Artificial Intelligence in Clinical Trials

Utkarsha Mane, Dr. Mahesh Thakare, Rushikesh Telore, Vaishnavi Hendge, Bhagyashri Thorbole,

doi:10.5281/zenodo.18073790

Review Paper | Open Access
Volume 02 | Issue 12 | Article Id IJSRT/250312134

Artificial Intelligence in Clinical Trials
Utkarsha Mane* Dr. Mahesh Thakare Rushikesh Telore Vaishnavi Hendge Bhagyashri Thorbole
Kasturi Shikshan Sanstha College of Pharmacy, Shikrapur, Pune-412208

Abstract

Improving the effectiveness, precision, and flexibility of clinical research procedures, artificial intelligence (AI) is quickly changing the clinical trial environment. AI provides creative answers to many of the conventional problems encountered in clinical trials, ranging from intelligent patient recruiting and protocol design to real-time data analysis and safety monitoring. Improved patient matching, quicker decision-making, and better trial results have all been made possible by recent developments in machine learning, natural language processing, and predictive analytics. Personalized treatment plans, decentralized trial models, and quicker drug development are all possible with the use of AI technology into clinical trials as they advance further. The main uses, current developments, and anticipated future developments of AI in clinical trials are highlighted in this paper, highlighting the technology's potential to transform clinical research and enhance patient care.

Keywords

Artificial Intelligence, Clinical Trials, Machine Learning, Predictive Analytics, Patient Recruitment, Adaptive Trial Design, Real-World Data, Natural Language Processing, Digital Health, Drug Development

Introduction

A key element of biomedical research, clinical trials serve as the foundation for the development of medical knowledge. These are methodical studies carried out on human subjects to assess the safety, effectiveness, and best practices of novel or pre-existing medications, biological products, medical equipment, or treatment approaches. By bridging the gap between laboratory research and clinical application, these trials guarantee that new medicines satisfy strict ethical and scientific requirements prior to being authorised for general use1-2. The creation of trustworthy evidence on the advantages and disadvantages of medical therapies is the main goal of clinical trials. This data eventually improves patient care by assisting medical professionals in making well-informed decisions. This helps improve existing treatment approaches in addition to assisting in the approval of novel medicines. James Lind carried out one of the first recorded clinical experiments in 1747 to investigate the impact of citrus fruits on scurvy in sailors, demonstrating the centuries-old notion of clinical trials. Since then, there has been a major evolution in the regulatory structure and conduct of clinical trials, especially in reaction to previous ethical scandals like the Tuskegee Syphilis Study. These historical turning points have influenced contemporary rules that place a strong emphasis on moral behaviours3-4. Strict international standards, such the Good Clinical Practice (GCP) guidelines set by the International Council for Harmonization (ICH), now regulate clinical studies. These studies are approved and monitored by regulatory bodies such as the Central Drugs Standard Control Organization (CDSCO) in India, the European Medicines Agency (EMA), and the U.S. Food and Drug Administration (FDA)4-5. Usually, there are four consecutive stages to the clinical trial process, each having its own objectives, target groups, and research strategies. Every stage is essential to guaranteeing that medical treatments are safe and effective for the intended population, from preliminary safety evaluations in healthy volunteers (Phase I) to extensive efficacy studies in patient groups (Phase III) and post-marketing surveillance (Phase IV)6. Clinical trials have grown more complicated in recent years due to the integration of real-world data, genetics, and artificial intelligence. Furthermore, in order to expedite the discovery of life- saving therapies and vaccinations, the COVID-19 pandemic highlighted the significance of adaptive trial designs and international cooperation. The goal of this study is to present a thorough analysis of clinical trials, emphasizing its many forms, design approaches, ethical and legal contexts, difficulties, and new developments. Researchers, medical practitioners, and legislators may help create safer, more efficient healthcare solutions by being aware of the complexities of clinical trials7-8

Fig.1: Clinical trials

Objective of the Review:

Assessing the safety, effectiveness, and tolerability of novel or current medical therapies in people is the main goal of clinical trials. The purpose of these studies is to collect high-quality, empirically supported data to aid in clinical judgement and regulatory approval.

