The study design is rarely the hard part. The data collection is.
Coordinators and fellows spend their time moving numbers between systems, and the study waits on them.
What investigators tell us
Information that could guide care or monitoring arrives a day or more after the event.
Quality metrics inside the hospital are limited to what existing reporting exposes.
Fragmented pipelines and separate review processes make multi-site collection the hardest part.
Studies in other specialties compete for smaller pools, so data work has to be cheap and reusable.
How many trials this applies to.
How many studies a year this applies to, counted from the public registry rather than estimated from market reports.
Interventional trials of any phase, and device, diagnostic or digital-health studies, that collect data from more than one electronic source and report into a study database.
Studies with wearables, home visits, several labs or many sites, where reconciling sources by hand is the bottleneck.
Registry-only and observational studies with a single data source, where there is little to reconcile.
The trial pool keeps growing: registrations have risen every year since the registry opened in 2000 13. Counts are from public registry data; the in-scope figure is our estimate, not a registry statistic.
One pipeline from your systems to the study database.
Nightingale reads your existing systems, checks the study documents against each other, cleans and standardizes every value, and delivers it with an audit trail. Sites keep their current workflow.
Check the rulebook
Every study document vs each other and vs FDA, EMA, ICH. Conflicts flagged before enrollment.
Example
Protocol in kilometers, lab manual in miles. Flagged on day one, not month four.Inventory what exists
Records, prior labs, device history and paper, filtered to what counts.
Example
Two years of blood pressure already in the chart get structured and counted.Find the gap
What the study needs minus what exists = what to collect.
Example
ECGs needed at weeks 4, 12, 24; records cover week 4 for a third of patients.Run the data
Map, convert, check, deliver into the EDC. No retyping.
Example
Watch reports 10 miles; 16.09 km lands in the right field, with the conversion recorded.Watch it daily
Gap re-checked on live data; early warning when an endpoint is at risk.
Example
Site 17's ECG timestamps drift 40 minutes. Flagged in week 3, not at lock.Prove every step
An audit trail for every data point, created or verified.
Example
Where did this glucose value come from? One click: device, time, conversion, reviewer.Any source in. Any destination out.
Connections are agreed with your IT team and run on patient-approved access. Nothing is limited to one vendor.
Epic, Cerner, Allscripts or anything else your sites run. With the patient's approval, Epic connects through MyChart using its standard data connection and returns the whole chart, not just one clinic's visits. New results come through as they happen, checked at least every five minutes.
CGMs, smartwatches, ECG and other connected devices. Device data comes through the manufacturer's cloud, such as Dexcom Clarity, once the patient consents; devices that pair with a phone app connect directly, and devices that only send to a handheld reader need one extra setup step.
LabCorp, Quest, central labs and your hospital lab systems, through connections agreed with your IT team.
Patient-reported outcomes, diaries and questionnaires, alongside paper forms and scanned source documents.
Delivered wherever the study keeps its data
A study database or research data store your team built, in the structure you specify.
Delivered into REDCap projects, field by field, with the conversions recorded.
Medidata Rave, Veeva Vault, Oracle Clinical, Castor and others, in the format the study requires.
In the standard format regulators expect, with the full record behind every value, for FDA or EU requirements.
Studies and tools this makes possible.
An automated pipeline keeps collection running while enrollment continues and the protocol changes, instead of a fixed extract per amendment.
Data arrives in minutes rather than the next day, so it can feed monitoring or a decision-support tool your team builds.
The same clean, structured feed supports quality metrics reporting, which is often easier to fund than a trial.
Each site connects its own systems; every site's data lands in one identical structure.
Inpatient records and at-home device data in one dataset, without a separate manual export.
The data work that usually needs a coordinator or a data manager is handled by the pipeline.
What your coordinators see at a visit.
One plain screen for visit days. It shows what needs doing, checks the timing and order the protocol sets while the visit happens, and sends checked values into the study database once the coordinator confirms them.
Today's visits and tasks in plain words, with every visit window counted from Day 1 and anything not booked at the top.
Seated rest before blood pressure, lying down before the ECG, blood before the morning dose, tubes spun in time. A timing problem shows up while the participant is still there.
Checked values wait for the coordinator's OK, then go into Rave or whichever EDC the sponsor uses. Nothing is typed twice.
Both demos run on a made-up knee arthritis study. No real patients or sites.
One study, several sites, one dataset.
Every site connects its own systems, and the data arrives in one structure with a single audit trail.
Your data stays with you.
The preferred model: the institution stores the data, Nightingale operates the pipeline.
The quality standard regulated life-science vendors are held to.
The FDA rule for electronic records.
An audit trail on every value, from source to submission.
Independent security audit underway.
Identifiable data minimized and handled only under a signed agreement.
Reviewed through the IRB process at our flagship deployment.
Study-specific logic you define, such as flagging possible AKI when creatinine crosses a threshold.
How this work usually gets paid for.
Nightingale is priced per study on participants, data sources and duration, so the cost is known before the work starts.
Data infrastructure written into the budget of a trial or pragmatic study application.
Framed around quality metrics and reporting for hospital leadership.
Institutional or departmental gift funding, where a named study has a clear patient-care case.
The sponsor or CRO pays, and the site gets the same clean data with no site-level cost.
From one study to a portfolio.
A 30-minute call on the study you have in mind, its sites and its data sources.
A complimentary review of one study: systems, formats, gaps and what the endpoints need.
Written findings, the connection plan, and a quote you can put in a budget or a grant.
Connect the sources and run alongside your current process, then compare before expanding.
