For CROs and sponsors in Japan
Keep every system. Change what connects them.
AccuraTrials reads the protocol once and builds one governed, machine-readable model of the study. From that model it generates the build, computes every amendment and checks every entry, each step citing its page. It sits above the Medidata, Veeva, Oracle and domestic systems you run today, and it never asks you to replace them.
Why Japan, now
Japan is joining global trials earlier. The wait has moved to execution.
Three dated facts from Japan’s regulators, and our reading of what they mean for the teams who build and run studies.
-
December 2023 · MHLW
A Japanese Phase 1 study is, in principle, no longer required before Japan joins a global trial.
Notification 1225-2, listed on PMDA’s multi-regional trials page
-
FY2025 H1 · PMDA
239of 328
initial clinical trial notifications in Japan were for multi-regional trials.
PMDA FY2025 business results, December 2025
-
FY2024 · PMDA
Review lag0 yrsDevelopment lag2.0 yrsOur reading: review is no longer the delay. Trial execution is.
PMDA FY2025 business results, FY2024 drug-lag estimate
Every week taken out of study build makes Japan a stronger place to run the next global trial.
Trials today
Your trial runs on one PDF and dozens of systems.
Every system is set up by a person reading the protocol again and typing it in. That relay is where startup time, rework and late deviations come from. The friction is between the systems, not inside them.
USD 50–250k
per Phase I or II study
EDC build, configuration, validation and site onboarding
3–6 months
protocol readiness to launch
Much of it spent re-reading and re-keying one document
20–50 systems
per trial
Sometimes more than 75, each set up by hand
Typical ranges reported to AccuraTrials by clinical operations leaders in Japan, 2026. Your own figures replace these in a proposal.
One protocol, start to finish
Six moments. Each with a page, a name and a receipt.
Follow one protocol from the day it arrives to the day a site saves a visit date. The fragments on the right are illustrative, not study records.
Read in full, never retyped.
The protocol, its amendments, the investigator’s brochure, the SAP, the consent form, the lab manual and the governing SOPs enter as one source set, exactly as written.
A page on every fact, or a person decides.
Every fact carries its page and passage, and every number in the quote is checked against the source. What the page cannot support goes to a person. It is never dropped or quietly fixed.
The AI proposes. A person approves.
The AI maps each fact into one governed model on CDISC USDM and ICH M11. A named reviewer approves it with the clause beside the item, and the history is append-only.
Weeks of configuration become days of review.
Forms, fields, checks and visits are generated from the approved model by versioned code, each one citing its clause. Four or more weeks of manual build becomes days of review.
Impact known before anything changes.
A full reissue or a standalone amendment enters the same path. Its impact on forms, fields, checks, schedules, the consent form and the SAP is computed first. Weeks of impact-matrix work becomes hours.
Found at the keyboard, not in a listing.
The protocol’s own rules become deterministic checks beside yours. A visit outside its window is flagged on save with the clause attached, in minutes rather than weeks later in a listing review.
Moments 01 to 03 run in the AccuraTrials platform. Moments 04 to 06 run in AccuraTrials EDC, built to 21 CFR Part 11 and live in a paid study: proof that the model runs, not a system you must adopt. In your own EDC and CTMS, each one is an integration path validated together in a pilot.
Why AI can be trusted here
Built for the point where AI fails.
Most AI answers from memory, sometimes invents, and cannot show its work. Here nothing touches a study without a page, a name and a receipt.
- 01
AI drafts. Code executes.
Deterministic, versioned code runs every rule, so the same input gives the same output. Whether a subject is included or excluded is decided by code, not by a model.
- 02
No source, no action.
What cannot cite its page is not released, and what cannot be verified is never marked safe. Cited or quarantined; there is no third bucket.
- 03
The audit trail is the product.
Every approval records its clause or SOP. Models are part of the record: they change visibly, never silently.
- 04
A person approves what matters.
Upstream, by name. In the EDC, with an electronic signature aligned with 21 CFR Part 11.
Built for Japan
One model above the systems you already trust.
Sponsors and CROs in Japan have standardized on Medidata, Veeva, Oracle and domestic platforms. Switching is expensive, and we do not ask anyone to switch. The model sits above the stack.
What we heard in Japan
How clinical operations leaders ranked their needs. The ranking is our requirements list.
- Setup automationHigh
Study build in days of review.
- Protocol-to-CRF traceabilityHigh
Every form, field and check cites its clause.
- AI-assisted configuration, human reviewGrowing
AI proposes; a named reviewer approves.
- Replacing an established EDCLow
So we do not.
The regulatory frame
Japan’s rules, mapped to controls.
Verified against PMDA, MHLW and JCROA sources in September 2026. Regulatory fit for your own study is reviewed together in the pilot.
- ER/ES guidelineMHLW sets the expectations for electronic records and signatures: authenticity, legibility, preservation.
- CDISC data acceptedPMDA begins accepting CDISC-standard electronic study data with new drug applications.
- CDISC data requiredElectronic study data becomes required for new drug applications filed from this date.
- Notification 1225-2A Japanese Phase 1 study is, in principle, no longer required before a global trial.
- JCROA commentaryThe ER/ES commentary is revised for cloud systems and audit trails.
- GCP Ordinance revisionMHLW is revising J-GCP to reflect ICH E6(R3).
| Requirement | What it asks for | How AccuraTrials answers |
|---|---|---|
| CDISC-standard study data | Electronic study data in CDISC standards with new drug applications, required since April 2020 | The study is modelled in CDISC USDM, with ODM and CDASH, from day one, so collection stays consistent to the last dataset |
| J-GCP and ICH E6(R3) | Risk-based quality management, proportionate oversight and data integrity | Critical rules are explicit before enrollment, each with its clause and the name of the person who approved it |
| MHLW ER/ES and 21 CFR Part 11 | Authenticity, legibility and preservation of electronic records, with audit trails | An append-only, tamper-evident audit trail, named approval, and EDC electronic signatures aligned with Part 11 |
What does not change, and what is out of scope
Stated plainly, so nothing has to be discovered in a pilot.
