1. Working with Agentic AI: Clinical Data Analyst

How human skills and agentic skills are dynamically integrated and coordinated across clinical data responsibilities

1.1. Clinical Data Analyst as an Occupation

A clinical data analyst designs, validates, standardizes, and analyzes patient and clinical research data to ensure scientific integrity, monitor health outcomes, and support regulatory compliance. This occupation connects clinical medicine, research protocols, and data governance by transforming raw electronic health records (EHR), electronic data capture (EDC) inputs, laboratory results, and patient-reported outcomes into rigorously validated, analyzable clinical datasets. To ensure that this data reliably supports medical decisions, research conclusions, and regulatory submissions, the analyst must understand how clinical protocols, patient safety requirements, medical terminologies, data standards, and regulatory frameworks collectively determine the meaning and acceptability of clinical information.

Typical responsibilities of a clinical data analyst include:

• clinical data capture design and system configuration;
• clinical data validation, cleaning, and query management;
• clinical data standardization and integration;
• clinical outcomes analysis and safety signal monitoring;
• regulatory compliance and clinical data governance; and
• clinical data reporting and research support.

These responsibilities are fulfilled through distinct tasks and subtasks, each requiring one or more actions with specific skill requirements. Human skills and agentic skills both contribute to these responsibilities, but their levels and forms of involvement differ according to the responsibility, task, action, data, clinical context, technical environment, and expected outcome.

1.2. Human Skills and Agentic Skills in Clinical Data Analysis

Clinical data analysis increasingly integrates both human skills and agentic skills.

Human skills are reusable capabilities possessed and developed by clinical data analysts. They include clinical terminology understanding, medical coding logic, regulatory reasoning, protocol interpretation, patient safety judgment, critical thinking, contextual interpretation, discrepancy resolution, and professional accountability.

Agentic skills are reusable capabilities available to AI agents through their underlying models, contextual information, instructions, constraints, tools, and computational resources. In clinical data analysis, these skills include natural language processing for clinical notes, automated anomaly detection, data extraction, standard mapping, edit check generation, query drafting, descriptive statistical calculation, pattern recognition, and documentation support.

Both skill types contribute to the same occupational responsibilities, but their involvement varies by action. Understanding a complex, unstructured adverse event narrative relies primarily on human knowledge of pathophysiology and clinical context; an AI agent may then use agentic skills to extract structured medical codes and cross-reference them against standardized safety dictionaries. In data cleaning, the analyst first determines acceptable physiological ranges based on the study protocol, and an AI agent subsequently scans millions of data points to flag violations and draft clarification queries for clinical sites. For regulatory compliance, an AI agent may continuously monitor audit trails for unauthorized access patterns, while the analyst interprets these flags to determine whether a genuine data integrity breach has occurred. Across these actions, human and agentic skills are activated sequentially or iteratively based on task requirements, with the human professional maintaining oversight and final judgment.

The relationship between these skills is not a fixed division where certain responsibilities permanently belong to humans and others to AI. Instead, human and agentic involvement shifts across tasks and between different actions within the same task.

1.3. Dynamic Human–Agentic Skill Integration and Collaboration in Clinical Data Analysis

Clinical data analysis functions can be understood as an occupation where responsibilities are fulfilled through tasks and subtasks, tasks and subtasks are executed through actions, and actions require appropriate configurations of human and agentic skills.

An occupation provides the overall professional context. For clinical data analysis, this context encompasses designing data capture tools, ensuring the accuracy and completeness of patient data, mapping information to universal clinical standards, monitoring safety signals, and preparing data for regulatory or clinical review.

A responsibility represents an expected area of professional work or an outcome that the clinical data analyst accomplishes or maintains. Responsibilities such as data validation, standardization, safety monitoring, and governance each contain multiple tasks.

A task is a context-dependent unit of work used to fulfill part or all of a responsibility. When a task is too broad or complex for reliable execution, it is decomposed into smaller subtasks.

Tasks and subtasks are executed through one or more actions. Actions represent reusable and configurable operations such as designing a case report form (CRF) field, writing a logical edit check, mapping a local laboratory code to a standard terminology, drafting a data query, calculating an adverse event frequency, or generating a data flow diagram.

Each action has skill requirements. These requirements describe the capabilities needed for effective and reliable execution. Depending on the action and its context, human skills, agentic skills, or an integration of both can satisfy these requirements.

