Pharma Focus Europe
ThermoFisher Scientific - Go beyond the catalog

AI-Driven Patient Recruitment

Dr Santhosh Kumar, VP, Enterprise Clinical Solutions, Indegene

Artificial Intelligence (AI) can revolutionize clinical trial patient recruitment by swiftly identifying candidates and cutting recruitment time. Natural Language Processing (NLP) can effectively extract vital data from diverse structured and unstructured sources, whereas Machine Learning (ML) can automate labor-intensive tasks, and predictive modeling can evaluate patient enrolment probability and compliance with trial protocols. AI accelerates trials, and ensures diverse participant pools, enhancing trial success.

1. How do you define AI-driven patient recruitment within the context of clinical trials, and what role does it play in enhancing the recruitment process compared to traditional methods? What are the primary challenges in conventional patient recruitment methods, and how can AI-driven approaches effectively address these challenges to streamline the process?

With more clinical trials getting decentralized and the adoption of AI/ML witnessing an upward trend, the clinical development landscape is evolving at a rapid pace demanding insightful decision-making for increased predictability and efficiency. Leveraging AI-derived insights, especially with Real World Data (RWD) fosters a safer, streamlined research environment thereby optimizing the clinical trial process and drug discovery process. In the patient recruitment space, AI-driven strategies are revolutionizing the entire process mainly by boosting the implementation of data-driven clinical trials’ design and patient enrolment activities respectively.

Some of the significant challenges pharmaceutical companies face with traditional patient recruitment methods are:

• Huge patient dropouts during initial screening, especially for rare diseases’ clinical trials
• Inability to meet patient enrolment timelines due to limited success in identifying potential participants
• Delayed responses from clinical research sites and unresponsiveness of research sites about the availability of potential target pool of participants

Integration of extensive structured and unstructured healthcare data from various RWD sources coupled with advanced analytics helps automate tasks while helping to elevate data quality in numerous activities across stages of clinical trials. Outlined below are a few examples of how AI/ML is improving trial design and patient enrolment in clinical trials:

Improvements in Study Design: AI enhances trial design and optimization by identifying patterns in data, enabling predictions about patient behavior and drug efficacy. AI/ML-enabled platforms analyze past studies to determine optimal patient populations, diagnosis, and prognosis requirements. This optimizes study design and improves decisions regarding country and site selection, enrollment models, and patient recruitment, eventually yielding predictable results, minimizing protocol amendments, and enhancing the overall efficiency of clinical trials.

Site identification and patient recruitment: AI and ML address challenges in site identification and patient recruitment for clinical trials. As studies focus on specific target populations of patients, achieving recruitment goals becomes more challenging, leading to increased costs and timelines. AI and ML mitigate these risks by identifying sites with high recruitment potential, suggesting effective recruitment strategies, and proactively targeting sites with predicted patient populations. This allows sponsors to prioritize sites and reach out to fewer sites with high enrolment probability. This accelerates recruitment and reduces the risk of under-enrollment, ultimately enhancing the efficiency and success of clinical trials.

Synthesizing disparate data elements, ML uncovers meaningful insights for precise site identification, ensuring access to ample patient populations. This approach significantly increases global enrollment rates compared to traditional experience-based site identification methods. Manual efforts in analyzing site risks and generating action items for clinical monitoring can be alleviated as advanced analytics offer composite site rankings, enabling precise risk identification. This accelerates decision-making, allowing for timely actions and issue avoidance in clinical trials.

2. Can you provide specific examples where AI has been successfully employed to accelerate patient recruitment in clinical trials, highlighting the key outcomes achieved? And, could you elaborate on specific instances where ML algorithms have significantly improved the efficiency of patient selection and screening for clinical trials, and how were these algorithms tailored to address specific challenges?

Integrating advanced AI and ML algorithms with existing data and domain expertise enhances clinical research efficiency. Unlike the traditional method of selecting and activating research sites based on historical data, using real-time site data enables just-in-time site activation. Prioritizing sites based on specific criteria, including target patient demographics, reduces recruitment forecasting time and resources. This approach offers a more accurate study completion timeline by activating sites based on current data, optimizing the clinical research process, and streamlining the path to patient enrollment.

Below are a few examples where with AI/ML some of the leading pharmaceutical companies garnered substantial process efficiencies in clinical development by meeting patient recruitment targets in record time.