Safety Assessment- To evaluate a medications or treatment's general tolerability, toxicities, and possible side effects. Aids in identifying any negative responses and determining the safe dose range.
Evaluation of Efficacy- To assess if, under controlled circumstances, the intervention yields the desired therapeutic benefit. Contrasts the intervention with either a regular treatment or a placebo.
Optimization of Doses - To determine the ideal dosage that provides the most benefit at the lowest possible risk. Uses dose-ranging research to determine the safest and most effective dosage.
Drug interactions and pharmacokinetics- To research pharmacokinetics, the study of drug absorption, distribution, metabolism, and excretion. To comprehend the pharmacodynamics, or the biological consequences and mechanism of action, of the medicine.
Effectiveness in Comparison- To assess the novel intervention's efficacy, adverse effects, and patient outcomes in comparison to either placebos or already available therapies.
Long-Term Results and Security-To track a medication's or treatments long-term effects (particularly in Phase IV studies). Finds uncommon or delayed side effects that weren’t seen in previous stages.
Development of Clinical Guidelines-Clinical trial data are used to inform national and international treatment standards and practices. Supports the use of evidence in medicine.
Regulatory Acceptance- For data to be submitted to regulatory bodies (FDA, EMA, CDSCO, etc.) for product approval and market authorization, clinical trial findings are necessary9-10.

Types of Clinical Trials:

Clinical trials can be generally categorized according to their interventional nature, design, and purpose. Choosing the best approach to answer a particular research topic requires an understanding of the many kinds of clinical trials. Each kind has unique uses in medical research as well as advantages and disadvantages.

Clinical Trials for Intervention11

Often referred to as experimental trials, they entail proactively allocating individuals to certain treatments (such as medications, equipment, or processes) in order to examine their impacts. Features: Participants are placed in either the control or intervention groups. Conditions and therapy are under the researcher's control. Used to evaluate causal linkages. Example: A study evaluating the lowering of blood pressure by contrasting a novel anti- hypertensive medication with a placebo.

Clinical Trials using Observation12

In these experiments, individuals are observed in their normal environment without receiving any kind of intervention. Features: The researcher didn't become involved. Beneficial for researching correlations, risk factors, or results. Either retroactive or prospective. Types: Cohort studies track a group across time to examine its results. In case-control studies, people with a condition (cases) are compared to people without it (controls). Snapshot studies that assess results at one particular moment in time are known as cross- sectional studies.

RCTs, or randomized controlled trials13

RCTs, which are regarded as the gold standard in clinical research, minimize bias by randomly allocating people to either the treatment or control group. Qualities: Comparability among groups is guaranteed via randomization. Blinding is frequently used to lessen prejudice. The findings are supported by statistics. A double-blind RCT comparing a COVID-19 vaccination to a placebo is one example.

Trials with Open Labels14

Both the participants and the researchers are aware of the therapy being given in these studies. Features: There is no blinding. Frequently seen in early-stage research or in situations when blinding is not feasible. Beneficial for long-term safety evaluations.

Trials that are both blind and double-blind15

By hiding group allocations, these designs help to lessen prejudice. Single-blind: The group assignment is just unknown to the individual. Double-blind: No one knows, not even the participant or the researcher. Triple-blind: Group allocations are likewise unknown to the data analysers. Benefits: Lessens observer bias and the placebo effect.

Trials of Crossover16

After a washout period, participants get several treatments in a sequential fashion. Features: Every participant is in charge of themselves. Beneficial for stable, long-term problems. Need fewer people, but timing is key.

Feasibility and Pilot Studies17

Before a full-scale trial, these small-scale exploratory investigations are carried out to evaluate the viability, time, expense, and possible results. Pilot Trial: Evaluates trial design and logistics. A feasibility trial establishes whether or not a complete trial is feasible.

Clinical Trials That Are Adaptive18

These provide adjustments to the trial's protocols (dosage, sample size, etc.) in response to intermediate findings without jeopardizing its validity. Features: Adaptable and effective. Beneficial in quickly developing domains such as infectious illnesses and cancer.

Trials of Equivalence and Non-Inferiority19

The purpose of equivalency trials is to ascertain whether a novel treatment is therapeutically equal to a conventional one. Non-Inferiority Trials: Prove that a novel therapy is not, within a certain range, inferior to the standard. Important: Frequently used in the creation of biosimilar medications.

