Digital: AI and ML

The use of artificial intelligence is becoming increasingly common in life sciences R&D. In what areas can it be utilised in the most helpful manner?

Milan Bhatt at Hexaware

The life sciences industry is turning a new page with the integration of artificial intelligence (AI) technologies. Arguably, one of the areas most impacted by AI is clinical R&D, where it brings significant value to clinical teams. These teams, which include functions such as data management, programming and statistics, safety, medical monitoring, medical writing, medical affairs and clinical operations, are seeing firsthand how AI is reshaping their work.

Clinical trials are complex by nature, involving multiple phases with their own set of challenges. AI concepts, such as intelligent agents and generative AI (genAI), are being leveraged to enhance the efficiency, accuracy and productivity of clinical teams at every phase. These technologies can automate routine tasks and provide real-time insights, freeing up human resources to focus on more strategic activities.

As we are still in the early stages of the AI revolution, it can be hard to pin down exactly how it will impact organisations. Many are still questioning how the use cases will solidify, and what outcomes they can expect to see once they do. Below are eight key teams in life sciences clinical R&D where AI usage will have an impact:

Data management is a critical function in clinical development, responsible for collecting, standardising and ensuring the study data is of high quality and devoid of any discrepancies. These teams collect data daily, using manual exports from electronic data capture (EDC) systems and merging it with other study sources like labs, biomarkers, and electronic patient-reported outcomes (ePRO) and electronic clinical outcome assessment (eCOA) systems.

• Automation of validation rules:

Setting up validation rules is essential to ensure that data meets a study’s

management plan and specific protocol requirements. But manually programming these rules can take a lot of time, especially when criteria are constantly changing. Automating this process helps streamline review cycles, making it easier to lock the database faster and reach important trial milestones on time

• Automating data review:

Traditionally, data managers perform manual reviews to check for missing, inconsistent or duplicate data. AI-powered automation can significantly streamline this process. It can reduce the manual effort required for data cleaning and validation, leading to reduced operational costs

• Anomaly detection:

Spotting unusual data patterns or outliers is key for quality control. Instead of relying on reactive, manual reviews, AI can catch these issues in real time. This proactive approach helps fix discrepancies before they snowball, reducing costly rework down the line.

Biostatistics and programming teams have to generate different data sets from data collected for a study, and ensure the samples comply with standards that are set by the Clinical Data Interchange Standards Consortium (CDISC) and other regulatory bodies. AI can help these teams with:

• Study data tabulation model (SDTM), analysis data model (ADaM) and tables, figures and listings (TFL) generation:

AI can automate the creation SDTM, ADaM and TFL data sets, ensuring compliance with regulatory standards and reducing manual effort and time

• Accelerated submission packets:

AI can streamline the creation of submission packets, expediting the approval process for both preclinical and clinical studies. Automation can increase the speed of statistical analysis and cut operational costs.

Ensuring patient safety is critical in clinical trials, particularly for detecting, assessing and preventing adverse drug-related events. AI significantly enhances pharmacovigilance by automating adverse event detection, signal detection and risk mitigation:

• Adverse event detection:

Natural language processing (NLP) algorithms can process clinical reports, electronic health records and ePROs in real time to identify potential adverse events (AEs). Early AE detection reduces patient risk and costs, improving outcomes and compliance

• Signal detection:

AI analyses large data sets to detect safety signals early, predicting AEs based on patient demographics, comorbidities and drug interactions for personalised safety monitoring. This can reduce pharmacovigilance monitoring time, enabling teams to focus on complex tasks.

Medical monitoring involves continuous oversight of clinical trial data to ensure patient safety and data integrity. A medical monitor, like a data manager, will be reviewing clinical data daily, looking at the clinical aspects of the data versus the correctness of data collection/data entry. AI can enhance medical monitoring in three key ways:

• Real-time monitoring:

AI algorithms can continuously monitor patient data, providing real-time alerts for any deviations from expected outcomes. This can reduce the time to identify and address issues and the need for manual monitoring, which can translate into reduced operational costs

• Predictive analytics:

Machine learning (ML) models can predict potential safety issues, enabling timely interventions

• Enhanced data integrity:

AI can ensure the accuracy and completeness of clinical data, reducing the risk of data discrepancies.

Medical writing

Medical writing can be time-consuming, from drafting clinical study reports to regulatory documents and scientific articles. But AI is changing the game by taking on much of the heavy lifting, helping teams create, edit and format content faster, with humans stepping in for the final touches.

• Automated content generation:

Generative models like GPT-4 can draft initial versions of study reports and regulatory documents, freeing up time and letting teams focus on refining instead of starting from scratch

• Editing and formatting:

AI can handle the tedious parts, like editing and formatting, to meet regulatory standards. This not only saves time but also cuts down on costs by reducing the manual effort involved

• Enhanced accuracy:

AI cross-references data from various sources to keep content accurate and consistent, giving teams confidence that everything checks out.

Medical affairs teams play a crucial role in bridging the gap between clinical development and commercialisation. They are the teams interfacing with healthcare practitioners (HCPs) and educating them about the science of the drug, so they rely on having insights from the scientific literature to hand.

• KOL identification:

AI can analyse vast data sets to identify key opinion leaders (KOLs) based on their influence, expertise and contributions to the field. This can reduce the time required to identify KOLs and extract insights from the literature

• Insights from scientific literature:

AI-powered semantic engines can extract relevant insights from a plethora of scientific publications, providing actionable intelligence. Access to real-time, comprehensive insights can improve strategic decision-making and potentially increase the success rate of product launches.

Clinical operations teams, particularly clinical research associates (CRAs), play a vital role in managing the logistics and execution of clinical trials. AI can act as a co-pilot for CRAs, helping them perform their tasks more efficiently and effectively. These teams can use AI for:

• Site selection and monitoring:

AI can analyse historical data to identify the best sites for clinical trials and predict potential issues, ensuring optimal selection and monitoring. This can greatly reduce the time required for site selection, monitoring and visit scheduling

• Visit scheduling and tracking:

Intelligent agents can automate the scheduling and tracking of site visits, reducing administrative burden and ensuring timely execution. Real-time data collection and reporting can enhance compliance with regulatory requirements, reducing the risk of costly delays.

Compliance in clinical trials isn’t just important, it’s essential. AI can give regulatory teams a serious boost in keeping everything on track. Here’s what that looks like:

• Automated regulatory submissions:

AI helps make sure everything is accurate and up to global standards, speeding things up and keeping submissions in line with the latest regulations

• Advertising and promotion compliance:

AI can also handle ad reviews, insuring those promotional materials meet all the right guidelines – local and international. With real-time updates and checks, there’s less risk of slipping up, and it helps avoid costly delays and fines

• Compliance monitoring:

Regulations change fast, and AI is constantly monitoring those updates. This keeps teams in the loop during trials and after products are on the market

• Risk assessment and mitigation:

AI doesn’t just flag potential risks – it suggests ways to reduce them. It’s like a built-in safety net, helping teams avoid compliance issues before they happen.

AI is steadily transforming clinical R&D in life sciences. By enhancing efficiency, accuracy and productivity, AI helps clinical teams accelerate the development of new therapies and improve patient outcomes. As AI technologies continue to mature, their role in clinical R&D will only become more central, making it imperative for organisations to adopt and integrate AI into their workflows to stay current with industry advancements.