보도자료

Jarame prepares an FDA pathway: 100,000-person RWD and a separated digital therapeutic module

Jarame will not file its whole platform as a medical device. It plans to isolate one therapeutic module for a U.S. FDA pathway. About 100,000 people of real-world data is a starting point, not an FDA-validated clinical result.

Author: 자람이팀

[Press Release · 2026-08-17]

Refine real-world data through the AI Digital Care Log. Separate one treatment module. Then pursue a U.S. device pathway.

Dream AI Lab’s developmental-disability platform Jarame is preparing for U.S. FDA medical-device review. The plan is not to file the whole platform at once, but to isolate clinically intended functions as an independent digital-therapeutic module and enter in stages.

Conceptual Jarame FDA diagram: home and clinic records become evidence, then a single sealed therapeutic module
Longitudinal records become evidence; one module is the first device candidate. Conceptual illustration, not an FDA certificate.

Jarame says about 100,000 people of developmental-disability real-world data have accumulated through the AI Digital Care Log. It also says improvement signals in the 30–65% range were observed on some treatment or function measures, depending on cohort and measurement conditions.

These figures are company-stated. They are not FDA-recognized clinical effectiveness. For a submission, analyzable subjects must be re-defined by diagnosis confirmation, baseline, actual intervention exposure, follow-up, instruments, concomitant therapy, missingness, and comparators—not by total users.

Why event counts are not clinical N

Jarame says data-processing methods used at Noah AI Labs—on the order of about 3.6 million events in a month—were applied to the AI Digital Care Log. Collected events are de-duplicated and normalized into person-level longitudinal records. For FDA use, the company plans a regulatory lineage that can trace source events, origin, decision rules, algorithm version, and edit history—not only the cleaned count.

Even millions of events are not millions of independent clinical samples. Total events, unique persons, completed therapy sessions, completed assessments, and long-follow-up subjects must be counted separately. Volume is a start. Lineage, completeness, bias control, and causal analysis must be joined before data can become RWE FDA can use.

RWE can support device decisions—if relevant and reliable

FDA’s December 2025 final guidance on real-world evidence for medical devices states that RWD may inform regulatory decisions when it is sufficiently relevant and reliable. Jarame’s large RWD could later support trial design, external controls, subgroup analysis, long-term safety, and post-market performance—after those quality tests, not because the count is large. (U.S. Food and Drug Administration)

First device: a working-name single module

The intended file is expected to be separated as working-name Jarame Therapeutic-01. The first product would not cover all developmental disability. It would freeze one age band, diagnostic condition, treatment goal, and clinical endpoint. Candidates include social communication, functional requesting, challenging behavior, or daily-living function—whichever currently has the strongest completeness and clinical validity.

Center search, home-therapy matching, scheduling, guardian–therapist communication, and general education stay on the existing platform. Patient-specific content selection, difficulty, intervention order, and next-step proposals that directly affect clinical outcomes move into separate device software. The architecture of that split is explained in Jarame’s FDA separation strategy.

Q-Submission first, pathway later

Before a formal file, Jarame plans FDA Q-Submission to discuss intended use, patients, endpoints, how far existing RWE can go, and what additional trial is needed. If no suitable predicate is found, a De Novo path is more likely for a novel low-to-moderate-risk device, but the final path must be set after predicate search and FDA discussion. (U.S. Food and Drug Administration)

Preparation covers more than clinical effect: software lifecycle, risk management, verification and validation, usability, cybersecurity, change control, complaints, CAPA, and post-market surveillance. FDA’s QMSR, effective 2 February 2026, incorporates ISO 13485:2016 by reference, so a quality system must cover development and operations. (U.S. Food and Drug Administration)

A platform of modules, not one frozen app

The strategy is to keep the AI Digital Care Log as developmental-disability RWD/RWE infrastructure and add indication-specific therapeutic modules on top. If the first module is authorized, later modules could cover communication, behavior, daily living, and sleep, with long-term effect and safety checked from actual use after market.

Evidence boundary: User counts, event volume, and improvement ranges are company materials today. They must be converted to regulatory-grade evidence through protocol registration, independent statistics, and FDA review. Jarame does not replace diagnosis, prescription, or therapy. This is not an FDA clearance or approval announcement.

Related tags

  • #자람이
  • #FDA
  • #디지털치료제
  • #DTx
  • #RWD
  • #RWE
  • #AI 디지털케어로그
  • #De Novo
  • #Q-Submission
  • #발달장애
  • #의료기기 소프트웨어
  • #Jarame Therapeutic

Related articles

NoahAI가 서로 다른 시장국면을 판단하고 LEARNING, PAPER, LIVE 경로와 AI 어시스턴트를 안전 게이트로 연결하는 모습
보도자료

NoahAI Makes It Easier for First-Time Users to Start Without a Ready-Made Strategy

Public release v3.9.1.46 lets users begin with baseline operation, a guided example or their own strategy, then learn and simulate before considering live trading.

Read more →
글로벌 파트너들이 공개 홈페이지, 국가별 서비스, 문의 CRM, 정산 원장과 개인정보 보호형 메신저 알림을 하나의 운영 플랫폼에서 관리하는 모습
보도자료

Global Couple Care Brings Agency Onboarding, Customer Inquiries and Settlement Records Into One Place

International marriage agencies can manage public profiles, customer inquiries, consultation progress and fee records in one service, with optional messenger alerts for new work.

Read more →
노아AI 상용 운영: 시장 관찰에서 추론·체결·재학습으로 이어지는 금융 AI 운영 현장
보도자료

NoahAI hits 3.44 billion live events in its first month, shifting the financial-AI paradigm

Noah AI Labs released 30-day live KPIs for NoahAI: 3.445 billion events, 99.9914% processing success, 21,000 MAU, 1.64 billion inferences and 1.65 billion learning records, and 115.38 million USDT in filled volume.

Read more →