Phase IV Trials for Post-Marketing Surveillance20

These are carried out to track a drug's long-term safety and efficacy after it has been authorised for commercialization. Qualities: Identify uncommon or postponed side effects. Evaluate practical efficacy in a range of demographics. Regulatory bodies frequently demand it.

Phases of Clinical Trials21-25:

Clinical trials are generally conducted in four sequential phases (Phase 0–IV), each with specific -objectives and increasing complexity. These phases ensure a comprehensive evaluation of a drug or intervention before and after it reaches the market.

Micro dosing Studies (Exploratory Trials) Phase 0

the goal is to do a preliminary analysis of human pharmacokinetics and pharmacodynamics. Participants: Very few (10–15 individuals in good health).

Design: Less than 1% of the dosage required to have a pharmacologic effect is given at sub- therapeutic doses.

Result: Assists in establishing if a medication acts in people in a manner consistent with preclinical research.

Relevance: Saves time and money before moving on to more extensive testing.

Phase I:

Dose determination and safety

Goal: Assess a novel drug's pharmacokinetics, safety, and tolerability. 20–100 healthy volunteers or occasionally patients (particularly in cancer) make up the participants.

Design: either single-blind or open-label.

The maximum tolerated dosage (MTD) is ascertained through dose-escalation experiments. Result: Identification of metabolism, excretion profiles, safe dose range, and adverse effects.

Phase II:

Profiling of Efficacy and Side Effects

Goal: Determine the drug's effectiveness and further assess its safety.

Participants: 100–300 individuals suffering from the intended illness. Design: Frequently controlled and randomized. Includes research on dose-response.

Result: Determines the short-term adverse effect profile and therapeutic dosage. Establishes if the medication has therapeutic value or biological action.

Confirmatory Trials, Phase III

Goals: Verify the medication's effectiveness, track adverse effects, and contrast with conventional therapies. 1,000–3,000+ patients from various centers participated.

Design: Double-blind, controlled, and randomized (RCT). Multinational, multicenter trials.

Result: The information gathered is submitted to and approved by regulatory bodies. Guarantees the authenticity of statistics.

Phase IV:

Monitoring After Marketing

Goal: After the medicine is approved, track its long-term safety, efficacy, and uncommon side effects.

Participants: People in general.

Design: either interventional or observational. May entail examinations of actual evidence.

Result: identifies drug-drug interactions, uncommon or delayed adverse events, and directs label changes.

Fig.2: Phases of clinical trials

Design and Methodology26-30:

Proper design and methodology are crucial for generating reliable, reproducible, and ethically sound results in clinical trials. A well-structured trial reduces bias, ensures participant safety, and strengthens scientific credibility. The definition of a trial protocol design is a comprehensive plan that describes the goals, design, technique, statistical considerations, and operational elements of the trial.

Contains:

Study justification

Criteria for inclusion and removal Comparators and interventions Measures of results Moral considerations

Using randomization

Goal: Balances confounding factors and removes selection bias. Types: Basic Randomization: Similar to tossing a coin. Block randomization guarantees that each group has an equal number of members. Using stratified randomization, groupings (such as age and gender) are balanced.

Masking, or blinding

Participants alone are not aware of their group in a single-blind study. Double-blind: No one knows, not even the volunteers or the researchers. Triple-blind: The data analyzer, researcher, and participant are all blindfolded. Relevance: Reduces bias in detection and performance.

Groups of Control

Comparison with an inert agent is known as placebo control. Active-Controlled: Evaluation vs conventional treatment. Particularly in non-pharmacological studies, no treatment or usual care control.

Calculating the Sample Size

Significance: To identify statistically significant changes, sufficient power is required. Factors include power (1−β), variability, significance level (α), and effect size.

Tools: Statistical programs such as SAS, nQuery, or G*Power.

Measures of Outcome

The main variable addressing the trial's goal is called the primary outcome (e.g., survival rate, symptom reduction). Secondary Outcomes: Additional effects (e.g., quality of life, side effects). Endpoints may be composite, surrogate (biomarkers), or clinical (mortality).

