New Frontiers in Technology: AI Use in Clinical Trial Patient Recruitment in Health Systems

Supporting clinical trials for hospitals and health-system pharmacies is crucial due to the significant financial risks involved in drug development. The average cost of bringing a drug to market nearly doubled between 2003 and 2013 to $2.6 billion, with 90% of drugs washing out during human trials.¹ This indicates the need for financial support to obtain safety and clinical efficacy data with limited bias.

Clinical trials represent a significant portion of the mean cost of developing a new drug for the US market, estimated at $879.3 million when considering drug development failure and capital costs.² Approximately 75% of hospital-based clinical trials are funded by pharmaceutical firms.³ Involving pharmaceutical firms in research activities can facilitate outreach into local regions, promote enrollment of diverse patients, and expedite research activities.

Moreover, the integration of artificial intelligence (AI) technologies in clinical trial data management has shown promising results. According to Novartis, these technologies can lead to a 10% to 15% reduction in patient enrollment times in pilot trials, which can enhance drug approval rates and reduce development costs.⁴

Research pharmacies play a crucial role in ensuring the safety of patients involved in clinical trials.5 These pharmacies ensure the safe and appropriate use of treatments, promoting patient safety, compliance with regulations, and the ethical conduct of research following protocols for investigational drugs.⁵

In recent years, AI has transformed the process for identifying and recruiting patients to clinical trials, particularly for rare and ultrarare diseases. The use of AI technologies can leverage advanced data analytics, machine learning algorithms, and natural language processing to streamline and improve the efficiency of the patient recruitment process. It involves meticulously matching patients’ clinical profiles with trial eligibility criteria derived from diverse data sources, including global registries, published natural history studies, and electronic health records.^6,7

Patient Identification

Global Registries

Health care entities worldwide maintain comprehensive registries documenting phenotypic presentations, genotypic confirmations, and natural progression of rare and ultrarare diseases. Registries are crucial as they provide rich data sets, including patient demographics, clinical characteristics, biomarkers, and longitudinal data on disease progression.⁸

Natural History Studies

Natural history studies offer baseline data on disease courses without therapeutic intervention, which is essential for understanding disease trajectories and identifying potential trial participants. For example, a registry might include data on Duchenne muscular dystrophy progression, tracking muscle strength, respiratory function, and mobility over time.⁹

Matching Eligibility Criteria

Interested patients must meet specific criteria pharmaceutical companies establish before enrolling in studies. This helps identify participants without confounding factors and ensures unbiased results. Additionally, conducting a thorough literature search is vital for understanding treatment effects and avoiding bias by harmonizing and recruiting patients who receive similar standards of care across different geographical regions.10

AI Algorithms for Data Harmonization

AI aggregates and harmonizes data from various registries and studies, overcoming challenges of disparate data formats and standards. This process involves standardizing data entries, resolving discrepancies, and integrating heterogeneous data sets into a cohesive database. For example, AI algorithms process and harmonize data from different countries’ cystic fibrosis registries, ensuring consistent data representation.⁸

Patient Matching

Clinical trials are time-consuming and take years to complete. Subject recruitment is the first and foremost crucial step in initiating a clinical study. However, it is the most time-consuming, and 1 in 5 studies needs to complete recruitment and fulfill the required power to analyze baseline data. These enrollment criteria sometimes require analyzing the patient’s past exposure to treatments and filtering subjects that have confounding effects on results.⁹

AI systems utilize advanced algorithms to analyze extensive registry data and identify patients whose clinical profiles align with specific trial criteria. This involves comparing patients’ phenotypic and genotypic data, biomarkers, and disease progression markers against clinical trial inclusion and exclusion criteria. For instance, in a trial for a new gene therapy for spinal muscular atrophy, AI swiftly identifies eligible patients by cross-referencing trial criteria with patient data on SMN1 gene deletions, motor function scores, and respiratory status.⁹

AI utilizes programs or input code not only to identify required criteria from medical records but also to match the severity of disease in subjects with that in other similar participants in registries or natural history databases; this allows visibility around patterns of changes to understand patterns of phenotypic presentation in similar genetic variants. Ultimately, this creates a particular cohort of similar participants.⁹

AI harmonizes clinical data from published articles to understand whether there is a similarity in the standard of care for rare diseases worldwide. Recruiting clinical trials ensures participants receive similar standard of care to avoid any bias in trial results and ensure the proposed investigational drug shows similar benefits with no confounding effects or influence of external factors.⁹

AI significantly reduces the burden of traditional recruitment methods, which are often time-consuming and resource intensive, by automating the matching process.^10-14 This expedites recruitment and lowers associated costs. In breast and lung cancer clinical trials, using AI led to a 24% to 50% increase in accurately identifying eligible patients, surpassing standard practices.^11,15 The Automated Clinical Trial Eligibility Scanner system increased enrollment by 11.1%, improved the number of patients screened by 14.7%, and reduced patient screening time by 34% compared to manual screening leading to quicker trial initiation.¹⁶

Practical Implementation, Impact, and Future Prospects

The application of IBM Watson for Clinical Trial Matching is a notable example that is utilized in oncology to identify eligible patients for cancer trials.¹⁷ Ni et al study suggested that AI can accelerate patient matching for clinical trials and streamline the process, potentially increasing trial enrollment and expediting the approval of lifesaving cancer drugs.¹¹