This case involves leveraging AI/ML with a geofenced strategy to boost participant recruitment for a phase three trial on cytokine storm during the COVID-19 pandemic. Faced with recruitment challenges, the approach involved creating hyperlocal campaigns within a radius of targeted clinical research sites. By analyzing real-time location data, demographics, and user content preferences, the team achieved remarkable outcomes. The geofenced strategy resulted in over 17,000 weekly unique visitors to the trial landing page. From these visits, 460 interested participants were identified, indicating a 7% conversion rate from website visits to secondary qualification - which is 50% higher than industry norms. The success demonstrates the effectiveness of AI and ML technologies in optimizing patient recruitment strategies and achieving superior conversion rates. This approach not only addressed the specific recruitment needs during the pandemic but also showcased the potential of innovative digital strategies to enhance clinical trial outcomes in a targeted and efficient manner.

AI/ML-driven methodologies have also been used extensively for ongoing clinical trials as well. In one such instance, the pharmaceutical company had already engaged a Contract Research Organization (CRO) to enroll patients for a drug trial. However, the traditional methods used by the CRO were not very effective in meeting the patient recruitment targets. By leveraging AI/ML and RWD the pharmaceutical company conducted a comprehensive site feasibility assessment, real-time analytics tracking, and the activation of accurate research sites. In addition to this, hyper-local, geo-fenced digital outreach campaigns prioritized patient qualification. Support by RN Concierge Services ensured swift participant engagement and handover to research sites. The result was the activation of 60+ research sites in approximately just three weeks and a remarkable 3x increase in participant enrollment rate. Thus with data-driven site prioritization, omnichannel marketing, and RN Concierge Services, the company synergized with the existing CRO model to enhance outreach, refine the secondary qualification of eligible clinical trial participants, and meet the patient enrolment targets on time.

In another instance, a leading Japanese pharmaceutical company achieved a 40% improvement in both the top of the funnel and final enrollment by leveraging a hybrid nurse concierge system for screening. The primary screening process, a digital website-driven experience, was followed by a handoff to a nurse concierge team for phone or chat-based secondary screening. Additionally, an auto-site visit scheduler minimized delays between participant expression of interest and contact from the research site, improving adherence and accelerating enrollment. The efficient workflow streamlined the screening process and ensured a prompt handoff to the research site upon participant qualification, minimizing the risk of attrition and optimizing the overall enrollment experience. This innovative approach highlights the transformative impact of combining digital technologies with human touchpoints in clinical trial recruitment.

In another example, an oncology biotech company used AI/ML and RWD for trial design optimization, feasibility assessments, and recruitment forecasting. The solution involved creating a clinical site prioritization dashboard, ranking sites based on factors like therapeutic experience, current trial status, and competing trials. Notably, considerations included the size of hospitals, recognizing that larger institutions often lead to longer contracting times, impacting trial startup delays. The meticulous site selection process addressed this concern, ensuring judicious choices. The result was a bespoke trial recruitment plan tailored for a successful phase two metastatic non-small cell lung cancer study. This strategic approach showcased the importance of informed site prioritization in mitigating delays and optimizing trial outcomes for pharmaceutical and biotech companies in the oncology field.

3. In what ways can NLP effectively extract pertinent information from structured and unstructured sources in the context of patient recruitment, and what are the benefits of utilizing NLP in this domain?

NLP has significantly improved the analysis of diverse data sources, including Electronic Health/Medical Records (EHR/EMR), insurance claims, and notably, social media data. This plays a pivotal role in targeted marketing and educating potential trial participants. With the explosive growth of RWD in this industry which largely encompasses information routinely collected outside clinical trial settings, spanning hospitals, labs, imaging centers, and patient-reported outcomes, NLP can be extensively leveraged to generate Real World Evidence (RWE) for many informed decisions in the clinical development space. RWE significantly impacts inclusion/exclusion criteria evaluation, ensuring the right data is collected to ease site and patient burdens. By leveraging RWE, protocol amendments in clinical trials may be reduced, streamlining timelines and alleviating burdens.

The COVID-19 era emphasized the importance of hybrid clinical trials, introducing flexibility with patient assessments outside research sites. Home health services and local healthcare facility visits enhance trial participation convenience without straining research infrastructure. RWE, informed by patient feedback, influences study design and protocol optimization, promoting diversity, equity, and inclusion. In summary, NLP-driven data analysis and the integration of RWE contribute significantly to clinical trial optimization. From refining inclusion/exclusion criteria to minimizing protocol amendments, shaping study endpoints, and fostering hybrid trial configurations, the multifaceted impact of RWD on study design is evident. This comprehensive approach not only improves the efficiency of clinical trials but also ensures patient-centricity and relevance in a rapidly evolving healthcare landscape.

4. What are the critical factors that need to be considered when implementing NLP techniques for data extraction from diverse sources such as EHR and social media, and how can potential challenges be mitigated?