Plan for Statistical Analysis (SAP)

a predetermined strategy for examining trial results. Contains: Techniques for dealing with missing data Analysis of subgroups Per-protocol (PP) versus intention-to-treat (ITT) analysis

Observation and Evaluation

Information Committees for Monitoring (DMCs): Examine current safety statistics. Analyses data at predetermined times using interim analysis. Inspections and Audits: Verify adherence to legal and GCP requirements.

Ethical Considerations31-32:

Ethical integrity is the backbone of clinical research. Clinical trials must respect the rights, dignity, and safety of participants. Ethical considerations are enforced by international declarations, guidelines, and oversight bodies.

Knowledgeable Consent

The trial's goal, potential dangers, and advantages must all be adequately disclosed to participants. Participation is entirely voluntary, and withdrawal is possible at any moment. Prior to involvement, consent must be acquired, recorded in writing, and described in a clear and concise manner.

Institutional Review Board (IRB) or Ethics Committee

An independent ethical body, or IRB, must examine and approve each clinical experiment. Their responsibility is to: Examine the consent forms and research procedure. Protect the rights and safety of participants. Keep an eye on the trial's conduct

Evaluation of Risk-Benefit

Participants should be at little risk during trials. The possible advantage (to the individual or to society) must be greater than the danger. It is crucial to continuously check any negative consequences.

Confidentiality & Privacy

Confidentiality of participant data is required. Coded IDs are used in place of private data. Compliance with data protection laws such as HIPAA (USA) or GDPR (Europe).

Populations at Risk

Enrolment is handled with extra care:

Kids

Women who are pregnant Senior Citizens

Groups that are economically or educationally disadvantaged

Declarations and Ethical Guidelines

The WMA, or Declaration of Helsinki, is the gold standard for research ethics. Guidelines for ICH-GCP: centered on ethical and scientific excellence. The USA's Belmont Report: focusses on justice, beneficence, and respect. There is more ethical monitoring.

Regulatory Framework33-35:

Clinical trials are regulated to ensure compliance, safety, and scientific integrity. Each country or region has its own regulatory agencies and submission requirements

International Regulatory Authorities
Key Regulatory Documents

Investigational New Drug (IND) Application: Required to begin human trials (USA).
Clinical Trial Application (CTA): Submitted to national authority before trial initiation.
New Drug Application (NDA) or Marketing Authorization Application (MAA): Submitted after successful Phase III trials.

3. Good Clinical Practice (GCP)

Harmonized set of ethical and scientific standards.
Ensures:
Protection of human rights
Accurate trial data
Responsibilities of sponsors, investigators, and monitors

Challenges in Clinical Trials36-37:

Recruitment and Retention

Slow enrolment due to strict eligibility criteria.
Dropouts due to side effects, long study durations, or loss of motivation.

2. High Costs

Clinical trials are expensive:
Multi-center studies
Regulatory compliance
Data management
Small companies often struggle with funding.

3. Globalization and Regulatory Heterogeneity

Variability in regulations between countries causes delays.
Harmonization remains a challenge for multinational trials.

4. Ethical Dilemmas

Balancing risk vs. benefit in terminal illnesses.
Use of placebos when effective treatment exists.
Consent challenges in low-literacy populations.

5. Data Integrity Issues

Incomplete or inaccurate data collection.
Risk of fraud, bias, or selective reporting.
Need for robust monitoring systems.

6. Long Duration

Trials can take 5–10 years from conception to approval.
Regulatory reviews and data analysis further extend timelines.

Recent Trends in AI for Clinical Trials38-40

AI-Driven Patient Recruitment and Matching

AI algorithms scan EHRs and health databases to identify eligible patients.
Reduces recruitment time and increases patient diversity.
Example: IBM Watson Health is used to match cancer patients with clinical trials based on genomic profiles.

2. Natural Language Processing (NLP)

Extracts meaningful insights from unstructured data like clinical notes, case reports, and publications.
Automates data entry and improves protocol feasibility assessment.

3. Predictive Analytics

AI predicts trial outcomes, patient dropout risks, and adverse events using historical data.
Helps in risk mitigation and better trial planning.

4. Adaptive Trial Design

AI supports real-time modifications to trial parameters (e.g., dosage, sample size) based on ongoing results.
Makes trials more flexible and responsive.

5. Wearable and Remote Monitoring Integration

AI processes data from wearable devices to monitor patient vitals and symptoms continuously.
Enables Decentralized Clinical Trials (DCTs) with real-time feedback and safety monitoring.

