RISCONAPOSTOLUS

Evidence

Items Needing Review

34 items collected for review: evaluation lanes waiting on data, research gates waiting on the auditor, verdicts a harness holds and inputs that sit outside the repository.

Lane notes were recorded in Korean and are shown as written.

Evaluation lanes that are blocked · 10

A reference label, a linkage between records or approved access is missing. The lane’s rule stays written so it can run when the data exists.

5-1Blocked lane

Direct thermodynamic measurements with independent scaffold groups

현재 D3R 농도 환산 결과로는 1 kcal/mol 직접 실측 목표를 판정하지 않는다.

18-2Blocked lane

MFDS pill identification and DUR with MIMIC-IV 3.1 as a candidate

식약처는 참조 이미지와 제품 ID를 제공한다. MIMIC-IV는 인증이 필요하고 시험에 필요한 독립 오류 정답이나 조제 낱알의 동일 기록 결합이 확인되지 않았다.

19-1Blocked lane

Paired patient-level VCF and HPO with adjudicated labels

현재 공개 시험에는 논문이 이미 보고한 원인 변이만 있고 환자별 전체 VCF와 독립 ACMG/AMP 정답은 없다.

16-2Blocked lane

ADNI longitudinal assessments and registered cognitive training trials

등록 임상연구는 실제 연구의 근거이며 원자료 수준 문항 로그는 확보하지 못했다. ADNI는 자격 승인과 이용 조건이 있다.

16-3Blocked lane

ADNI / OASIS-3 / DementiaBank

검토한 자료에서 교차 모달리티의 동일 환자 키와 읽을 수 있는 종점이 확인되지 않았다. 이 목표의 외부 AUROC는 아직 측정할 수 없다.

18-1Blocked lane

FARSEEING observed falls and eICU-CRD 2.0

FARSEEING은 실제 센서 낙상을 요청으로 제공하며 모든 사건의 병동·독립 영상 시작·3초 전 관측이 확인되지 않았다. 자격이 필요한 eICU에서 사전 센서와 연결된 독립 낙상 시각을 확인하지 못했다.

18-3Blocked lane

Independent instrument and sponge counts linked to video

확보한 데이터에서 원래 요구한 수량 누락률 0%의 적격 실측 정답은 확인되지 않았다.

12-1Blocked lane

MIMIC-IV-Note

공개 설명에는 퇴원 요약·영상의학 보고서만 있고, 정규 정신과 관찰 시계열과 독립 급성 위기 시각 기록이 없다. 행 접근은 자격·교육·이용 동의가 필요하다. MHRN의 90일 연구는 사건 정의의 참고다.

12-2Blocked lane

Danish linked-register bereavement study

덴마크 레지스트리는 적격 기관 신청이 필요하다. 원저는 사건이 발생한 사례의 시간 비교 연구라 자체로 분류용 음성군을 제공하지 않는다. 공개 C-SSRS 분류 도구는 유가족 응답의 금기 판정표가 아니다.

12-4Blocked lane

AI-Hub Child and Adolescent Counselling Dataset

AI-Hub 안내는 만 7~12세와 다섯 상담 위기단계, 상담 일시·발화 경계 시각을 제공한다. 이 정보에서 학교별 12~17세 코호트 전체와 독립 사건·인계 시각은 확인되지 않았다. 이용 신청은 내국인 대상이다.

Lanes whose rows await access · 4

The split rule is locked. The rows are available only after the provider approves access.

16-3Rows awaiting access

DementiaBank ADReSS Challenge 2020

미디어와 공식 훈련 ID는 DementiaBank 멤버 접근이 필요하다. 공개 시험 ID만 로컬에서 대조했다. 한국어·기관 간·MRI 결합 외부 성능을 추정하지 않는다.

18-3Rows awaiting access

CholecTrack20 CVPR 2025

공식 접근 신청 및 DUA가 필요하다. 기구 궤적과 수술 도구 총수는 봉합 전 거즈 잔류의 독립 정답을 포함하지 않는다.

12-2Rows awaiting access

NIMH NDA ED-SAFE Trial

NDA 공개 목록만 확인했다. 환자 자료는 승인형이며 동일 사람의 퇴원 시각과 임상 종점 연결은 취득 전이다. 이 코호트는 유가족이나 생성 응답을 평가하지 않는다.

