Patient Journey Intelligence Cookbook: How to Use

Patient Journey Intelligence is best understood as a set of clinical and operational workflows. This cookbook gives you 16 recipes and 10 end-to-end scenarios covering the path from raw clinical data to cohorts, datasets, curated fields, de-identified outputs, and registry exports. This overview explains how the library is organized; each recipe and scenario has its own page with the full step-by-step workflow.

How the tutorials library is organized

• Scenarios are complete, end-to-end demo paths that chain several building blocks together into a single narrative, from source ingestion through to a reviewed, exportable result. Five scenarios have an accompanying video walkthrough. Each scenario page ends with a Recipe reference section documenting the goal, features, edge cases, and value of each building block it uses on its own.
• Standalone recipes are the few building blocks that don't belong to any single end-to-end scenario - they keep their own dedicated page.
• Review each recipe or scenario in the same four layers: the goal, the required preparation, the Patient Journey Intelligence features involved, and the expected output.
• The steps are reference paths. In a live deployment, some steps may be skipped, renamed, unavailable, or handled by different roles depending on configuration, data readiness, and governance requirements.

Recipe library

Most recipes are demonstrated end-to-end inside a single scenario, so their full detail - when to use them on their own, features involved, edge cases, and value - lives in the Recipe reference section of the scenario page(s) that use them, right alongside the steps that implement them.

#	Recipe	Documented in
1	Ingest clinical documents from S3 and prepare them for analysis	Scenario: S3 to cohort to dataset to Data Curation
2	Create a patient cohort from the Assistant	S3 to cohort to dataset to Data Curation, Cohort to care gap to Patient Timeline, Assistant cohort to dataset to De-Identification
3	Build a dataset from a cohort or ingestion results	S3 to cohort to dataset to Data Curation, Trial match to cohort to dataset, Assistant cohort to dataset to De-Identification, Dataset to De-Identification to external sharing
4	Run Information Extraction on a dataset	Trial match to cohort to dataset
5	Start Data Curation from a dataset	S3 to cohort to dataset to Data Curation
6	De-identify a dataset before sharing or review	Assistant cohort to dataset to De-Identification, Dataset to De-Identification to external sharing
7	Run care gap analysis from a saved cohort	Cohort to care gap to Patient Timeline, Guideline question to cohort criteria to care gap
8	Match patients to a clinical trial and reuse the cohort	Trial match to cohort to dataset
9	Curate a registry project with human review	Registry project to case finding to export
10	Ask guideline questions and apply findings to a cohort workflow	Guideline question to cohort criteria to care gap
11	Review a Patient Timeline before taking action	Cohort to care gap to Patient Timeline, Patient Timeline to source document review

Standalone recipes

These recipes aren't demonstrated inside one specific end-to-end scenario and keep their own page:

#	Recipe	Best for
12	Use Information Extraction results to guide Data Curation	Reducing reviewer noise before full curation

Advanced Data Curation: capturing facts that do not exist in OMOP

Most recipes and scenarios above mostly move and reshape data the OMOP Common Data Model already holds. Recipes 13–16 use Data Curation to pull clinically decisive facts that live only in free-text notes and have no native home in OMOP's coded domains, such as condition, drug, measurement, procedure, visit, or observation.

Because these fields are not standard OMOP concepts, each recipe defines a custom registry in Data Curation Studio rather than using a built-in preset. The field lists on each page are illustrative starting points, not shipped presets.

Secondary-use rigor applies to every recipe in this section:

• Version the custom registry so an extraction run is reproducible and auditable.
• Keep evidence spans so every curated value links back to the source sentence for review and provenance.
• Treat "not documented" as absence of documentation, not absence of the fact.
• Map free-text terms back to standard concepts before joining curated output to the cohort.
• Use human-in-the-loop review before downstream research, safety, or quality decisions.

#	Recipe	Best for
13	Curate reason for medication discontinuation or switch	Tolerability, adherence, and pharmacovigilance analysis
14	Curate implanted-device / UDI details for safety surveillance	Device-safety surveillance and recall response
15	Curate cancer recurrence events	Disease-free-survival endpoints for oncology research
16	Curate family history with pedigree detail	Hereditary-risk stratification and genetic-counseling referral

Scenario list

Scenario	Best for	Workflow	Video
S3 to cohort to dataset to Data Curation	Full journey from raw files to curated clinical data	S3 ingestion -> Cohort -> Dataset -> Data Curation	Yes
Trial match to cohort to dataset	Trial feasibility and recruitment planning	Trial Match -> Cohort -> Dataset	Yes
Structured FHIR to OMOP	Fast ingestion of structured clinical facts into OMOP	FHIR ingestion -> OMOP CDM -> Patient Journey	Yes
Import an external OMOP CDM database	Connecting an existing OMOP CDM database to Patient Journey	PostgreSQL OMOP connection -> Connection test -> Import -> Patient Journey Gold OMOP	Yes
Cohort to care gap to Patient Timeline	Quality improvement and care-management workflows	Cohort -> Care Gap -> Patient Timeline	Yes
Assistant cohort to dataset to De-Identification	Privacy-first document preparation	Assistant cohort -> Dataset -> De-Identification	-
Registry project to case finding to export	Registrar workflows and human-in-the-loop abstraction	Registry Project -> Case Finding -> Curation -> Export	-
Guideline question to cohort criteria to care gap	Clinical evidence-driven quality workflows	Guideline Question -> Cohort Criteria -> Care Gap	-
Patient Timeline to source document review	Reviewer workflows and data-quality triage	Patient Timeline -> Source Document Review -> Correction or Escalation	-
Dataset to De-Identification to external sharing	Governed external sharing of reviewed de-identified data	Dataset -> De-Identification -> External Sharing Package	-