Addressing data privacy challenges in accessing diverse data sources, especially patient information, involves considerations of regulatory requirements and regional laws. Amidst the scrutiny faced by social media companies, protecting patient identity and health information (PHI) is paramount. To mitigate concerns, a focus on patient consent should be emphasized, employing a double opt-in mechanism. Beyond initial website consent, patients should receive an email requiring explicit permission, ensuring a robust consent process in adherence to local laws and platform frameworks. The objective is to assist patients rather than treating them as retail consumers. By seeking explicit permission and consent, concerns are alleviated, healthcare information is safeguarded, and actions are taken after their consent, maintaining a patient-centric and ethical approach in the realm of data access and utilization. Apart from the issue of data privacy, the presence of extensive data often results in significant noise, underscoring the need to discern and prioritize pertinent information over extraneous details.

5. How can ML techniques be utilized to automate labor-intensive tasks in the patient recruitment process, and what advantages does this automation bring to the overall efficiency of the recruitment process?

ML techniques play a pivotal role in patient recruitment by:

• Analyzing patient data against eligibility criteria, efficiently identifying potential study participants, and excluding ineligible candidates
• Streamlining scheduling and follow-up on appointments for clinical trial participants
• Automating patient qualification and various workflow activities

Further, regardless of data sources, AI/ML-enabled RWD platforms can ingest and unify datasets into a common data model. This eliminates the need for extensive time spent on data preparation, with the platform focusing on analytics and insights. This paradigm shift - from spending 80% of time on data preparation, to 80% on analysis, brings great efficiencies in clinical trials. The solution is crafted to assist data scientists with a powerful tool for data harmonization, analysis, and actionable insights. This approach aims to accelerate the drug development process from inception to market, aligning with the overarching objective of bringing effective pharmaceuticals to market promptly and efficiently.

6. What are the potential challenges in implementing predictive modeling techniques for patient recruitment, and what strategies do you suggest to mitigate these challenges effectively during the implementation process?

While predictive modeling enhances trial planning, it’s delayed application during the trial, rather than beforehand, poses challenges. The significance and quality of data for predictive modeling are crucial, given the potential disparities between predictions and actual outcomes. Analysis often involves eliminating outliers, potentially skewing results towards an idealized rather than realistic scenario.

7. How important is interdisciplinary collaboration between AI experts, healthcare professionals, and regulatory authorities in the development and implementation of AI-driven patient recruitment strategies? Can you provide examples of successful collaborations that have yielded significant advancements in this field?

Like with all industries, the adoption of AI/ML for any aspect of clinical development requires very strong interdisciplinary collaboration between AI experts, healthcare professionals, and regulatory authorities. While there have been great innovations by the AI experts’ community and extensive support by Healthcare Professionals (HCPs) the life sciences industry is yet to witness direct collaboration with regulatory authorities. However, it is extremely encouraging to see the frequent guidance the industry receives from regulators pertaining to the application of AI/ML for drug development activities.

8. Looking ahead, how do you envision the continued evolution of AI-driven patient recruitment shaping the landscape of clinical trials and healthcare research? What key steps do you believe are necessary to maximize its potential impact in the field?

The future of AI-driven patient recruitment holds promise for advancing clinical trials in continuing to accelerate drug development and improve the quality of clinical studies. To fully realize this potential, it is crucial to establish a supportive regulatory environment, foster collaboration, prioritize data privacy, and continuously assess and refine these approaches to align with the evolving needs of the healthcare and research communities.
While change is inevitable, the industry is cautious in implementing AI/ML innovations, emphasizing incremental transformations in each aspect of the clinical trial continuum. This approach ensures impactful changes are implemented gradually, fostering industry trust, comfort, and confidence over time, while maintaining patient safety as the paramount concern throughout.

--Issue 03--

Author Bio

Dr Santhosh Kumar

Dr. Santhosh has over two decades of experience in the clinical research industry, specifically clinical data management. Most recently, he was VP at Accenture Services, leading a large pharma client involving clinical study set-up activities. He has previously worked at TCS and IQVIA respectively, where he was responsible for clinical development solutioning and delivery. At Indegene, he is responsible for growing/streamlining Indegene’s Enterprise Clinical BU Global Operations.

Latest Issue
Get instant
access to our latest e-book
Lab Indonesia 2024Sartorius Lab Balance CleaningWorld Vaccine Congress Europe 2024World Orphan Drug Congress 2024Advanced Therapies 2024Equinix accelerated medical research...World Vaccine Congress Washington 2024EUROPEAN PHARMA OUTSOURCING SUMMIT 2024Pharma 2024