NIH NIDA SBIR Phase I  ·  1R43DA062307-01

The AI-native DCT platform for
behavioral health trials

Trialli is the first decentralized clinical trial platform purpose-built for substance use disorder research. AI imports your protocol, captures participant responses by voice, and runs your SUD trial — so coordinators focus on science instead of transcription.

Request a Demo See Voice Mode ↓
0 SUD-specific
competitors
$18.8B DCT market
by 2030
25–70% SUD trial
dropout rate
NIH NIDA SBIR
funded
Trialli Admin Portal
Trialli Patient App
The Problem

SUD trials fail at retention —
because existing tools weren't built for this population

Every existing DCT platform was designed for oncology and pharmacology trials in non-stigmatized conditions. None of them understands SUD.

25–70%

Dropout rate

SUD trial dropout is the #1 cost driver. Housing instability, stigma, and variable shift work mean participants cannot keep clinic appointments — and existing platforms assume they can.

$50–100K

Per coordinator per year

Study coordinators spend most of their time on structured check-ins, form administration, and protocol reminders — tasks that AI can handle at a fraction of the cost.

+40 days

Retention from clinician referral

Systematic coordinator contact adds 40 days of retention. AI can scale that contact to every participant, every week — without adding headcount.

REDCap

The status quo

The NIDA Clinical Trials Network — 16 nodes, hundreds of SUD trials — runs on REDCap. No mobile app. No AI. No voice. Trialli is the purpose-built replacement.

Differentiation

Three capabilities no other
DCT platform has

01

Voice-first data collection Beta

Participants speak their answers instead of typing. The app captures speech on-device, a rule-based matcher handles common cases, and an LLM fallback parses ambiguous transcripts into the structured score the form expects — with confidence scoring. Validated for SUD populations with low digital literacy.

Voice Capture
02

Purpose-built for SUD & behavioral health

Built-in library of behavioral-health instruments — ISI, BAM, PHQ-9, GAD-7, AUDIT-C, PCL-5, FTND, PROMIS Sleep — and protocol-import logic that recognizes a further dozen by alias (ASI, DAST-10, PSQI, MoCA, etc.). Designed around the community treatment-center workflow that NIDA-CTN sites actually run.

SUD Vertical
03

One-click protocol import

Upload the IRB-approved protocol PDF. The platform extracts the 12 ICH-GCP sections, the visit schedule, eligibility criteria, and the form list — and one click creates the entire EDC structure. Weeks of study-build work compressed into minutes.

Setup in Minutes
Competitive Landscape

How Trialli compares

Platform SUD Focus Voice AI Assessment Patient-Facing AI Community Site Model Pricing
Trialli You are here ✓ Purpose-built ✓ Full interview ✓ AI Coordinator ✓ Native Per-participant SaaS
Medidata Rave ✗ ✗ ✗ ✗ $200K–$2M/study
Medable ✗ ✗ Investigator only ✗ $150K+/study
Castor EDC ✗ ✗ ✗ ✗ Custom
REDCap ✗ ✗ ✗ ✗ Free / self-hosted
ObvioHealth ✗ ✗ ✗ ✗ Custom

Source: Competitive analysis conducted March 2026. No major DCT platform has SUD-specific positioning, instruments, or voice AI capability.

Voice Capture Beta on TestFlight

Speak your answer.
The form fills itself.

SUD trial assessments are 5–9 instruments per visit. Typing on a phone with shaky hands, blurred vision, or low literacy is the wrong burden to place on someone in early recovery. Voice mode removes it.

📱

Participant taps "Voice mode"

Available 24/7 in the iOS / Android / Web app. The current question is read aloud via on-device text-to-speech.

→
🎙️

Participant speaks the answer

Platform speech recognition transcribes the response — iOS Speech Framework, Android Speech Recognizer, or Web Speech API. On-device Whisper available as offline fallback.

→
📊

Answer extracted & confirmed

A rule-based matcher resolves the common cases instantly. An LLM fallback parses ambiguous transcripts into the structured value the question expects — number, choice, yes/no, or scale.

→
🔒

Into the EDC with audit trail

DataRecord stored with source = AI_ADMINISTERED and the original transcript. 21 CFR Part 11 audit log.

