Clinical trial demonstration
AI prescreening for clinical research sites
Volunteers connect health records, match recruiting studies, and hand off pre-filled referrals to research coordinators.
Live process demonstration
Volunteer prescreening from inquiry to appointment
One volunteer asks about diabetes trials, connects health records through SMART on FHIR, matches local studies, and receives a confirmed visit time from a research recruiter.
- 1
Connect health records securely
Volunteers sign into their patient portal, approving read-only access to conditions, medications, vitals, and lab results through SMART on FHIR standards.
- 2
Generate cited eligibility checklists
Assistant matches local recruiting studies, evaluating criteria line by line and citing exact chart entries for every met or unmet condition.
- 3
Hand off pre-filled referrals
Recruiters receive complete referral files with chart citations and open questions, then confirm the screening visit with the volunteer.
Five steps of AI prescreening
Review how data moves from volunteer questions to coordinator schedules during an automated prescreening interaction.
Select a step to inspect
Volunteer sees
Engine executes
Rules enforced
Human oversight
Volunteer Emery Purdy, Harbor Marsh Research, and coordinator Renee Alston are invented demonstration profiles.
Verified chart citations for study checklists
Every eligibility criteria line matches against specific reference codes from patient records. Missing items or old lab values mark criteria as unknown or unmet rather than making unsafe assumptions.
Automated checks and human decisions
Prescreening logic enforces strict data validation while keeping site staff in full control of patient referrals and eligibility decisions.
Citation verification server checks
Server logic rejects checklists citing nonexistent chart codes or marking criteria without citations, sending queries back to the assistant for correction.
Strict unknown handling rules
Lab results older than three months or unlisted medications render criteria unknown or unmet, preventing invalid eligibility assumptions from chart omissions.
Read-only patient authorization
Data extraction operates through volunteer consent, restricting access strictly to approved categories without storing patient records long term.
Human referral approvals
Site recruiters decide every referral and screening call, while Azure Prompt Shields screens all conversational messages against adversarial inputs.
One agent across every system
Most research-site software keeps data isolated, and vendors build AI tools only inside their own products. CloudCentric builds one unified agent that works across the separate applications a site already runs every day.
Volunteer chart prescreening
Automated prescreening connects volunteer health records to protocol criteria, generating cited match summaries for recruitment teams.
Recruiter study board copilot
Upcoming demonstration features an intelligent assistant assisting recruiters with candidate pipelines, trial matching, and call preparations across study boards.
Coordinator inbox automation
Future demonstration shows coordinator email inboxes converting unstructured messages into organized tasks and properly filed trial documents automatically.
Frequently asked questions
- Is this connected to Epic?
Epic is not connected yet. Demonstration sign-ins use the SMART Health IT public sandbox. Production deployments connect to health system patient portals supporting SMART on FHIR standards.
- What does a prescreen cost?
Prescreening workflows execute 16 to 20 steps over 30 to 38 seconds. AI model processing costs approximately 1 to 2 cents per volunteer prescreen.
- Who decides if a volunteer is eligible?
Site recruiters and study coordinators make every referral and eligibility decision. Assistant outputs serve strictly as pre-filled summaries and prescreening tools for human review.
- Where is the volunteer's health data kept?
Health data remains inside active chat sessions. System architecture stores no patient charts and builds no permanent data indexes.
- What does a real deployment need first?
Real deployments require a executed business associate agreement with the covered entity and IRB approval of prescreening chat scripts before volunteer interactions begin.
- Whose data is shown in the demonstration?
Demonstration uses synthetic patient Emery Purdy from a public test server and invented site Harbor Marsh Research. No real patient data is used.
Schedule a trial software demonstration
Call (844) 422-7000 or book a conversation to review AI prescreening workflows for your research site network.