6. Fraud Detection and Data Validation

AI algorithms flag inconsistent or fabricated data, ensuring data integrity.
Supports regulatory compliance and ethical conduct.

Future Perspectives41-42:

Personalized and Precision Trials

Integration of AI with genomics and biomarkers to create highly individualized treatment trials.
Expected to improve response rates and reduce trial failures.

2. Automated Protocol Generation

AI will assist in designing trial protocols based on disease trends, historical data, and real- world evidence.
Will drastically reduce setup time.

3. Fully Virtual Trials

Combining AI with telemedicine and mobile health apps could allow clinical trials to be conducted entirely online.
Increases accessibility for patients in remote areas.

4. AI-Augmented Regulatory Submissions

AI tools will help in organizing and validating regulatory documents, streamlining the approval process.
Speeds up time to market for new drugs.

5. Real-Time Data Analytics and Decision Support

Continuous data analysis through AI can guide immediate clinical decisions during trials.
Reduces adverse effects and enhances patient safety.

6. AI-Powered Synthetic Control Arms

AI can generate synthetic control groups using historical data, reducing the need for placebo groups and improving ethical standards.

CONCLUSION:

Artificial Intelligence (AI) has emerged as a transformative force in the clinical trial ecosystem, offering unprecedented opportunities to enhance efficiency, accuracy, and decision-making across all stages of drug development. By enabling faster patient recruitment, predictive modeling, real- time monitoring, and automated data management, AI significantly reduces trial timelines and operational costs. Machine learning and deep learning algorithms support more precise patient stratification, helping researchers identify ideal candidates based on genomic, phenotypic, and clinical characteristics. This leads to improved trial outcomes, reduced dropout rates, and more personalized therapeutic strategies. AI-driven tools also play a crucial role in detecting anomalies, predicting adverse drug reactions, and improving protocol adherence through digital health technologies and remote patient monitoring. Natural language processing accelerates literature analysis, regulatory documentation, and safety reporting, further streamlining the clinical trial workflow. Despite these transformative advantages, challenges remain. Issues such as data privacy, algorithmic bias, regulatory acceptance, model transparency, and interoperability of digital systems require strong governance frameworks and ethical oversight. Ensuring high-quality datasets and fostering collaboration between AI developers, clinicians, and regulatory authorities are essential to building reliable and clinically valid AI systems. AI will continue to evolve as a cornerstone of next-generation clinical research. Integration of wearable devices, digital biomarkers, federated learning, and adaptive trial designs will further advance precision medicine and patient-centric trials. With robust validation, transparent algorithms, and ethical safeguards, AI has the potential to revolutionize clinical development, accelerating the availability of safer, more effective therapies for patients worldwide.