How to read this cookbook

This cookbook is organized as a library rather than a sequential manual. Start with the goal, persona, or workflow that needs to be supported, then open the recipe or scenario that most closely matches it.

For demos, begin with the scenario list and pick the path that best represents the intended workflow, such as full curation, trial matching, registry operations, care gaps, evidence review, or de-identification. For implementation planning, focus on required inputs, permissions, edge cases, and deployment-specific limitations on each recipe page.

Common workflow dependencies:

1. Ingest data before you can search, curate, or analyze it.
2. Create a cohort before creating a dataset from a patient population.
3. Create a dataset before starting Data Curation or De-Identification.
4. Use extraction, curation, registry, or care-gap outputs as review artifacts, exports, or follow-up analysis inputs depending on what is enabled in the deployment.

Workflow prerequisite map

Ingestion
  -> Cohort
      -> Dataset
          -> Information Extraction
          -> Data Curation
          -> De-Identification
              -> Review
                  -> Export / Share / Follow-up Analysis

Alternative dataset path:

Ingestion
  -> Dataset from ingestion filters
      -> Information Extraction / Data Curation / De-Identification

Structured FHIR path:

Structured FHIR
  -> OMOP CDM
      -> Cohort / Patient Journey / Database Explorer

Key dependency rules:

• Cohorts are reusable patient populations.
• Datasets are working sets of patients, documents, and images.
• Data Curation and De-Identification require a dataset.
• Information Extraction can be used before Data Curation to focus review, but direct result handoff should be confirmed per deployment.
• Structured FHIR ingestion can populate OMOP directly when the source bundle already contains the required coded clinical facts.

Persona-based scenario guide

Persona	Most relevant workflows	Typical starting point	Common output
Clinician	Assistant cohort creation, guideline questions, care-gap review, patient timeline review	Clinical question or patient list	Cohort, care-gap list, patient context
Data Curator	Dataset creation, Information Extraction, Data Curation, De-Identification	Cohort or ingestion-filtered documents	Reviewed extracted data, de-identified package
Registrar	Registry project setup, case finding, curation, validation, export	Registry project and patient cohort	Reviewed registry cases and export-ready results
Data Engineer	Source setup, S3 ingestion, ingestion monitoring, dataset readiness	Source files or configured data source	Ingested documents and usable patient/document assets
Admin	User access, configuration, audit, readiness checks	Deployment or team setup	Secure, governed Patient Journey Intelligence workspace

Feature-to-use-case mapping

Patient Journey Intelligence feature	User-facing use cases	Typical user
Data Sources	Bring clinical files, FHIR resources, and external OMOP databases into Patient Journey Intelligence	Data engineer, admin
Ingestion Jobs	Monitor, retry, and validate data loading, including structured FHIR and external OMOP ingestion	Data engineer, admin
Assistant	Ask clinical data questions and create cohorts	Clinician, analyst
Cohort Builder	Save reusable patient populations	Analyst, researcher
Dataset Builder / Explorer	Prepare the required input for curation, extraction, or de-identification	Data curator, analyst
Information Extraction	Find targeted facts in documents	Data curator, SME
Data Curation Studio	Produce structured, evidence-linked outputs	Data curator, registrar
De-Identification	Review and remove sensitive information before sharing	Privacy officer, curator
Clinical Trial Matcher	Create cohorts from trial eligibility criteria	Research coordinator
Care Gap Analyzer	Identify patients missing recommended care	Quality team, care manager
Clinical Guidelines	Ask cited guideline questions	Clinician, medical affairs
Patient Timeline	Review patient history and supporting context	Clinician, registrar, analyst
Patient Registry	Manage case finding, abstraction, review, validation, and export	Registrar, reviewer
Standard / Custom Registries	Define or manage abstraction requirements	Registry admin, SME
Medical Terminology	Resolve clinical terms and support criteria building	Analyst, clinician
DB Explorer	Validate data and inspect analysis outputs	Analyst, admin
Admin, access, audit, license	Manage users, permissions, compliance, and configuration	Admin
Custom Registry for OMOP-gap fields	Extract decision-critical facts absent from coded data	Data curator, researcher, SME
Curated Output Export / OMOP Join	Join note-derived structured fields back to a cohort	Data curator, analyst, researcher
Evidence and Version History	Provide source-linked provenance and reproducible runs	Data curator, registrar, auditor

Workflows that need product confirmation

These items should be confirmed before being presented as fully supported customer workflows:

