The Site Perspective: Why Ambient AI Is The Missing Link In Clinical Trial Data Integrity
By Mikel Daniels, DPM, MBA, president and chief medical officer, WeTreatFeet Podiatry
If you’ve ever watched a doctor during an exam, hunched behind a keyboard, barely looking up, you know that sinking feeling. You want the doctor’s real attention, not just a box checked on a computer. In routine practice, that dynamic is frustrating. In a clinical trial, it is a liability.
As a site investigator, I live this reality every day. When I am recruiting for a study or conducting a follow-up visit, my attention is split. I am trying to build rapport with a nervous patient while simultaneously ensuring I capture every needed detail required by the protocol, not to mention other regulatory bodies. Questions are always on my mind, as I don’t want to miss anything. Did they take the study medication? Any new concomitant meds? A slight headache last Tuesday? If I miss those details because I’m staring at a screen, the data quality suffers.
That’s where AI ambient listening has changed everything for our clinical research at WeTreatFeet Podiatry. It allows our interactions to flow naturally, without awkward pauses for typing. It lets me keep my eyes (and mind) on the patient, not the keyboard, ensuring that the "source data" for a trial is actually coming from the patient, not my memory of the conversation 10 minutes later.
The Tech In The Trenches
Here’s how it works in our clinic. With a patient’s explicit consent, I use an AI-powered app to listen to our entire visit. It acts like a tireless medical scribe that doesn’t miss anything or need something repeated. Afterward, the system creates a draft of my clinical note based on a template I created. It separates the chit-chat from the crucial health details. I read over the draft, clean up anything odd, and sign off.
We are currently applying this workflow to our participation in a trial testing a skin substitute for diabetic foot ulcers. In a wound care trial, the details are everything. The size of the wound, the exudate, and the patient’s compliance with offloading are the small details that often get lost in check-box-heavy EDC systems.
With ambient AI, I am actually present. I can read a patient’s body language, spot hesitations when I ask about adverse events, and take in the things they’re not saying. My patients notice. They feel like I am understanding every word, which builds the trust necessary for retention. Because the AI logs everything in real time, the record is less likely to miss important details such as a new allergy or a fleeting symptom that might be a safety signal.
Navigating The Regulatory Minefield
I know what the clinical operations and quality assurance leaders reading this are thinking: “That sounds great, Dr. Daniels, but what about FDA compliance? What about 21 CFR Part 111? What about source data verification?”
These were my questions, too. Participating in regulated research means we can’t just download a cool app and hope for the best. Here is how we navigate the compliance landscape at the site level while using these tools:
1. The "Human in the Loop" is the Validator
The biggest misconception is that the AI is writing the medical record. It is not. The AI is a drafting tool. It produces a preliminary note, but that note does not become a legal medical record or source document until I review, edit, and sign it.
From a 21 CFR Part 11 perspective, the only electronic signatures that matter are mine. This is applied within our Part 11-compliant EHR. The AI is simply a sophisticated dictation microphone. The "source data" is the verified note in the EHR, authenticated by the PI, not the temporary audio file processed by the AI.
2. Addressing "Hallucinations" and Omissions
We know large language models can occasionally "hallucinate" or invent details. In a trial, that is dangerous. However, in my experience, the greater risk in clinical research is omission — the doctor forgetting to write down a reported side effect because they were rushed.
The AI catches 99% of the conversational details. My job shifts from generating content to verifying it. It is much easier for me to spot a hallucination in a draft than to remember a forgotten detail from scratch hours later. This workflow actually increases the accuracy of AE reporting because the AI catches the patient’s offhand comment about feeling dizzy that I might have mentally discarded as unrelated or potentially missed when pressing buttons on a computer.
3. Specific Consent is Nonnegotiable
HIPAA allows me to use tools for treatment operations, but clinical trials are different. We require specific, distinct consent from trial participants to have their study visits processed by ambient AI. We explain that the audio is processed securely and then deleted, leaving only the text. Transparency here doesn’t just satisfy the IRB; it respects the patient’s data rights.
The ROI For Sponsors And Sites
There is also a financial layer here. Medical coding is notorious for being a money pit when mistakes happen. AI-assisted coding changes the math. When the AI listens and drafts, it suggests billing codes and flags missing info. This is not a huge problem for our clinical trial work, as we are not submitting billing to insurance. However, for the pharmaceutical sponsor, they see a return on investment from our data integrity. We provide richer narratives, fewer queries about missing data, and a PI who isn’t burned out by administrative burden.
Bottom line? This AI is not about replacing care; it’s about giving it back its human touch and making the business side (and the compliance side) less of a headache. That’s something patients, doctors, and the whole life sciences ecosystem can feel good about.
Automation And The Future Of AI
Data entry for a clinical trial site is often burdensome. At times, it is repetitive and time-consuming, and patients sometimes feel that us asking the same questions indicates either a lack of listening or a lack of comprehension. AI and automation reduce this. Having a patient fill out the screening questions on a laptop and having AI incorporate that information into the draft note can both speed up the visit time and improve the accuracy. Most of the data entry can be automated in this manner. It can then be screened, placed into the correct documentation platform, and satisfy the requirements for the study and the data integrity.
Embracing The Shift From Data Entry To Data Verification
For clinical operations leaders and site investigators, the hesitation to adopt ambient AI often stems from a fear of losing control over the source document. However, the reality is that we are currently losing control through distraction and omission. The industry must stop viewing AI as a replacement for the investigator and start viewing it as the ultimate safeguard for data integrity. By shifting our focus from manual data entry to expert data verification, we not only ensure compliance with 21 CFR Part 11 but also restore the patient-physician relationship that drives retention. I urge sponsors and sites to start piloting these workflows now, not just to save time but to ensure that the data we collect is complete and sound.
Reference:
About The Author:
Mikel Daniels, DPM, MBA, is the president and chief medical officer of WeTreatFeet Podiatry and an investigator with The Clinical Trials Network (CTNx). With over two decades of experience in podiatric medicine and surgery, Dr. Daniels specializes in complex wound management, diabetic limb salvage, and the integration of advanced technologies into clinical practice.
As a site investigator, Dr. Daniels actively conducts FDA-regulated research, including trials for novel skin substitutes in diabetic foot ulcer treatment. His work focuses on bridging the gap between clinical care and regulatory compliance, specifically regarding data integrity and source documentation. He is a fellow of the American College of Foot and Ankle Surgeons and holds board certifications in foot surgery and wound care.
Beyond his clinical role, Dr. Daniels applies his business acumen to healthcare operations as a consultant and a venture adviser for MBX Capital. Dr. Daniels earned his bachelor's degree from Muhlenberg College and doctor of podiatric medicine from Temple University.