REFERENCE

Topol, E. J. (2019). High-performance medicine: the convergence of human and artificial intelligence. Nature Medicine, 25(1), 44–56.
Esteva, A., et al. (2019). A guide to deep learning in healthcare. Nature Medicine, 25(1), 24– 29.
Weng, C., et al. (2020). Artificial intelligence for clinical trial design. Trends in Pharmacological Sciences, 41(12), 1050–1061.
Shivade, C., et al. (2014). A review of approaches to identifying patient phenotype cohorts using electronic health records. Journal of the American Medical Informatics Association, 21(2), 221–230.
Sarker, I. H. (2021). Machine learning: Algorithms, real-world applications and research directions. SN Computer Science, 2, 160.
Liu, Y., et al. (2020). Artificial intelligence in clinical trials: a review. Trends in Pharmacological Sciences, 41(12), 882–891.
Mak, K. K., et al. (2019). Artificial intelligence in drug development: present status and future prospects. Drug Discovery Today, 24(3), 773–780.
Beam, A. L., & Kohane, I. S. (2018). Big data and machine learning in health care. JAMA, 319(13), 1317–1318.
Rajkomar, A., Dean, J., & Kohane, I. (2019). Machine learning in medicine. New England Journal of Medicine, 380(14), 1347–1358.
Dinerstein, S. (2021). Artificial Intelligence (AI) and the Future of Clinical Trials. Clinical Trials Arena.
Xie, J., et al. (2020). Deep learning for natural language processing in electronic health records: a systematic review. JAMIA, 27(3), 457–470.
Soroushmehr, S. M. R., & Najarian, K. (2021). Artificial intelligence in clinical decision support systems. Computers in Biology and Medicine, 127, 104104.
Chan, A. W., et al. (2013). Increasing value and reducing waste: addressing inaccessible research. The Lancet, 383(9913), 257–266.
Wang, Y., et al. (2019). The role of AI in drug discovery and development. Drug Discovery Today, 24(4), 1066–1075.
Kommrusch, R., et al. (2021). Machine learning for clinical trial prediction: a systematic review. Journal of Biomedical Informatics, 113, 103639.
Ghassemi, M., et al. (2018). Opportunities in machine learning for healthcare. arXiv preprint, arXiv:1806.00388.
Hernandez-Boussard, T., et al. (2020). Predicting patient outcomes using AI. Annual Review of Public Health, 41, 505–516.
Wang, F., et al. (2018). Clinical risk prediction with machine learning: a guide for clinicians. BMJ, 361, k1494.
Zarin, D. A., et al. (2017). Update on trial registration 11 years after the ICMJE policy was established. New England Journal of Medicine, 376(4), 383–391.
Wenzel, M., et al. (2021). Improving patient selection in clinical trials using AI. Artificial Intelligence in Medicine, 118, 102112.
Yim, J., et al. (2020). Predicting conversion to wet age-related macular degeneration using deep learning. Nature Medicine, 26(6), 892–899.
Bhattacharya, S., et al. (2022). AI and real-world evidence in clinical trial design. Frontiers in Pharmacology, 13, 832348.
Barda, N., et al. (2021). Developing a COVID-19 risk prediction model using machine learning. Nature Medicine, 27(3), 568–576.
Yu, K. H., Beam, A. L., & Kohane, I. S. (2018). Artificial intelligence in healthcare. Nature Biomedical Engineering, 2(10), 719–731.
Goecks, J., et al. (2020). How machine learning will transform biomedicine. Cell, 181(1), 92– 101.
Subramanian, J., & Simon, R. (2010). Overfitting in prediction models–Is it a problem only in high dimensions? Contemporary Clinical Trials, 31(6), 588–595.
Luo, Y., et al. (2016). Natural language processing for EHR-based phenotyping: a systematic review. Journal of the American Medical Informatics Association, 23(1), 212–220.
Zame, W. R., et al. (2020). Machine learning in healthcare: a critical review. Nature Medicine, 26(4), 389–400.
Shah, N. H., et al. (2019). Artificial intelligence and clinical trials: the promise and the challenges. American Journal of Managed Care, 25(3), 123–126.
London, A. J. (2019). Artificial intelligence and black-box medical decisions: accuracy versus explainability. Hastings Center Report, 49(1), 15–21.
Topol, E. J. (2020). The convergence of human and artificial intelligence: implications for clinical trials. Nature Reviews Drug Discovery, 19(5), 279–280.
Jiang, F., et al. (2017). Artificial intelligence in healthcare: past, present and future. Stroke and Vascular Neurology, 2(4), 230–243.
Miotto, R., et al. (2018). Deep learning for healthcare: review, opportunities and challenges. Briefings in Bioinformatics, 19(6), 1236–1246.
Andreu-Perez, J., et al. (2015). Big data for health. IEEE Journal of Biomedical and Health Informatics, 19(4), 1193–1208.
Roberts, M., et al. (2021). Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19. Nature Machine Intelligence, 3(3), 199–217.
Brennan, P. F., et al. (2021). Using AI for adaptive clinical trials: benefits, risks, and challenges. Clinical Trials, 18(4), 391–400.
Norori, N., et al. (2021). Addressing bias in AI in health care: a new frontier. Nature Medicine, 27(12), 2061–2063.
Perakslis, E. D. (2020). Digital health and clinical trials: AI-based recruitment. NPJ Digital Medicine, 3(1), 1–3.
Rieke, N., et al. (2020). The future of privacy-preserving AI in clinical research. Nature Machine Intelligence, 2(6), 341–343.
Banerjee, I., et al. (2020). Reading race: AI recognises patient racial identity in medical imaging. The Lancet Digital Health, 4(2), e79–e90.