Ninety-six percent of clinical trials now include at least one risk-based monitoring component, up from 88% just a year earlier, according to the ACRO’s 2024 RBQM report. That shift makes the question of how source data verification gets configured, and where AI fits into it, more than an operational nicety. Get the SDV plan wrong and the consequences are regulatory, not just operational: a May 2024 FDA warning letter stemming from a BIMO inspection cited inadequate source documentation as an objectionable condition, a reminder that accurate transcription is still the floor regulators check first.
The practical problem with traditional SDV was volume. Applying uniform verification across every data field in a large trial burns monitor time on low-risk entries while higher-risk data, eligibility criteria, primary endpoints, critical safety fields, gets the same treatment as a routine demographics form. A configurable approach flips that: the monitoring plan defines verification intensity by site, participant group, visit, or individual field, and the intensity adjusts as the trial runs. A site with repeated eligibility errors draws more scrutiny; one with a clean track record holds at its planned level. The key discipline is that a risk signal should trigger a formal assessment before any SDV plan change, not an automatic escalation. Signal detected, assess significance, decide response, then adjust if needed.
Remote SDV fits inside that same logic rather than beside it. It is an execution method for verification the monitoring plan already requires, not a separate strategy. Where sites have electronic health records with appropriately restricted read-only access, verification that previously needed an on-site visit can happen remotely. A study published in Therapeutic Innovation & Regulatory Science found that remote risk-based monitoring saved between 9 and 41 on-site monitoring visits per trial site, corresponding to cost savings of $13,500 or more per site. Most studies will use both methods depending on what is being reviewed and what the monitoring need requires at that moment.
AI enters at the point where monitors process source documents against EDC entries, a step that is time-consuming but structured enough for automation to help. The strongest use case is extraction, comparison, and prioritization for human review, not autonomous verification decisions. A system that flags probable matches and surfaces discrepancies for monitor judgment compresses the manual effort without removing the human accountability that regulators inspect for. The SDR/SDV distinction matters here too: a correctly transcribed value can still represent a procedure that fell outside the protocol window, and no extraction model catches that without clinical context. Whether AI-assisted SDV systems in production actually reduce discrepancy rates, rather than just reducing monitor hours, is the number worth demanding from any vendor evaluation.
Source link: https://www.clinion.com/insight/source-data-verification-clinical-trials/
Moe Alsumidaie is Chief Editor of The Clinical Trial Vanguard. Moe holds decades of experience in the clinical trials industry. Moe also serves as Head of Research at CliniBiz and Chief Data Scientist at Annex Clinical Corporation.