The skills available for clinical data analysis reside in a skill space containing both human skills and agentic skills. Relevant skills in this space can serve as candidate skills for executing the actions required by a particular task. A human skill in medical coding logic, for example, can serve as a candidate for resolving an ambiguous investigator diagnosis, while an agentic skill in natural language processing can serve as a candidate for extracting that diagnosis from unstructured clinical notes.

Candidate skills are evaluated relative to the requirements of the action and its occupational and task context. Appropriate skills are then selected and organized into a skill configuration for execution. Depending on these requirements, a configuration may rely mainly on human skills, mainly on agentic skills, or on an integrated arrangement of both.

1.3.1. Skills Competition and Skills Collaboration

Two important relationships can occur between human skills and agentic skills when constructing a skill configuration: skills competition and skills collaboration.

Skills competition occurs when both a human skill and an agentic skill can satisfy the same or similar requirements for an action, but simultaneous application of both is unnecessary. Under a particular occupational and task context, one skill type can provide a better fit for the action than the other.

For example, both a clinical data analyst and an AI agent can generate a summary of missing data fields across multiple clinical trial sites. When the database structure, missingness definitions, and reporting templates are clearly specified, the agentic skill provides fast and scalable execution. When the summary requires interpretation of why a specific site is failing to collect a particular biomarker—perhaps due to a localized supply chain issue or an ambiguous protocol instruction—the analyst’s human skills provide the necessary contextual understanding.

The purpose of skills competition is not to establish whether human or agentic skills are generally superior. Rather, it identifies which available skill provides the better fit for a particular action under particular conditions.

Skills collaboration occurs when one skill type alone cannot effectively satisfy the requirements of a task or its actions, or when combining complementary human and agentic skills produces stronger task performance.

Collaboration does not require human and agentic skills to operate simultaneously. Instead, they are activated at different stages of action execution in a sequential, iterative, or recursive manner according to evolving task requirements.

For example, an analyst first defines the clinical logic required to identify potential drug-drug interactions in a patient registry. An AI agent then scans millions of longitudinal patient records, applies the logic, and flags suspected interaction events. The analyst subsequently reviews a sample of these flagged events to determine whether the algorithmic logic correctly captures true clinical risk or merely coincidental prescribing patterns. Based on the analyst’s feedback, the agent refines its natural language processing parameters to reduce false positives. Finally, the analyst evaluates the resulting safety signal and determines whether it warrants escalation to the medical monitor. Throughout this sequence, human and agentic skills are activated iteratively as task requirements evolve, while the analyst directs the investigation and evaluates its outcomes.

This dynamic relationship allows clinical data analysts to leverage the complementary strengths of both skill types rather than relying on a permanent allocation of work.

1.3.2. Skill Configuration and Task Execution

A skill configuration specifies which human skills and agentic skills are used to execute the actions associated with a task or subtask and how those skills are coordinated.

The configuration depends on factors such as action requirements, data sensitivity, patient safety implications, regulatory constraints, protocol complexity, expected reliability, available tools, information standardization levels, decision consequences, and intermediate results.

For one action, an agentic skill may independently satisfy the requirements. For another, a human skill may provide the better fit. A complex task may require several configurations across multiple actions, with human and agentic skills repeatedly interacting as the task progresses.

The configuration is therefore dynamic rather than permanent. It may change when the action changes, new protocol amendments become available, an agentic result fails to satisfy expected clinical logic, or human interpretation identifies a new patient safety concern.

Throughout this process, the clinical data analyst remains the professional responsible for directing and controlling the work associated with the occupation and its responsibilities. Agentic AI provides powerful complementary capabilities that expand what the professional can extract, validate, map, calculate, monitor, visualize, and document.

1.3.3. Task Performance Evaluation and Skill Space Adaptation

The performance produced through a selected skill configuration is evaluated against predefined evaluation metrics. These metrics determine whether the human skills, agentic skills, or their integrated configuration are sufficiently capable of performing the relevant task and its required actions.

Evaluation criteria depend on the work being performed. In clinical data analysis, they include data accuracy, protocol adherence, coding correctness, patient privacy preservation, query resolution timeliness, standard compliance, reproducibility, audit readiness, and other task-specific expectations.

