Introduction
Artificial intelligence and digital health applications (known in Germany as DiGA — Digitale Gesundheitsanwendungen), including AI-assisted diagnostics, digital therapeutics, wearable-derived endpoints, and AI-powered chatbots, are reshaping how healthcare is delivered and how evidence is generated.
For Medical Science Liaisons (MSLs), these technologies are not abstract concepts discussed in boardrooms; they directly affect day-to-day scientific exchanges with healthcare professionals (HCPs), investigator education at trial sites, and the standards to which every field interaction is held.
This article translates the ethical and compliance framework for AI in pharma into actionable guidance for MSL field teams. It addresses five core questions and, for each, highlights what the answers mean for the MSL sitting across the table from an oncologist, a clinical trialist, or a digital-health innovator.
1. Ethical Principles That Shape Every MSL Conversation on AI
The four-principle framework of medical ethics “autonomy, beneficence, non-maleficence, and justice,” remains the reference point, but AI adds new dimensions to each:
– Autonomy and informed consent: When an MSL discusses an AI-enabled companion diagnostic or a DiGA with an HCP, the conversation should go beyond efficacy data. HCPs need clear, understandable explanations of what the tool does, what data it processes, where its limitations lie, and how human oversight is maintained. Consent – whether from a patient using a digital therapeutic or an investigator adopting a novel digital endpoint – is only genuine when these points are transparent. MSLs are often the first scientific peer an HCP turns to with questions about a new technology; ensuring that the information shared enables truly informed decisions is a core MSL responsibility.
– Non-maleficence beyond clinical errors: In the AI context, harm extends to privacy risks, false alarms, alert fatigue, and cybersecurity vulnerabilities. An MSL presenting data from an AI-supported tool should be prepared to discuss not only clinical performance but also these broader safety dimensions. For example, if a wearable generates excessive false-positive alerts in a monitoring study, this creates clinical burden and a type of harm that deserves the same candid discussion as an adverse drug reaction.
– “Justice” e.g., regarding bias and equitable access: If the training data behind an AI tool underrepresent certain populations, the tool may perform differently across patient subgroups. MSLs supporting investigator-initiated trials or advising HCPs on digital tools should proactively share what is known about validation cohorts and potential bias, so that clinicians can make informed decisions about applicability in their own patient populations.
Practical Takeaway for Field Teams:
Before discussing any AI or digital health tool with an HCP, prepare a concise “transparency brief” covering: intended use, data inputs and outputs, known limitations, validation populations, and the role of human oversight. Treat it as you would a benefit-risk summary for a new molecule.
2. Compliance Guardrails: What Industry Codes Mean for MSL Interactions
Pharmaceutical industry codes, including those of IFPMA and EFPIA mandate the strict separation of scientific/medical information from promotional content, factual balance, substantiation of claims, and transparency in HCP interactions. When AI enters the picture, these guardrails require specific translation for the field:
– No “black-box” marketing promises: MSLs should avoid positioning AI capabilities in vague, unsubstantiated terms. If an AI tool claims to improve diagnostic accuracy, the MSL should be ready to share the published evidence, including study design, endpoints, comparators, and confidence intervals, as exactly as they would for a therapeutic claim.
– Clear labeling of intent: Every MSL interaction on AI should make explicit whether the content is medical/scientific information or relates to a product with promotional dimensions. If an AI chatbot delivers content to HCPs on behalf of the company, that content should have gone through legal, medical and regulatory (LMR) review, with documented version control and a defined update cadence.
– Transparency of value transfers: Digital channels should not become vehicles for hidden influence. Sponsorship of AI tools, provision of digital health platforms, or data-sharing arrangements all fall under transfer-of-value reporting. MSLs should be aware of these obligations and be ready to disclose relevant relationships proactively.
Practical Takeaway for Field Teams:
When an HCP asks about an AI feature or digital health application, apply the same compliance checklist you use for any scientific exchange: E.g. Is your response on-label? Is it substantiated by approved evidence? Has the material been through LMR review? If any answer is “no” or “unclear,” escalate before responding.
