential and suggesting appropriate recruitment strategies. This involves mapping patient populations and proactively targeting sites with high predicted potential to deliver the most patients—before a single site is opened—and identifying the best avenues to recruit them. This means sponsors can open fewer sites, accelerate recruiting and reduce the risk of under-enrollment.
3. Pharmacovigilance
To ensure drug safety, massive amounts of structured and unstructured data must be integrated and reviewed. PV units that seek to harness the power of this data find that AI and ML technologies address many of the challenges they face by providing new levels of insight and predictive analytics. These tools can automate manual processing tasks and translate and digitize case safety reporting and adverse drug reaction documents. They can also monitor digital conversations on social media and other platforms to ensure that adverse events are promptly identified.
Natural language processing, Optical character recognition, and deep neural networks are being used to analyze and format structured and unstructured data for faster and more efficient safety reviews.
The insights derived from using AI for PV tasks lead to faster assessment of subject, site and study risks and overall study performance by domain experts. This allows project leads to increase efficiencies as well as patient safety.
4. Clinical monitoring
Tremendous manual effort is spent analyzing site risks and generating “action items” to mitigate those risks. AL and ML concepts can alleviate these pressures by assessing the risk environment and delivering predictive analytics to generate more effective clinical monitoring insights.
Advanced analytics provide composite site rankings for holistic risk assessments, allowing for more specific identification of risks and removal of false positives. Using the composite evaluation of site risks across the study quickly shows high-risk sites, key risk indicators and site risk rank. Evaluations can also be used to proactively identify which sites are more likely to have recruitment and performance issues, or which patients are at higher risk for potential AEs. These insights facilitate faster action and avoid potential problems.
5. Patient care
Disease detection algorithms are now being designed to leverage medical information, such as symptoms and procedures that typically precede a diagnosis, to identify patients who are very likely to develop diseases. This allows for proactive care, as well as new recruiting insights for prodromal or early-stage disease studies and those that require treatment-naive patients.
One area where this type of model is having a profound impact is Alzheimer’s disease research. Due to the exponential effects of a delayed AD diagnosis, much clinical research activity in AD is focused on the prodromal stage. Yet, traditional screenings for prodromal AD patients deliver only a 20 percent precision rate.
AI and ML: Augmenting the Future of Clinical Development
The clinical development landscape is changing quickly, and sponsors need to have the best insights to make the right decisions to increase predictability, reduce time to market and improve efficiencies. By using these methods in clinical development, we can introduce change to prove—and improve—how AI-derived insights can be applied to many aspects of development.
These technologies aren’t going to replace human expertise. But they are going to accelerate the ability to analyze the data and take meaningful action in response to create a safer, more streamlined research environment. As the volume and complexity of clinical data continue to increase and more real-world evidence is introduced, AI applications have great potential value for the clinical trial process and the pharmaceutical drug discovery industry.
Talent, Technology and Expertise
These technologies hold huge promise for clinical development, but that promise can only be achieved when organizations have the tools, talent and partners to leverage these technologies in relevant ways. To achieve the full potential of AI and ML in clinical development, pharma companies need partners who can provide:
- Deep pharmaceutical knowledge and domain expertise. This expertise should include in-depth understanding of healthcare data for insightful analyses, understanding of global regulatory environments, payer expectations, physician and patient behavior and therapeutic knowledge.
- State-of-the-science AI and ML technologies. The solution should be capable of mining multiple data sets with speed and scale to identify high-level global and regional trends, as well as detailed physician- and patient-level insights while applying rigor through the GxP Software Development Life Cycle approach to produce top quality models.
- Data analytics and machine learning experts. The customer team should include technical experts who are skilled in crafting machine learning algorithms relevant to the clinical development process.
- Access to vast healthcare data sets. Machine learning algorithms are only as good as the data they can access. The most effective platforms will provide access to multiple global healthcare databases, including prescription data, EMRs, pharma sales data and patient and disease trend data that are updated regularly.
- The ability to integrate these disparate data sets. Many healthcare data sets are unstructured or inconsistent in form and formatting, making them difficult to analyze. A good partner will have strategies in place to clean the data, so the algorithms can interpret results and translate them into actionable insights that drive better results.
The proven promise and huge potential of using AI and ML to accelerate drug discovery while cutting costs and risks is a tremendous catalyst for innovation. By continuing to leverage data, intelligence, analytics and domain expertise, there is an enormous opportunity for the industry to fundamentally transform the clinical development landscape in ways that will greatly benefit patients, sponsors, payers and physicians alike.
References
- Source: Accenture, “Artificial Intelligence: Healthcare’s New Nervous System,” https://www.accenture.com/au-en/insight-artificial-intelligence-healthcare
Lucas Glass is the Global Head of IQVIA’s Analytics Center of Excellence; responsible for researching, developing and operationalizing machine learning and data science solutions in the R&D business.
Gary Shorter, Head of Artificial Intelligence, Research & Development Solutions, IQVIA. Gary Shorter holds an MSc and has served as global biostatistics lead for multiple compounds in clinical trials.
Rajneesh Patil, Head of Process Design & Analytics, Centralized Monitoring Services, IQVIA. Rajneesh Patil’s expertise spans process design, project portfolio management, risk-based monitoring and advanced analytics for clinical trial applications.
