Clinical trial app development usually starts with the patient app, because that is the part a sponsor can picture and a supplier can demo. It is also the smallest part of the job. The FDA's own list of what the electronic systems behind a decentralised trial do runs to seven functions, and two of them face the participant; the rest belong to site staff, research nurses, monitors, sponsors and the service providers in between. Regulators inspect the platform those people use, and the phone in the patient's hand is only where some of the data starts. Build the permissions, the audit trail and the data flow for them first, and the patient app becomes the easy part it should be.
What does a clinical trial app have to do?
A lot more than the patient sees. The FDA's 2024 guidance on trials with decentralised elements lists what the electronic systems behind a trial are used for:
- managing electronic consent, including approved versions and signed forms
- capturing reports from remote trial staff, local clinicians and laboratories
- managing electronic case report forms
- scheduling visits and other trial activities
- tracking investigational product shipped to participants' homes
- syncing data from apps, wearables and sensors
- communicating between trial staff and participants
Only the last two touch the participant directly. The record does not live on the phone either. The FDA's guidance on digital health technologies (December 2023) treats the source data as whatever lands in the sponsor's "first durable electronic data repository", and says it does not intend to inspect individual devices when the data reaches that repository securely and as planned. The thing an inspector opens is the platform.
Who uses a clinical trial platform?
Half a dozen kinds of user at least, each with different rights to see and change data. PatientGo, the platform we have built and run for Illingworth Research Group for more than five years, has more than eight types of user, runs in 36 countries and 46 languages, and has more than seven external systems plumbed into it. The participant is one of those users.
User | What they do in the platform | What the rules ask of it |
|---|---|---|
Participant | Visit details, questionnaires, reminders, expenses | Once it collects trial data, the app is a "data acquisition tool" |
Research nurse on a home visit | Patient-specific forms, often with no signal | Personal details shared only with those authorised |
Site staff | Patients, visits, data entry, setting patients up on devices | Access matches the investigator's delegation |
Investigator | An overview of each participant's data | Data that is "relevant, meaningful and manageable" |
Monitor | Reviews data and raises queries | No unblinding information unless the design needs it |
Sponsor and service providers | Oversight, logistics, payments | A record of every system, user, role and permission |
What runs through all six rows is the permission model. ICH E6(R3), the good clinical practice guideline adopted in January 2025, asks for access assigned "based on a user's duties and functions, blinding arrangements and the organisation to which users belong", revoked when it is no longer needed. It asks for logs of every account created and every role changed, and for audit trails that show who entered or changed data, when and why, and that nobody can switch off. Every one of those requirements sits on the staff side of the platform, which in our experience is where most of the build goes.
What changed for trial software in 2026?
UK law caught up with the guideline. Since 28 April 2026 the amended clinical trials regulations have made ICH good clinical practice a legal requirement in the UK, and they name "the development and maintenance of trial specific computerised systems" as a sponsor function in so many words. They extend retention of the trial master file, electronic documents included, from 5 years to 25, and they require sponsors to publish a summary of the results within 12 months of the trial ending and to offer participants a version written for a lay reader. A platform built now has to keep its records readable through several generations of technology, and has to be able to reach every participant at the end.
The next change is already published. Annex 2 of E6(R3), covering trials with decentralised elements, comes into effect in the EU on 15 January 2027. It classes a patient app that collects trial data as a data acquisition tool, which brings it under every computerised-system requirement above, and asks sponsors to make sure that data reaches the investigator "in a way that is relevant, meaningful and manageable". A patient app that streams readings nobody at the site can make sense of fails that test, however good it looks in the demo.
Where do clinical trial apps go wrong?
At the site, more often than on the patient's phone. A Tufts CSDD survey of 355 research sites (2024) found that half of them saw decentralised-trial technology as more burdensome than running a trial the traditional way. A 2025 Tufts study heard sites describe the work of setting patients up on electronic questionnaires, helping them through it, and managing a different system for every sponsor. When participants do hit a problem with trial technology, the site is who they call: in CISCRP's 2025 survey of past participants, 49% of those with a technical problem contacted the site or study staff.
The same survey puts the patient app in proportion. 72% of participants who used trial technology had no problems with it, and among those who found taking part disruptive, the most common reason was travelling to the clinic (49%). Meanwhile the data keeps growing: Tufts puts the average phase III protocol at 5.9 million datapoints, up 11% a year since 2020, and every one of them has to land somewhere a site can manage it.
Where should a clinical trial app build start?
With the permission model, the audit trail and the data flow: who can see what, who can change what, and where every piece of data ends up. Then the site and nurse tools, then the patient app. That order is ours, from running PatientGo and building SourceGo, the forms research nurses fill in on home visits, which work offline because a patient's home may have no signal (the subject of our post on offline mobile apps). On that foundation the patient app can stay simple, which is what participants need, and it can be translated and checked screen by screen in each language, as the 2025 good practice for electronic outcome assessments asks. Built the other way round, the audit trail gets retrofitted and the site ends up keeping a spreadsheet beside the platform.
Regulated software costs more for reasons we have set out before. This is where in a trial platform that work sits, and it is most of what our healthcare and life sciences work involves.
If you are planning a clinical trial platform, or have a patient app the sites struggle to work with, get in touch.