Another case is the European Rare Disease Registry, which utilizes AI to harmonize and analyze data from multiple rare disease registries across Europe, facilitating efficient patient identification for clinical trials and research studies.¹²

With advancements in machine learning and natural language processing, the incorporation of AI in patient recruitment is expected to grow. For example, predictive analytics can forecast disease progression and patient outcomes, improving the accuracy of trial matching. Additionally, AI-driven virtual trials and decentralized trial models are beginning to emerge as well, offering new patient participation and data collection possibilities.¹³

Conclusion

AI is crucial in identifying suitable patients in clinical trials, particularly in rare and ultrarare diseases. By utilizing global registries and natural history studies, AI systems adeptly align patient cohorts with trial eligibility criteria, expediting recruitment and optimizing trial outcomes. This technological advancement bolsters the efficiency of clinical trials and enhances patient access to potentially life-saving treatments, underscoring the profound impact of AI in modern health care.

REFERENCES

1. Freedman DH. The promise of artificial intelligence in drug discovery. Nature. 2019;576(Suppl):S51-S56. doi:10.1038/d41586-019-03846-0.
2. Lopez-Rico I, Vives R, Cruel M, et al. Financial impact of the hospital pharmacy's participation in clinical trials. Eur J Hosp Pharm. 2021;28(Suppl 2):e185-e190. doi:10.1136/ejhpharm-2020-002601
3. Ni Y, Bermudez M, Kennebeck S, Liddy-Hicks S, Dexheimer J. A Real-Time Automated Patient Screening System for Clinical Trials Eligibility in an Emergency Department: Design and Evaluation. JMIR Med Inform. 2019;7(3):e14185. doi:10.2196/14185
4. Deloitte. Intelligent Clinical Trials: Transforming the Clinical Development Process. Deloitte Insights. Accessed August 11, 2024. https://www.deloitte.com/insights
5. Ismail A, Al-Zoubi T, El Naqa I, Saeed H. The role of artificial intelligence in hastening time to recruitment in clinical trials. BJR Open. 2023;5(1):20220023. May 16, 2023. doi:10.1259/bjro.20220023
6. Mai B, Roman A, Suarez A. Forward thinking for the integration of AI into clinical trials. Clinical Researcher. June 20, 2023. Accessed July 17, 2024. https://acrpnet.org/2023/06/forward-thinking-for-the-integration-of-ai-into-clinical-trials/
7. Harrer S, Shah P, Antony B, Hu J. Artificial intelligence for clinical trial design. Trends Pharmacol Sci. 2019;40(8):577-591. doi:10.1016/j.tips.2019.05.005
8. Sarno DL, Silver EM, Goldstein R, Frontera WR, Silver JK. Rehabilitation clinical trials in global registries: reporting of participant inclusion by sex, age, race, and ethnicity. Disabil Rehabil. 2024;46(13):2946-2954. doi:10.1080/09638288.2023.2231844
9. Nickel M, Schulz A. Natural history studies in NCL and their expanding role in drug development: experiences from CLN2 disease and relevance for clinical trials. Front Neurol. 2022;13:785841. doi:10.3389/fneur.2022.785841
10. Corbaux P, Bayle A, Besle S, et al. Patients’ selection and trial matching in early-phase oncology clinical trials. Crit Rev Oncol Hematol. 2024;196:104307. doi:10.1016/j.critrevonc.2024.104307
11. Haddad T, Helgeson J, Pomerleau K, Preininger A, Roebuck M, Dankwa-Mullan I, Jackson G, Goetz M Accuracy of an Artificial Intelligence System for Cancer Clinical Trial Eligibility Screening: Retrospective Pilot Study. JMIR Med Inform. 2021;9(3):e27767.
12. Calaprice-Whitty D, Galil K, Salloum W, Zariv A, Jimenez B. Improving Clinical Trial Participant Prescreening with Artificial Intelligence (AI): A Comparison of the Results of AI-Assisted vs Standard Methods in 3 Oncology Trials. Ther Innov Regul Sci. 2020;54(1):69-74. doi:10.1007/s43441-019-00030-4
13. European Platform on Rare Disease Registration. European Commission. Accessed July 17, 2024. https://eu-rd-platform.jrc.ec.europa.eu/_en
14. Fogel DB. Factors associated with clinical trials that fail and opportunities for improving the likelihood of success: a review. Contemp Clin Trials Commun. 2018;11:156-164. doi:10.1016/j.conctc.2018.08.001
15. Coustasse A, Kimble C, Penematsa V, Pile TL. Optimizing pharmaceutical innovation through AI: a pathway to expedited drug discovery and development. Pharmacy Times. May 23, 2024. Accessed July 17, 2024. https://www.pharmacytimes.com/view/optimizing-pharmaceutical-innovation-through-ai-a-pathway-to-expedited-drug-discovery-and-development
16. Haddad T, Helgeson JM, Pomerleau KE, et al. Accuracy of an Artificial Intelligence System for Cancer Clinical Trial Eligibility Screening: Retrospective Pilot Study. JMIR Med Inform. 2021;9(3):e27767. doi:10.2196/27767
17. Chopra SS. Industry Funding of Clinical Trials: Benefit or Bias? JAMA. 2003;290(1):113–114. doi:10.1001/jama.290.1.113
18. Sertkaya A, Beleche T, Jessup A, et al. Costs of Drug Development and Research and Development Intensity in the US, 2000-2018. JAMA Netw Open. 2024;7(6):e2415445. doi:10.1001/jamanetworkopen.2024.15445