3. Data Privacy and Patient Sovereignty in the Field
Confidentiality is a cornerstone of medical practice, and regulations like the GDPR codify the principles of purpose limitation, data minimization, and integrity. For MSLs, these are not abstract legal concepts, but they surface in concrete field scenarios:
– Discussing digital therapeutics: When an HCP asks how a DiGA handles patient sleep data, the MSL should be able to explain what is collected (sensor data, usage patterns), why, who may access the information, and how patients can opt out without compromising care. This “privacy-by-design” narrative demonstrates that the company treats patient sovereignty as a design principle, not an afterthought.
– Supporting clinical trials with digital endpoints: Wearable-derived endpoints such as continuous heart rate monitoring, actigraphy, or stride velocity generate vast volumes of patient data. MSLs educating investigators on digital trial protocols should clearly communicate data governance: how data flows from the device to the database, what pseudonymization or anonymization steps are applied, and how retention and deletion policies protect participants. Investigators are gatekeepers of patient trust; equipping them with clear privacy information strengthens site engagement.
– Real-world evidence discussions: When MSLs facilitate conversations about real-world data generated by AI tools, they should be prepared to address whether patients provided granular, understandable consent and whether the data can be used for purposes beyond the original intent.
Practical Takeaway for Field Teams:
For every digital health or AI-enabled product in your portfolio, maintain a one-page data-privacy fact sheet covering: data types collected, purpose, third-party access, anonymization methods, retention periods, and patient opt-out options. Use it as a reference during HCP and investigator meetings.
4. Clinical Accountability, Human-in-the-Loop, and the MSL’s Role in Investigator Education
A foundational principle: clinical accountability remains with human practitioners. AI can support clinical judgment but should not replace it without clearly defined roles and responsibilities. This principle has direct implications for MSL activities:
– Educating investigators on AI-supported trial tools: As digital endpoints and AI-assisted data analysis become more common in clinical trials, MSLs increasingly serve as the bridge between the sponsor’s digital-health team and the investigator. This means helping investigators understand: (a) the intended use and validated performance of the AI component, (b) how uncertainty is communicated (e.g., confidence scores, flagged outliers), (c) what “human-in-the-loop” means in practice and that the clinician retains the authority and responsibility to override AI-generated outputs, and (d) how performance drift is monitored and managed over time. MSLs supporting site education should ensure investigators know exactly where the AI’s role ends and the clinician’s judgment begins.
– Discussing AI-driven recommendations with HCPs: When an HCP in clinical practice encounters an AI tool that suggests a diagnosis or treatment pathway, the MSL’s role is to provide the evidence behind the tool’s algorithm, including how it was trained, validated, and what its known error rates are. The message should always be clear: the AI is a decision-support tool, not a decision-maker.
– Regulatory context for software as a medical device (SaMD): MSLs should be conversant in the regulatory classification of AI tools they discuss. Risk management, usability engineering, cybersecurity requirements, and post-market surveillance obligations for learning systems are all fair topics for investigator education. For adaptive or continuously learning AI systems, additional controls apply: governed updates, re-verification and re-validation after model changes, and traceability of model and dataset versions.
Practical Takeaway for Field Teams:
When preparing for investigator meetings involving digital endpoints or AI-enabled tools, build a “human-in-the-loop” explainer into your site education materials. Cover: what decisions the AI informs, what decisions remain with the clinician, how to report unexpected AI behavior, and where to escalate concerns. This is analogous to the adverse-event reporting education you already provide and extend it to AI incidents.
5. Managing AI Risks and “Digital Side Effects” – The MSL as an Early Warning System
Pharmacovigilance is well established in pharma; for digital health solutions, it is complemented by post-market surveillance, incident management, and cybersecurity reporting. MSLs are uniquely positioned to serve as an early warning system:
– Detecting signals in the field: Just as MSLs capture unsolicited safety reports during HCP interactions, they should be alert to reports of AI malfunctions for example, an algorithm systematically producing incorrect recommendations, a digital endpoint device generating unreliable data, or a cybersecurity incident affecting patient data. These observations should be routed through defined reporting pathways (Medical, Safety, Compliance, IT Security) with the same urgency as pharmacovigilance signals.
– Transparent communication: If an AI tool is found to have a systematic flaw, ethical practice demands prompt correction and clear communication to users with no delay driven by reputational concerns. MSLs may be asked to communicate corrective actions to HCPs or investigators; doing so openly and factually reinforces the trust that underpins every MSL/HCP relationship.