Live on real patient data.
EmpowerMBC at Emory University, a Gilead-funded oncology study, is our live deployment and the reference we will put on a call. We baseline against your current process before claiming anything.
$340K contracted across signed agreements: Emory University plus two medical-device trials via Thinq CRO. Harvard iLab company, onboarding into the Georgia CTSA Program, and in conversation with several academic centers about multi-site studies.
What research teams ask first.
No. Nightingale is not a record system and not a database. It reads the systems you already run and delivers into whichever study database the protocol uses, in the format the study requires.
The preferred model is that your institution stores the data and Nightingale operates the pipeline. Identifiable data is minimized and handled under a signed HIPAA business associate agreement.
Through the standard connection the record system already offers, with the patient's approval, rather than a one-off export. New results come through as they happen, checked at least every five minutes. Device and lab data come through the manufacturer or lab connection, with consent.
Nothing to install and no second place to enter data. Coordinators get one plain screen for visit days: what needs doing, the timing the protocol sets, and checked values to confirm instead of retype. Manual transcription and query chasing go down. Try the coordinator demo.
We have been through the IRB process at our flagship deployment, hold a completed GxP vendor audit and Part 11 alignment, and have SOC 2 Type II in progress. We provide the documentation your review needs.
On participants, data sources and duration, so a quote is reproducible and can go into a grant or a departmental budget before the study starts. For industry-sponsored trials, the sponsor or CRO pays.
Start with one study.
Thirty minutes on the study you have in mind and the systems it would pull from.
Pick a time that works for you. We will walk through the study, the data sources and what a first connection looks like.
Book a callWhat happens next
A 30-minute call on the study and the systems behind it.
A complimentary review of the sources, formats and gaps.
Connect the sources, then measure against how you run today.
Sources behind every figure.
Show all 20 sources
- Tufts Center for the Study of Drug Development (2021): Phase III protocols average 3.6 million data points, three times the volume of ten years earlier.
- Marra C. et al. (2020), npj Digital Medicine: connected digital products in trials grew ~34% per year, 2000-2017.
- Tufts CSDD and Veeva Systems, eClinical Landscape Study (2017): 257 organizations; median 8.1 days from patient visit to EDC entry; 36 days on average from last patient last visit to database lock.
- Advarra, Study Activation Survey 2023: N>500 sites; ~69% use six or more sponsor systems.
- Garza M. et al. (2026): manual transcription error rate 8.23%; integrated capture showed no detected transcription errors.
- Stokman P. et al. (2021), Journal of the Society for Clinical Data Management 1(1), DOI 10.47912/jscdm.20: 20 Phase III studies, 1,939,606 queries; ~96,980 per study; 42% resulted in a data change.
- Smith Z., DiMasi J., Getz K. (2024), Therapeutic Innovation & Regulatory Science 58(5):855-862: direct cost of trial delay $55,716/day; opportunity cost on the order of $500,000/day.
- Salary.com, US Clinical Data Manager compensation, September 2026: average $144,625, ~$70/hour.
- Vendor-sponsored industry survey, n=88: ~12 hours/week per data manager per study on query and reconciliation work. Indicative.
- Moore T. et al. (2018), JAMA Internal Medicine: median cost per patient in pivotal trials ~$41,117. Per-source and fixed-cost components are estimates.
- IQVIA Institute, Global Use of Medicines: global medicine spending ~$1.6T. Medical device market ~$600B (industry estimate). WHO: global health spending ~$10T.
- Evaluate Pharma / IFPMA: pharmaceutical R&D ~$300B per year; medtech and digital-health R&D ~$50B (estimate). IQVIA Technologies, Veeva and Medidata combined revenue ~$10B.
- ClinicalTrials.gov: more than 35,000 new studies registered per year. The ~20,000 addressable interventional figure is an estimate.
- ClinicalTrials.gov sponsor records: Pfizer ~423 tracked / 251 active; Eli Lilly ~333. Order-of-magnitude proxy.
- ICH E6(R3): §3.16.1(c), §3.16.2(c), §2.12. Step 4 6 Jan 2025; EU 23 Jul 2025; FDA 9 Sep 2025; Annex 2 adopted 3 Jun 2026, effective 15 Jan 2027.
- FDA Digital Health Technologies guidance (Dec 2023); FDA Part 11 Q&A (Oct 2024); FDA and EMA AI principles (Jan 2026); FDA Real-Time Clinical Trials (28 Apr 2026).
- Disclosed transactions: Change Healthcare/Optum $13B; Datavant/Ciox ~$7B; Komodo Health $3.3B; Flatiron Health/Roche $1.9B; Truveta >$1B; Tempus AI market capitalization ~$10.6B.
- Public revenue multiples: Snowflake at IPO >100× trailing; Palantir ~80× (2025); Cloudflare ~35× (2025); Databricks ~27× (private, 2025).
- Private comparables: a September 2026 software seed round at $7M on $100M post-money with ~$500K lifetime revenue; a clinical-trial software company that raised under $400K in total and sold for roughly half a billion dollars. Details available on request.
- FDA, Considerations for the Design and Conduct of Externally Controlled Trials, draft guidance, February 2023. EMA, qualification opinion on prognostic covariate adjustment using digital twins (PROCOVA), September 2022.
- Assessment of Trends in the Design, Accrual, and Completion of Trials Registered in ClinicalTrials.gov by Sponsor Type, 2000-2019, JAMA Network Open (2020): of 245,999 interventional studies, 70,329 (28.5%) were industry-sponsored.