When a selected skill or skill configuration produces performance above the defined threshold, it passes the performance evaluation and is retained for future applications under appropriate conditions. Repeated successful use provides additional evidence about where that skill or configuration performs effectively.

When performance falls below the required threshold, the corresponding skill or configuration is not reused without modification. The relevant human or agentic skills may be refined or reconfigured, existing skills can be combined differently, or new human and agentic skills may be developed or constructed from scratch when the existing skill space lacks adequate capability.

For example, an AI agent first generates automated data queries to clinical sites regarding out-of-range vital signs, but repeatedly issues queries for physiologically normal variations that the specific protocol intentionally allows. Task performance evaluation identifies this over-querying problem as falling below the required threshold, as it burdens clinical sites unnecessarily. The agentic skill is then refined through clearer protocol-specific constraint instructions, improved context-aware logic, or access to approved medical rules engines. The revised skill is evaluated again, and once it satisfies the required performance criteria, it is retained for appropriate future data cleaning tasks. Human skills follow a corresponding development process: when an analyst lacks the knowledge required to map a newly introduced genomic biomarker to international data standards, the relevant capability is developed through learning, collaboration with subject matter experts, and practical experience before its performance is evaluated through subsequent work.

Refined and newly constructed skills that satisfy the required performance criteria are added to the skill space. In this way, task performance evaluation provides feedback not only for completing current work but also for continuously optimizing and expanding the available skill space.

The skill space is not constructed as an isolated collection of occupation-specific representations. Many human and agentic skills used in clinical data analysis are reusable across occupations. Information retrieval, data extraction, standard mapping, statistical reasoning, calculation, summarization, visualization, documentation, and tool use contribute to work in health informatics, biostatistics, epidemiology, pharmacovigilance, public health, and many other fields.

Reusable skills can therefore be represented once and configured differently according to the responsibilities, tasks, actions, and contexts of different occupations. This allows the skill space to serve a growing range of occupations without unnecessarily duplicating or exploding the number of skill representations. Occupational specialization is introduced through task context, action requirements, parameters, data sources, tools, constraints, and skill configurations rather than by creating a completely separate skill representation whenever the same underlying capability appears in another occupation.

The resulting process is adaptive: occupational responsibility → task and subtask → actions → skill requirements → candidate human and agentic skills → skill configuration → task execution → task performance evaluation → skill refinement, reconfiguration, or construction → optimized and expanded skill space

This cycle allows human and agentic capabilities to evolve alongside the requirements of clinical data work.

1.4. Essential Responsibilities of Clinical Data Analyst

1.4.1. Clinical Data Capture Design and System Configuration

Clinical data capture design and system configuration establishes the digital infrastructure required to collect patient and research data accurately. This responsibility includes translating clinical protocols into case report forms (CRFs), defining logical edit checks, configuring electronic data capture (EDC) systems, establishing user roles, and designing data entry interfaces that minimize site burden and prevent entry errors.

Agentic skills contribute extensively to protocol parsing, draft CRF generation, edit check logic suggestion, interface layout optimization, and repetitive system configuration actions. Human skills remain essential for interpreting the clinical intent of the protocol, ensuring that data collection aligns with patient safety objectives, resolving ambiguities in medical definitions, and deciding whether a proposed edit check appropriately balances data quality with clinical site workflow.

For example, a research team first finalizes a protocol amendment introducing a new patient-reported outcome measure. The clinical data analyst clarifies the clinical purpose of the new measure and defines the acceptable scoring ranges. An AI agent then parses the protocol text, generates draft CRF fields, and proposes logical edit checks to prevent impossible scores. The analyst reviews these proposals and identifies that one proposed edit check would incorrectly block a valid clinical scenario involving pediatric dosing. After the analyst revises the logic, the agent updates the configuration scripts and executes validation tests. Finally, the analyst verifies that the configured system accurately reflects the approved clinical protocol.

1.4.2. Clinical Data Validation, Cleaning, and Query Management

Clinical data validation, cleaning, and query management ensures the accuracy, completeness, and reliability of aggregated clinical data. This responsibility includes reviewing data discrepancies, managing automated and manual edit checks, drafting clarification queries to clinical sites or healthcare providers, resolving missing data issues, and preparing databases for formal lock and analysis.