Forms in the built-in library

PHQ-9 GAD-7 ISI BAM AUDIT AUDIT-C DAST-10 PCL-5 FTND PROMIS Sleep APA DSM-5 L1 Sleep Diary Treatment Satisfaction Global Assessment ASI-Lite (Substance Use) TLFB Daily COWS CIWA-Ar Columbia SSRS PACS (Penn Alcohol Craving) Full ASI-Lite (7 sections, planned) PSQI (planned) BDI-II (planned)
🛣️

Phase II: multi-turn AI-conducted interviews

Today's voice mode is one question at a time. Our Phase II SBIR aim builds the full conversational version: the AI conducts the entire ISI or BAM interview turn-by-turn, asks clarifying follow-ups when confidence is low, screens every response in parallel for C-SSRS suicidality signals, and routes flagged sessions to a human coordinator before the participant ever sees the next question.

Workflow

From IRB to data lock —
every step supported

01

Sponsor creates trial & invites team

Draft protocol, configure sites (treatment centers, MAT clinics, residential programs), assign site PIs and coordinators, invite IRB reviewers — all by email. New users get automatic account setup.

02

IRB approves in-app

Reviewers log in, read the protocol, and approve or reject with comments. No DocuSign. No email threads. Only approved trials can activate enrollment.

03

Coordinators enroll participants

Send invitations by email. Participants receive a link to the Trialli app, create an account, and see their invitation. No app store account required for web.

04

Digital informed consent

Participants review the full consent document, agree to terms, and sign electronically — drawn or typed signature. Consent status updates in real time for coordinators.

05

Participants complete assessments by voice or tap

Scheduled forms (ISI, BAM, PROMIS, AUDIT-C…) appear in the patient app with overdue tracking. Participants either tap their answers or switch to voice mode and speak. Phase II adds the AI coordinator that initiates check-ins, detects missed visits, and escalates safety signals before a coordinator has to look.

06

Coordinators monitor & manage exceptions

Review AI-generated assessments, manage flagged items, report adverse events, track compliance — spending time on clinical judgment, not administrative contact.

The Platform

Built for every stakeholder

Sponsor & Coordinator Portal

A full-featured web application for sponsor administrators and site coordinators. Manage trials, review AI-generated assessments, track compliance, report adverse events — from a single dashboard.

  • Role-based access: Sponsor Admin, Site PI, Coordinator, IRB Reviewer
  • Real-time participant enrollment, consent, and compliance status
  • AI assessment review queue with transcript + confidence scores
  • Audit log for all critical actions (21 CFR Part 11 aligned)
  • Adverse event reporting with SAE escalation workflow
Trialli Admin Portal

Participant Mobile App

A consumer-grade mobile experience — iOS, Android, and web — designed for the realities of SUD populations: low digital literacy, variable schedules, housing instability.

  • Digital consent with e-signature (drawn or typed)
  • Scheduled forms with overdue tracking and reminders
  • Voice check-in: speak naturally, AI handles the rest
  • Document and image upload from camera or phone storage
  • AI assistant for trial questions (privacy-first, self-hosted LLM)
Trialli Patient App

HIPAA-eligible AI Assistant

Participants get answers to their trial questions in plain language, grounded in the protocol context. Today the assistant runs on Google Vertex AI's Gemini 2.5 Flash under a HIPAA BAA; a Phase II RAG pipeline will add retrieval over the protocol, consent, and a curated NIDA SUD literature corpus with source citations.

  • Gemini 2.5 Flash via Vertex AI (HIPAA-BAA eligible)
  • Per-feature provider override — admin can switch to Claude or GPT-4o
  • Voice input via platform STT + on-device Whisper fallback
  • Source-cited RAG Phase II
  • Air-gapped self-hosted GPU tier Phase II
Trialli AI Assistant

Wearable & Device Data

Collect continuous or episodic health data from participants using consumer wearables, native phone health stores, or manual entry. Sleep and actigraphy data map directly to BBTI and insomnia trial outcomes.

  • Apple HealthKit / Google Health Connect via the health Flutter plugin
  • Heart rate, blood pressure, sleep, weight, SpO₂
  • Manual entry with pending/sync status tracking
  • History view with date-organized records
  • Empatica Embrace2 integration Phase II
Trialli Wearable Data
Full Feature Set

Everything a behavioral health trial needs,
built in from day one

Voice-First Data Capture Beta

Participants speak instead of type. The current question is read aloud via on-device TTS, the participant's response is transcribed locally, and the answer is matched to the question's expected type — number, choice, scale, yes/no. Ambiguous transcripts fall back to an LLM extractor that returns a structured value with a confidence label.