- Nothing is replaced
- Your systems and their records stay the record. Your SOPs remain the controlling documents.
- Open standards
- OpenStudyBuilder, the open-source study definition platform started by Novo Nordisk, with CDISC USDM, ICH M11, ODM, CDASH, LOINC and UCUM.
- Out of scope today
- Medical coding with MedDRA and WHODrug, and submission datasets (SDTM, Define-XML). The platform works upstream: protocol to build, amendments and capture.
- Your data stays yours
- No training on protocols or patient data. Subject-identifiable data never enters the shared model.
A proof of concept on a protocol you choose
Bring one protocol. Leave with a machine-readable study.
Phase I or II to start, with English protocols today. Scope, duration, cost, NDA, hosting region, model provider, data flow and validation responsibilities are agreed on paper before work starts.
You bringWhat we ask of you
- One protocol with its amendments
- One named reviewer for approvals
- One EDC or CTMS contact
- Sponsor permission for the protocol and any sponsor-owned system
You keepWhat you leave with
- The study as CDISC USDM
- The cited fact ledger
- The amendment diff
- The generated build
- An audit trail export
FAQ
AccuraTrials in Japan, answered plainly.
Updated October 2026
Does AccuraTrials work with CROs and sponsors in Japan?
Yes. AccuraTrials offers CROs and sponsors in Japan a proof of concept on a Phase I or II protocol they choose. The platform builds one governed, machine-readable model of the study from the protocol, then generates the build, computes amendment impact and checks entries, each step citing its page. The product interface and documentation are in English. AccuraTrials, Inc. is based in Madison, Wisconsin, USA.
Do we have to replace our EDC, such as Medidata, Veeva, Oracle or a domestic EDC?
No. AccuraTrials sits above the systems you run and does not ask anyone to switch. Your systems and their records stay the record, and your SOPs remain the controlling documents. Connections to your EDC and CTMS are integration paths validated together in a pilot. AccuraTrials EDC, built to 21 CFR Part 11 and live in a paid study, is optional.
Does AccuraTrials support PMDA’s requirement for CDISC electronic study data?
PMDA has accepted CDISC-standard electronic study data with new drug applications since October 2016 and has required it for applications filed from April 2020. AccuraTrials models the study in CDISC USDM, with ODM and CDASH, from day one, so collection stays consistent with the standards to the last dataset. Producing submission datasets (SDTM and Define-XML) is out of scope today.
How does AccuraTrials fit J-GCP and ICH E6(R3)?
MHLW is revising the GCP Ordinance to reflect ICH E6(R3), with its emphasis on risk-based quality management, proportionate oversight and data integrity. AccuraTrials makes a study’s critical rules explicit before enrollment, each with the protocol clause behind it and the name of the person who approved it. Regulatory fit for a given study is reviewed together in the pilot.
Does AccuraTrials support Japan’s ER/ES guideline for electronic records?
Japan’s ER/ES guideline (MHLW, 2005) and JCROA’s 2024 commentary for cloud systems ask for authenticity, legibility and preservation of electronic records, with audit trails. AccuraTrials provides an append-only, tamper-evident audit trail, approval by a named person and, in its EDC, electronic signatures aligned with 21 CFR Part 11. These controls are designed to support the guideline; fit for your system is reviewed together in the pilot.
Can AccuraTrials read Japanese-language protocols?
English protocols are supported today. Japanese-language protocols are a joint objective with pilot partners, written into the pilot’s success criteria rather than promised in advance.
What does an AccuraTrials proof of concept involve?
You bring one Phase I or II protocol with its amendments, one named reviewer for approvals and one EDC or CTMS contact, with sponsor permission for the protocol and any sponsor-owned system. You keep the study as CDISC USDM, the cited fact ledger, the amendment diff, the generated build and an audit trail export. Scope, duration, cost, NDA, hosting region, model provider, data flow and validation responsibilities are agreed on paper before work starts.
Is our protocol or patient data used to train AI models?
No. AccuraTrials does not train models on protocols or patient data, and subject-identifiable data never enters the shared model.
Why does Japan matter for global clinical trials now?
In December 2023, MHLW Notification 1225-2 removed, in principle, the need for a Japanese Phase 1 study before Japan joins a global trial. In the first half of FY2025, 239 of 328 initial clinical trial notifications in Japan were for multi-regional trials, according to PMDA. PMDA put the FY2024 review lag at 0 years and the development lag at 2.0 years. Our reading is that the bottleneck is now trial execution, not review.
Sources
- PMDA, Multi-Regional Clinical Trials: PSB/PED Notification 1225-2, 25 December 2023.
- PMDA FY2025 business results (December 2025): clinical trial notification counts and the FY2024 drug-lag estimate.
- PMDA, New Drug Review with Electronic Data.
- MHLW PFSB Notification 0401022 (2005), the ER/ES guideline; JCROA ER/ES Guideline Commentary, revised 26 March 2024.
- Setup ranges reported to AccuraTrials by clinical operations leaders in Japan, 2026.
Fragments on this page are illustrative and are not study records. This page describes a system under development and is not a regulatory representation.
Read once. Never retyped.
Bring one protocol and one workflow your team wants to improve. Nami, our guide, can answer the common questions and pass your note to the team, who reply by email.
Or email us: jgui@accuratrials.com