Agentic skills are well suited to continuous database scanning, anomaly detection, cross-form consistency checking, automated query drafting, and tracking query resolution metrics. Human skills contribute to determining whether a data anomaly represents a true medical event or a transcription error, formulating polite and clinically precise queries, interpreting complex medical narratives, and deciding when to escalate unresolved discrepancies to medical monitors.

For example, an AI agent continuously monitors an incoming clinical trial database and flags a series of identical, highly unusual laboratory values across multiple patients at a single site. The agent drafts automated queries asking the site to verify the data. The clinical data analyst reviews the flags and recognizes a pattern suggesting a localized calibration error in the site’s laboratory equipment rather than true patient pathology. The analyst instructs the agent to pause automated queries, drafts a comprehensive communication to the site’s principal investigator regarding equipment calibration, and updates the data validation plan. Human and agentic skills interact iteratively as the process moves from broad algorithmic detection to nuanced clinical investigation.

1.4.3. Clinical Data Standardization and Integration

Clinical data standardization and integration transforms disparate, localized data sources into unified datasets compliant with universal clinical data models. This responsibility includes mapping local laboratory codes, electronic health record (EHR) terminologies, and wearable device outputs to standardized dictionaries, ensuring interoperability, and merging external data feeds into the primary clinical database.

Agentic skills assist with large-scale terminology mapping, natural language processing for unstructured clinical notes, code translation, schema comparison, and automated alignment to international data standards. Human skills remain important for resolving ambiguous medical synonyms, understanding the clinical equivalence of differing diagnostic criteria, governing data mapping rules, and ensuring that standardized outputs preserve the original clinical meaning.

For example, an organization first acquires a real-world evidence dataset from a network of hospitals using varied local coding systems. The analyst defines the mapping strategy and identifies high-risk therapeutic areas requiring strict manual review. An AI agent then processes millions of records, mapping local diagnostic codes and medication names to standardized international terminologies. The analyst reviews the agent’s confidence scores for complex, unmapped oncology regimens and manually resolves the clinical equivalencies. The agent then applies these human-verified rules to the remaining dataset and generates a mapping validation report. Finally, the analyst approves the standardized dataset for downstream epidemiological analysis.

1.4.4. Clinical Outcomes Analysis and Safety Signal Monitoring

Clinical outcomes analysis and safety signal monitoring examines validated clinical data to track patient progress, identify adverse events, and detect early indicators of efficacy or toxicity. This responsibility includes slicing data by patient cohorts, calculating descriptive safety statistics, monitoring protocol deviations, visualizing longitudinal patient trends, and preparing data packages for biostatisticians and medical monitors.

Agentic skills retrieve relevant patient histories, calculate incidence rates, generate safety visualizations, detect temporal clusters of adverse events, and summarize protocol deviation trends. Human skills evaluate the clinical significance of statistical signals, distinguish between expected disease progression and unexpected drug reactions, interpret patient narratives, and determine which safety signals require immediate regulatory or medical intervention.

For example, a medical monitor requests an urgent review of hepatic laboratory trends in a specific patient cohort. The analyst defines the clinical parameters for liver toxicity and specifies the required time windows. An AI agent then extracts the relevant longitudinal lab data, calculates shifting averages, and generates visual trend lines highlighting patients crossing predefined toxicity thresholds. The analyst reviews these visualizations alongside the patients’ unstructured adverse event narratives to determine whether the lab shifts represent a genuine safety signal or expected fluctuations related to underlying comorbidities. The agent subsequently compiles the confirmed cases into a standardized safety report format for regulatory review.

1.4.5. Regulatory Compliance and Clinical Data Governance

Regulatory compliance and clinical data governance ensures that clinical data handling adheres to strict privacy laws, ethical guidelines, and regulatory agency expectations. This responsibility includes maintaining audit trails, managing user access controls, ensuring the anonymization or pseudonymization of protected health information (PHI), preparing for regulatory inspections, and enforcing data lifecycle policies.

Agentic skills contribute to continuous audit log monitoring, automated PHI detection and redaction, access pattern anomaly detection, compliance documentation generation, and tracking regulatory guideline updates. Human skills define acceptable data governance policies, interpret ambiguous regulatory findings, understand the ethical implications of data use, prioritize compliance risks, and authorize consequential database locks or unblinding procedures.

