It is critical to acquire the right data to deliver the right insights to drive the right actions in order to shape the future of clinical monitoring.
Clinical trials are undergoing digital transformation in every aspect. Over the past few years and further accelerated by the COVID-19 pandemic, Risk Based Quality Management (RBQM) – in which trial management can adjust and allocate monitoring resources based on the level of risk identified for a specific trial or site – has utilized this digital evolution to emerge as a more efficient way to facilitate trial delivery. RBQM strategies optimize data flows, increase oversight, enhance data quality, and improve subject safety.
Vast amounts of data are collected through clinical studies. A lot of time and effort is spent sourcing this data from disparate systems, as well as standardizing and cleaning it for analytics, rather than generating the wealth of insights this data could provide if companies have the time to comb through it. However, the focus always needs to be on quality and efficiency. Advanced analytics tools using artificial intelligence (AI) and machine learning (ML) enable research teams to make sense of this information. To keep up with this digital transformation, sponsors need better data management technologies that utilize AI and ML to not only accelerate data collection but also streamline the review and decision-making process.
It is critical to acquire the right data to deliver the right insights to drive the right actions.
The Right Data
One of the biggest challenges in the realm of clinical trials is data spread across disparate systems with minimal interoperability. Point solutions hinder centralized access to data and restrict efficiency in decision-making. This makes it difficult to scale when organizations grow. Trial data captured in disparate systems prevent a cohesive view of the subject. To fully review the critical data at a subject level, monitoring teams must have access to a single data profile of the subject – sourced from several systems.
The ideal solution should be able to integrate disparate data sources such as EDC, CTMS, eCOA, labs, connected devices, IRT, and others in real-time. Once the data is curated, cleaned, harmonized, and housed quickly in a single repository for stakeholder use, it is easier to use custom artificial intelligence models and machine learning algorithms to connect the dots across different point solutions and make holistic assessments to identify data errors, outliers, and false entries. The longer it takes to generate the insights, the bigger the risk becomes. This in turn, enables more informed decision-making and reduces trial costs by enabling the team to focus on operational and strategic business processes. The learnings from the historical trials could be used to teach algorithms, which could be leveraged at a program or port.
The Right Insights
RBQM relies on access to the “right data” to systematically generate the “right insights,” enabling oversight, risk mitigation and prevention, and issue resolution with respect to sites and subjects.
Early generations of these tools and algorithms focused on single parameters to identify issues, which, out of context, often led to false positives. Advanced analytics have improved the user’s ability to apply tribal knowledge and contextual information to generate more accurate insights.
For example, two sites – one with 1 subject enrolled and the other with 150 subjects enrolled – were historically flagged to be at similar risk for not reporting, or underreporting, adverse events irrespective of the number of subjects enrolled in each of the sites or the time for which the subjects have been exposed to the drug. Hence, Central Monitors (CMs) and Clinical Research Associates (CRAs) spent a lot of their time responding to triggers that may not have required much action. However, advanced analytics models can differentiate this relative risk between the two sites and indicate that the second site in this example is at higher risk. These normalization factors applied with trained and validated models in past studies have benefited the overall accuracy and effectiveness of trial oversight. CMs/CRAs now do not have to split their time evenly between every site looking for anomalies that often didn’t exist. Algorithms provide a list of potential outliers, which can be supplemented, validated, and prioritized by the stakeholders. Systemic observations with a particular issue can be used to reprioritize risks based on the study design.
Over the past few years, further enhancements have been made to these models by incorporating ML, which uses algorithms and decision support systems to learn from past searches, making the system ‘smarter’ with every review. This learning isn’t just limited to the current trial. These platforms are able to learn from historic clinical data sets, making them increasingly ‘knowledgeable’ with every application and resource. That translates to greater oversight, higher-quality data, and a safer trial environment. It also means sponsors can utilize clinical research staff more efficiently and focus on more value-added tasks, such as ensuring sites are operating within trial parameters and compliance measures.
The Right Actions
By automating the data collection, cleaning, and review process, operational stakeholders are free to focus on more value-added tasks, and sponsors and sites get faster access to information. It also brings greater efficiency and job satisfaction.
AI algorithms, coupled with workflow automation, have resulted in more proactive risk management. The algorithms can help in reviewing the appropriate insights and flag risks automatically. These insights can then be used by the stakeholders to verify the risks and take the appropriate action, freeing monitors and site staff to focus on solutions, rather than reviewing all data and then deciding whether a risk warrants an action or not.
Each issue receives the necessary level of scrutiny, resulting in the most optimal corrective and preventive actions that ensure the identified issue is mitigated and a similar issue does not repeat. The primary value drivers in terms of time savings are the prevention of repeated issues leading to a reduced time to clean, lock, and submit the data.
Shaping The Future
The overall success of an RBQM implementation depends on its ability to highlight high-priority areas that require immediate attention. The effective techniques of AI and ML are a sponsor’s best chance to rapidly achieve goals, employing advanced algorithms and statistical models for faster and more informed decision-making. With an increased demand for unsupervised learning, AI/ML has been leveraged for risk identification in multiple ways.
A few use cases include:
- Holistic site risk assessment using composite risk indicators for identification and monitoring of high-risk sites
- Monitoring high-risk subjects through early signal detection of subject outliers for Labs and Vital Signs
- Predicting Protocol Deviations through historical trending allows the monitoring team to proactively implement mitigation actions
- Identification of duplicate or professional subjects with workflow alerting
Technology as an Enabler
The technology landscape is continuously changing with the advent of mobile health (mHealth), wearable technologies, connected devices, and improved analytics. Accessibility to more patient data flowing from newer data sources could potentially eliminate the need for source verification. The need for CRAs to spend time onsite for data review is being replaced by centralized data review processes, as evidenced by the recent implementations of RBQM models.
The key areas of focus for technology strategy could be:
- Advanced Analytics including implementation of machine learning and artificial intelligence
- Internet of Things platforms enabling the use of wearables technologies, connected devices, and smartphones
- Automation of workflows with suggested actions for quick issue escalation and management
- Prioritization of risks based on system-generated insights from historical and current trial information
Changing the Role of a CRA
The role of a traditional CRA has gone through a series of changes since the advent of RBQM. This shift in focus to the most critical risk identification means that the CRA has to manage compliance and relationships at the site. The role of the CRA would be geared more towards the safety of the patients rather than actions and mitigations that could be deployed remotely.
The paradigm shift in considering targeted monitoring that requires visits to the site based on risk data, instead of the traditional approach of reviewing all data on-site, can be attributed to improved ability in remote communication with site personnel through the different virtual channels. CRAs need to address complex issues virtually through a hybrid approach, which places a demand for a different set of communication skills from the CRAs.
With access to data and analytics, the CRAs must be increasingly analytical with the interpretation of the data and insight to identify outlier trends. This entails the need for additional training, not only for the CRAs to embrace data and insights but also to prepare the stakeholders on the ground to be ready to use them. Critical thinking is crucial to the management of the site’s quality and maintaining patient safety.
Accelerating Innovation
In summary, the current successes and enormous potential of using technology, AI, and ML to accelerate drug discovery while cutting costs and risks is a tremendous catalyst for innovation. By continuing to leverage data, digital intelligence, analytics, and domain expertise, there is a huge opportunity for the industry to fundamentally transform the clinical development landscape in ways that will greatly benefit trial participants, sites, and sponsors.
Visit AITechPark for cutting-edge Tech Trends around AI, ML, Cybersecurity, along with AITech News, and timely updates from industry professionals!