12-3Rows awaiting access

NHATS Sample Person and Other Person files

Public Use도 등록과 이용 조건 적용 대상이다. 환자·개인 수준 결과나 행을 외부 모델에 보내지 않는다. 고립 라벨과 결합 가능한 수동 센서가 확인되지 않았으며 이 종점은 자살 사건을 사용하지 않는다.

Research gates awaiting the auditor · 4

The four gates of the common protocol. Automated checks do not open them.

Auditor review pending

G1 · Data integrity and leakage control

Synthetic input hashes were verified. Research data eligibility and the split require auditor approval.

Blocked

G2 · Consistency of domain rules and evidence

Review of all research outputs follows G1 approval.

Pending

G3 · Counterexamples and independent cohort evaluation

The frozen synthetic counterexamples ran. The independent cohort and research model evaluations are pending.

Pending

G4 · Isolated replay and auditor sign-off

The synthetic results include a replay command. Isolated research execution and auditor sign-off are pending.

Held verdicts and outside inputs · 15

Items a harness session recorded as held or as depending on something outside the repository.

19-1Public-record audit

ACMG criteria claims audited against public records

The second model has not re-reviewed the literature step since it was fixed (checking and admission split into separate functions; no route into the ledger without the second reader). The last recorded verdict on that step is FIX_REQUIRED.

19-1Public-record audit

ACMG criteria claims audited against public records

The language-model trial ran through Claude Code subagents. The direct API route in model_runs.py has not been run, so token counts and list price are not recorded.

16-2Synthetic harness

Adaptive difficulty control on a psychometric virtual cohort

Stratum means, covariances and latency distributions are placeholders for ADNI and DementiaBank summary statistics that have not been imported.

16-2Synthetic harness

Adaptive difficulty control on a psychometric virtual cohort

Fatigue, time-of-day and lapse parameters are not identified by public summary statistics and remain assumptions.

16-3Synthetic harness

Multi-site MRI and speech fusion: cross-cohort validation on synthetic cohorts

External AUROC on real cohorts has not been measured. ADNI and OASIS-3 require approved access, and DementiaBank access differs by corpus.

16-3Synthetic harness

Multi-site MRI and speech fusion: cross-cohort validation on synthetic cohorts

The mapping of diagnostic labels and visit dates across cohorts is to be registered before any data are imported.

18-2Synthetic harness

Dispensing audit kernel: pill verification and DUR cross-check

No cases exist for LIQUID_FORM; its rate is held and not reported as zero.

18-2Synthetic harness

Dispensing audit kernel: pill verification and DUR cross-check

Replacing the simulated vision outputs with a model trained on the fixed ePillID split, and building a domestic gallery from the MFDS pill identification data, have not been done.

18-2Synthetic harness

Dispensing audit kernel: pill verification and DUR cross-check

The rule table must be generated from published DUR rule codes with their source versions; the table used here is virtual.

18-2Synthetic harness

Dispensing audit kernel: pill verification and DUR cross-check

An execution-time watchdog and a release interlock are design requirements that the harness does not implement.

12-2Synthetic harness

Risk trajectory and rule-constrained safety-plan support

Clinical review of the approved template wording is an external step (specification assumption A1).

12-3Synthetic harness

Consent-based on-device detection of isolation transitions

The relation that maps regional statistics to prevalence is an external input.

12-4Synthetic harness

Robust screening and immediate counsellor handoff

An authorised dictionary and an eligible corpus are external inputs.

8-1Synthetic harness

Privacy-preserving synthetic data engine and TSTR validation harness

The run on the fixed TCGA-BRCA institution split and the training of the main generator have not been done; they need a GPU environment.

8-1Synthetic harness

Privacy-preserving synthetic data engine and TSTR validation harness

Release of synthetic tables derived from MIMIC-IV depends on review of its data use agreement, which the release rule tested here does not cover.

Signatures · 1

Signature

Auditor sign-off

No result on this site carries an auditor signature.

© 2026 RISCON Co., Ltd.Privacy PolicyTerms of UseCookie SettingsContact

RISCON Labsrisconlabs.com