For example, an automated compliance scan alerts the team that a database export intended for an external research partner may contain unredacted patient identifiers hidden within free-text investigator notes. The analyst confirms the risk and directs an AI agent to trace the origin of the unredacted text and apply advanced natural language processing models to identify and mask all remaining PHI elements. The agent executes the redaction and generates a compliance validation certificate. The analyst then reviews a randomized sample of the redacted notes to ensure clinical utility was preserved without compromising patient privacy, ultimately authorizing the secure data transfer.

1.4.6. Clinical Data Reporting and Research Support

Clinical data reporting and research support converts validated clinical data into structured evidence that supports clinical study reports, regulatory submissions, and internal research decisions. This responsibility involves generating data flow diagrams, creating summary tables and listings, responding to data requests from biostatisticians or investigators, documenting database lock procedures, and ensuring analytical outputs are fully traceable to source data.

Agentic skills summarize database metrics, draft data flow documentation, generate standard summary tables, format listings for regulatory submission, and organize evidence for audit defense. Human skills evaluate the completeness of the reporting package, ensure alignment with the original clinical protocol, resolve conflicting data requests from different stakeholders, and determine whether the final data package is sufficiently robust to support scientific conclusions.

For example, a clinical trial reaches its final database lock milestone. The analyst defines the required documentation package to prove data integrity to regulatory auditors. An AI agent then aggregates all query resolution metrics, audit trail summaries, and mapping validation reports, drafting the initial technical appendices for the clinical study report. The analyst reviews the draft and identifies that the rationale for a specific batch of excluded patient data lacks sufficient clinical justification in the text. The analyst provides the missing context and specifies how the exclusion aligns with protocol deviations. The agent regenerates the documentation, and the analyst provides final sign-off for the regulatory submission.

1.5. Changing Human and Agentic Involvement Across Clinical Data Responsibilities

Human skills and agentic skills are involved throughout clinical data analysis, but their relative involvement differs across responsibilities and changes during task execution.

These involvement levels are not permanent allocations. A single responsibility may contain actions where human and agentic skills compete as alternative capabilities and other actions where they collaborate sequentially, iteratively, or recursively.

The practical objective is to construct skill configurations that use the most suitable available capabilities for each part of the work and revise those configurations when task performance evaluation shows that another arrangement performs more effectively or reliably.

1.6. Working as a Clinical Data Analyst with Agentic AI

Agentic AI is becoming an increasingly powerful component of clinical data analysis. Clinical data analysts should therefore adopt and integrate AI agents as complementary participants in professional workflows rather than treating them only as occasional tools for generating summaries, mapping codes, or formatting tables.

This shift does not reduce the importance of clinical and medical expertise. Human professionals remain responsible for the occupation and its corresponding responsibilities. They understand clinical protocols, define meaningful safety questions, interpret medical context, determine appropriate data standards, evaluate patient risks, communicate with medical monitors, control consequential database locks, evaluate task performance, and decide how human and agentic skills are configured and improved.

At the same time, agentic skills greatly expand the volume and variety of clinical information that can be processed. AI agents retrieve and organize massive volumes of patient records, generate and test logical edit checks, map heterogeneous data sources to universal standards, calculate descriptive safety metrics, detect anomalies across millions of data points, draft regulatory documentation, and synthesize extensive clinical evidence at a scale difficult to achieve through direct human effort alone.

Strong clinical data practice is therefore not based on maximizing human work or agentic work independently. It is based on finding appropriate configurations of human skills and agentic skills for the responsibilities, tasks, subtasks, and actions being performed.

Where a human skill and an agentic skill can independently perform the same action, skills competition identifies which is better suited to the current context. Where work requires complementary capabilities, skills collaboration results in human and agentic skills being activated sequentially, iteratively, or recursively to achieve stronger task performance than either skill type achieves alone.

Task performance evaluation then determines whether selected skills and configurations satisfy predefined expectations. Successful skills are retained and reused, while inadequate skills are refined, reconfigured, or newly developed. The resulting capabilities return to an increasingly useful skill space that supports clinical data analysis and, where capabilities are reusable, other occupations as well.

For a clinical data analyst, working effectively with agentic AI means treating AI as a powerful complementary capability integrated into everyday professional routines while maintaining human control of occupational responsibilities. Through appropriate skill selection, competition, collaboration, evaluation, and continuous improvement, human professionals and agentic AI work together to achieve clinical data analysis that is more efficient, scalable, adaptable, timely, compliant, and fundamentally protective of patient safety.