Coverage
Plausible clinical/source-system coverage runs from June 18, 2002 to May 6, 2026. Structured symptom and rating measurements run from June 18, 2002 to November 3, 2025.
Behavioural health data, prepared for controlled use
Privacy-first behavioural health data for research, analytics, and AI development.
DJO DataLabs prepares de-identified and synthetic clinical datasets from real-world psychology-supervised care, with structured privacy review, governance, and utility validation before release.
Privacy first
Structured controls for direct and indirect identifiers.
Utility aware
Datasets shaped for analysis, modelling, and product testing.
Governed release
Review steps before data leaves the protected environment.
Currently available
Tier A De-identified Longitudinal Behavioural Health Dataset.
Tier A is our first available de-identified real-world behavioural health dataset, built from psychology-supervised care delivered in Canada. It contains structured longitudinal symptom/functioning ratings, treatment/session metadata, and constructed episode-level outputs designed for outcome analysis, utilization-response modelling, cohort analysis, and controlled AI product evaluation.
The release is positioned as a behavioural health outcomes and utilization dataset, not a claims, pharmacy, laboratory, or full EHR export. Raw narrative notes, psychological reports, identifiable documents, qEEG files, and unstructured free text are excluded from the current commercial package.
112,877
Repeated structured clinical rating measurements
8,234
Clients represented in treatment/session metadata
9,993
Constructed treatment episodes
10
Primary clinical rating domains with confirmed directionality
Outcome layer
The buyer-facing release includes 8,693 strict QA outcome rows after excluding sparse non-primary OCB and Other domains from outcome tables.
Privacy posture
The package excludes direct identifiers, exact dates, raw free text, start and stop times, provider names, reports, source files, and raw file paths.
Upcoming
Broader clinical data lake under development
Beyond Tier A, DJO DataLabs is developing a broader clinical data lake that consolidates de-identified, real-world behavioural health data across multiple service lines. Future layers may include de-identified demographics, referral sources, presenting concerns, DSM diagnostic impressions, standardized assessment results, symptom and functioning measures, treatment episodes, session-level service utilization, report metadata, and longitudinal outcome measures.
Where available and appropriate for release, the broader lake may also support datasets derived from psychoeducational, psychological, MVA, WSIB, and related assessment workflows, including psychometric test outputs, clinical interview variables, qEEG/neurofeedback-related data, treatment recommendations, functional impairment indicators, and diagnosis-linked assessment profiles.
The objective is to convert historically fragmented clinical information into structured, governed data assets suitable for internal quality improvement, cohort analysis, research collaboration, AI-readiness, and controlled commercial evaluation.
PHIPA-aligned planning
PIPEDA-aware controls
Canadian de-identification expectations
What we do
We turn real behavioural health data into governed de-identified and synthetic datasets.
DJO DataLabs helps convert historically fragmented behavioural health records into structured, privacy-reviewed data assets. We work with real-world clinical datasets, remove or generalize identifying signals, document the release structure, and prepare governed packages for research, analytics, model development, and commercial evaluation.
De-identified real-world datasets
Structured clinical data prepared with direct identifiers excluded and release documentation attached.
Synthetic data derivatives
Fit-for-purpose synthetic datasets designed to preserve analytic utility while reducing exposure risk.
Governed evaluation packages
Dataset cards, data dictionaries, sample packages, and controlled access pathways for qualified partners.
Who it is for
Built for qualified teams evaluating behavioural health data products.
Mental health AI teams
Evaluate outcome-oriented workflows, prototype model concepts, and test behavioural health analytics before pursuing deeper data partnerships.
Clinical research groups
Study symptom trajectories, treatment response, utilization patterns, and episode-level outcomes in real-world psychotherapy and assessment-linked care.
Digital health and therapeutics companies
Explore longitudinal behavioural health patterns that can support product validation, evidence planning, and research collaboration.
Commercial analytics teams
Build dashboards, forecasts, cohort models, and market analyses using realistic care-pattern data with governed release controls.
Governance and release controls
Prepared for controlled buyer review.
DJO DataLabs is being developed around a privacy-first release model. Commercial or research packages are prepared with direct identifiers excluded, high-risk fields removed or generalized, and release documentation provided to support buyer review.
- Direct identifiers excluded before release
- No raw clinical notes, reports, source files, provider names, raw file paths, or unreviewed free text in commercial packages
- Exact dates, start/stop times, and high-risk temporal fields removed, shifted, or generalized where appropriate
- Dataset card and data dictionary available for controlled evaluation
- NDA, paid evaluation access, and partner-specific review pathways available where appropriate
- Governance approach designed with PHIPA, PIPEDA, and Canadian de-identification expectations in mind