  • Works against any form in the library — PHQ-9, GAD-7, ISI, BAM, AUDIT-C, PCL-5 …
  • Platform STT primary, on-device Whisper as offline fallback
  • LLM fallback is HIPAA-eligible and admin-configurable per feature
  • DataRecord stored with source = AI_ADMINISTERED + full transcript

Trial Lifecycle Management

DRAFT → PENDING → APPROVED → ACTIVE with full audit trail. Multi-site with per-site personnel, IRB approver assignment, sponsor + coordinator role hierarchy.

Digital Informed Consent

Participants review, agree, and sign electronically — drawn or typed signatures — right from their phone. Consent status tracked in real time for coordinators.

Safety & Adverse Events

Structured AE reporting with severity classification and SAE escalation. Built-in C-SSRS safety screen during every AI check-in. ICH E6(R2) GCP-aligned.

Participant Mobile App

iOS, Android, and web. Designed for SUD populations: simple interface, voice-first check-ins, offline form support, and crisis resources always one tap away.

AI-Assisted Coordinator Workflows Shipped

Coordinators today get an AI-assisted dashboard: real-time visit-compliance, AI-extracted protocol structure from PDF, instrument-library matching, and a participant chat assistant. Phase II expands this into a multi-agent backend that handles the participant relationship autonomously.

  • Protocol PDF import → 12 ICH-GCP sections + visit schedule (Gemini 2.5 Flash)
  • Visit compliance dashboard — overdue forms, missed visits, completion rate
  • AI form generation (GPT-5 strict-JSON)
  • Assessment / Follow-Up / Compliance / Safety agents Phase II

Structured eCOA & Forms

Configure validated forms (rating scales, single/multi-choice, free text, numeric) with per-form scheduling, overdue tracking, and revision workflows.

Audit Log & Compliance

Tamper-evident audit log on all critical models. 21 CFR Part 11 electronic signature support. Full trail for AI decisions, coordinator overrides, and data edits.

Architecture

Built to comply. Built to scale.

🤖

HIPAA-eligible LLM Layer

Per-feature provider routing: Gemini 2.5 (Vertex AI BAA) for protocol import, chat assistant, and close-out; GPT-5 for form generation; GPT-4o-mini for voice answer extraction. Admin can swap providers per slot from the AI-config panel.

🎙️

Platform & On-Device STT

iOS Speech Framework, Android Speech Recognizer, Web Speech API as primary; on-device Whisper (ggml-base-q5_1) bundled for offline fallback. Phase II: AWS Transcribe Medical for the multi-turn voice interview path.

⚡

Django REST Framework

Battle-tested Python backend with role-based permissions, full audit logging, and structured output extraction using OpenAI strict JSON mode.

🌐

Angular 18 Admin Portal

Standalone-component admin portal with Material Design — fast, type-safe, accessible. AOT-compiled for production.

📱

Flutter (iOS / Android / Web)

Single codebase across platforms. Native performance. Designed for low-bandwidth and variable connectivity common in SUD participant populations.

🔒

Compliance-Ready Infrastructure

Railway today (staging + production), with per-request email-capture middleware for safe E2E testing. 21 CFR Part 11 audit log on every clinical model. AWS migration for PHI-scale + SOC 2 path planned for Phase II.

Our Team

Clinical researchers.
Software engineers.
Mission-driven.

Trialli is built by a multidisciplinary team at the intersection of addiction medicine, clinical trial operations, and AI engineering — with NIH NIDA funding and Johns Hopkins University collaboration.

NIH Award
1R43DA062307-01
FOA
RFA-DA-25-052 (DCT for SUD)
Collaborating Institution
Johns Hopkins University
Trial Partner
Helping Up Mission, Baltimore
XF

Xuefeng Feng, PhD

Principal Investigator & Lead Engineer
Carina Medical LLC

Placeholder — add bio and institution here.

DA

Denis G. Antoine, MD

Co-Investigator
Johns Hopkins University

Placeholder — add bio and institution here.

+

Team Member

Role / Institution

Edit marketing/website/index.html — Team section to add members.

Get Started

The SUD trial platform
you've been waiting for.

We're actively piloting with NIDA-funded trials and NIDA CTN-affiliated sites. Request a demo and we'll walk you through the workflow live: protocol import, voice-first data capture, and the Phase II AI coordinator roadmap.

hello@trialli.io
Baltimore, MD / Remote
NIH Award 1R43DA062307-01

Request a Demo

We'll respond within one business day.