– Distinguishing event types: It is important to differentiate clinically relevant AI events (e.g., a recommendation leading to a medication error) from privacy or security incidents (e.g., a data breach). Both require rigorous response, but through appropriate and potentially different processes. MSLs should understand these distinctions and know which internal pathway to activate.
– Audit trails and documentation: Every MSL interaction related to an AI tool should be documented in the CRM with the same diligence as any other HCP engagement. This creates the audit trail necessary for compliance review and, if needed, regulatory investigation.
Practical Takeaway for Field Teams:
Add “digital adverse events” to your field reporting awareness. If an HCP reports an unexpected AI behavior, a device malfunction affecting a digital endpoint, or a data-privacy concern, treat it as a reportable event. Document it, route it to the appropriate internal function, and follow up. Your CRM note should capture: what happened, which product/tool was involved, what the clinical or data impact was, and what the HCP’s expectation for follow-up is.
Quick-Reference:
MSL Field Checklist for AI and Digital Health Interactions
Key Questions to ask yourself before the HCP interaction:
Ethics: Can I clearly explain the tool’s purpose, limitations, data handling, and human oversight to this HCP?
Compliance: Is my content on-label, substantiated, LMR-reviewed, and clearly labeled as scientific/medical (not promotional)?
Data Privacy: Can I describe what patient data the tool collects, why, who accesses it, and how patients can opt out?
Clinical Accountability: Am I making clear that the AI supports but does not replace the clinician’s judgment?
Risk Reporting: Do I know the internal pathway for reporting AI-related incidents, data breaches, or device malfunctions?
Documentation: Will my CRM entry provide a complete, audit-ready record of this interaction?
Closing Perspective
The ethical and compliance landscape for AI and digital health is evolving at pace. For MSLs, this is not a distant regulatory exercise but it is the fabric of daily HCP engagement, investigator education, and evidence interpretation. By grounding every AI-related interaction in the principles of transparency, patient sovereignty, human accountability, and rigorous documentation, MSLs ensure that digital innovation strengthens rather than erodes the trust that makes scientific exchange possible.
The MSL who can articulate not only what an AI tool does but also how it is governed, how patients are protected, and where the human remains in the loop will be the most valued scientific partner in the age of digital health.
About the author:
Dr. Hans Joachim Hutt, MD
Dr. med. Hans Joachim Hutt is a former Medical Affairs, Market Access, Governmental Affairs Leader in distinguished pharmaceutical companies as Boehringer Ingelheim, sanofi, Pharmacia, BMS, SPMSD, and LEO Pharma. He is a Medical Doctor from the University of Mainz, Germany, and holds a medical degree from the University of Essen, Germany. After years of medical training in internal medicine and clinical pharmacology, he joined the pharmaceutical industry in 1989.
During his career, he has managed MSL teams with other departments as Regulatory Affairs, Pharmacovigilance, Clinical Operations, and Medical Affairs in Germany, Austria & Switzerland. He has facilitated the successful launch of pharmaceutical products for big and mid-sized pharmaceutical companies. He contributed to evidence generation, guideline appraisal, lifecycle management, and held responsibility/accountability for value-added dossiers and pricing negotiations for new molecular entities.
Dr. Hutt has coached, interviewed, and reviewed the CVs of many aspiring MSLs. His insights and guidance have resulted in MSLs successfully breaking into their first roles and fostering careers within the industry.
He has been a keynote speaker and moderated many conferences on various industry–related topics, including creating teams, management, MSL training, and proper utilization of a competency framework used to measure MSL performance.
Dr. Hutt has also served as a resource and consultant on many compliance-related projects for several organizations, including the German Code of Conduct Body (AKG e. V.), chairing a committee assessing audited organizations regarding HCP-/HCO-Compliance. For the German trade association BPI, he acts as a senior expert focusing on patient support programs, medical education, and AI.
He was recently announced as a member of the expert group „off-label“ at the German drug authority and acts as Information Officer according to the German Drug Law. As a consultant, he assesses business plans of start-ups developing digital health applications/devices in Germany and Singapore.
Have an opinion on this article? Send a message to the editor.