The AI decides who enters, not what is given — a trial registered as a drug
A trial using a predictive model on electronic health record data to select surgical patients at high risk of persistent opioid use, then testing a non-opioid analgesic in them. What is registered as the intervention is the drug, not the AI.
Trial overview (primary data)
- StatusRecruiting
- ConditionsOpioid Consumption, Postoperative
- InterventionsDRUG: Suzetrigine, DRUG: Standard Care (in control arm)
- SponsorUniversity of California, San Diego
- Target enrollment100 participants
- Period2026-07-02 〜 2027-06-01
Key points
- A validated predictive model called SurgNet-POU predicts risk of persistent postoperative opioid use from electronic health record data.
- Persistent use is defined as requiring opioids three months or more after surgery, and those at high risk become eligible candidates.
- Suzetrigine is an approved non-opioid oral analgesic acting as a selective Nav1.8 sodium channel blocker.
- Participants are randomized to standard care alone or standard care with a fourteen-day regimen of suzetrigine.
- The primary outcome is total opioid consumption over fourteen days, the phase is 4 and target enrolment is 100.
1What part the AI plays
Across the six registrations so far, the AI analysed, tested, judged, supported, was consulted, and operated. Here it does none of those. The AI sits at the point of deciding who enters the trial. What is registered as the intervention is the drug.
2Three roles held apart
The accuracy of the model is not an outcome of this trial. The primary outcome is total opioid consumption, and the AI does its work upstream, narrowing who is considered.
3Why gather only those at high risk
- 1Query every patientHealth record data from all prospective orthopedic surgery patients goes into SurgNet-POU
- 2Predict the riskThe model predicts persistent use, defined as needing opioids three months or more after surgery
- 3Narrow the candidatesThose identified as high risk become eligible for the study
- 4RandomizeTo standard care alone, or standard care with fourteen days of suzetrigine
The design selects, in advance, the group where an effect should show. Offering a non-opioid option to those most likely to keep using opioids makes any difference easier to see, and the AI performs that narrowing.
4The drug box, and the phase
Only 32 of 776 registrations (4.1%) carry a phase at all, and most of those sit in the drug or radiation boxes. The box settles which framework of assessment applies. The next article sets all eight side by side.
Why it matters
An AI can enter a clinical trial without being the intervention. Acting at the point where candidates are selected leaves the drug as the registered intervention, and what gets measured is the effect of that drug rather than the accuracy of the model. That only 32 of 776 registrations carry a phase, most of them in the drug or radiation boxes, shows the registered box settling which framework of assessment applies.
FAQ
Is the AI not the intervention?
What is suzetrigine?
Why enrol only high-risk patients?
Sources (primary)
Source: ClinicalTrials.gov (U.S. NIH/NLM, public domain). This site does not provide medical advice. Verify the latest and exact details with the official source. This site is not endorsed or certified by the NIH/NLM.
- ClinicalTrials.gov (study record, original)
- NCT ID: NCT07672015