1. Data Curation results to Assistant cohort creation: Querying curation outputs in the Assistant and immediately creating a cohort from those results may require export or additional configuration.
2. Information Extraction to Data Curation handoff: Extraction results can guide review, but direct reuse inside a curation job should be confirmed per deployment.
3. De-Identification export/save: Review may be available before full export or permanent save is enabled.
4. Registry external submission: Project export may be available even if direct submission to an external registry is not configured.
5. Custom registry lifecycle: Draft, test, publish, archive, and reuse behavior should be clarified.
6. Dataset dependency behavior: Deleting or changing datasets used by curation, extraction, or de-identification jobs should be documented clearly.
7. Assistant behavior over exported curation data: Availability depends on whether exported results are registered and queryable in the target environment.
8. Care-gap prerequisites: Required source data for each condition should be defined so you understand open or missing results.

Suggested customer demo scenarios

Demo 1: raw documents to curated oncology data

• Ingest oncology documents from S3.
• Create a breast cancer cohort in the Assistant.
• Build a dataset from the cohort.
• Start Data Curation with an oncology or registry template.
• Review extracted fields with evidence.
• See Scenario: S3 to cohort to dataset to Data Curation for the full walkthrough.

Demo 2: fast biomarker discovery before curation

• Build a dataset from recent pathology reports.
• Run Information Extraction for biomarker values.
• Use the findings to refine the dataset.
• Start Data Curation on the focused dataset.
• See Recipe 12: Use Information Extraction results to guide Data Curation.

Demo 3: privacy-first data sharing

• Create a cohort for demo or external review.
• Build a dataset from the cohort.
• Run De-Identification.
• Review and export de-identified outputs if enabled.
• See Scenario: Assistant cohort to dataset to De-Identification.

Demo 4: trial feasibility

• Search for a clinical trial.
• Run eligibility matching.
• Save matched patients as a cohort.
• Build a dataset for deeper document review.
• See Scenario: Trial match to cohort to dataset.

Demo 5: care gap worklist

• Create a chronic disease cohort.
• Run Care Gap Analyzer.
• Review a patient timeline for an open gap.
• Export the outreach list.
• See Scenario: Cohort to care gap to Patient Timeline.

Demo 6: registry abstraction

• Create a registry project.
• Select a cohort.
• Run case finding and extraction.
• Review evidence, validate fields, and export results.
• See Scenario: Registry project to case finding to export.

Demo 7: guideline-driven care gap

• Ask a guideline question.
• Turn the recommendation into cohort or gap criteria.
• Run Care Gap Analyzer.
• Review one patient timeline for supporting context.
• See Scenario: Guideline question to cohort criteria to care gap.

Demo 8: patient evidence review

• Open a patient from a care-gap, registry, or curation result.
• Review the timeline.
• Open the supporting source document.
• Correct or escalate the result where the workflow allows.
• See Scenario: Patient Timeline to source document review.

Demo 9: external sharing package

• Build a dataset from a cohort.
• Run De-Identification.
• Review the de-identified output.
• Export or save the package if enabled.
• See Scenario: Dataset to De-Identification to external sharing.

Demo 10: structured FHIR to OMOP

• Register or select a structured FHIR R4 source.
• Run structured FHIR ingestion.
• Map Conditions, Medications, Observations, Procedures, Encounters, and Visits into OMOP.
• Validate the imported facts in Patient Journey, Cohort Builder, or Database Explorer.
• See Scenario: Structured FHIR to OMOP.

Demo 11: external OMOP CDM import

• Connect to an external PostgreSQL OMOP CDM database.
• Test the connection and preview importable OMOP tables.
• Configure patient ID remapping.
• Import demographics and clinical data into Patient Journey Gold OMOP.
• Validate imported patients in Patient Journey, Cohort Builder, or Database Explorer.
• See Scenario: Import an external OMOP CDM database.

Frequently asked questions

A cohort is a reusable patient population saved for later use across workflows. A dataset is the working set of patients, documents, and images built from a cohort or ingestion filters - Data Curation, Information Extraction, and De-Identification all require a dataset, not just a cohort. See the Recipe reference sections in Scenario: S3 to cohort to dataset to Data Curation.

You need a dataset. A cohort defines the patient population; a dataset adds the document, image, and note-type scope that Data Curation and De-Identification require as input. See the "Build a dataset from a cohort or ingestion results" recipe reference in Scenario: S3 to cohort to dataset to Data Curation for how to build one from a cohort or directly from ingestion filters.

Extraction results can guide review and help you narrow a dataset before curation, but direct reuse inside a curation job should be confirmed per deployment. See Use Information Extraction results to guide Data Curation and the Workflows That Need Product Confirmation section above.

Use structured FHIR ingestion when the source bundle already contains the coded clinical facts you need: conditions, medications, observations, procedures, encounters, and visits map directly into OMOP CDM without NLP processing. Free-text notes still require document ingestion and NLP if you need to extract facts from narrative text. See Structured FHIR to OMOP.

Confirm the dataset scope matches the approved sharing request, review every detected identifier against the original documents, and validate outputs before external delivery. Export and save behavior vary by deployment, so confirm what's enabled before promising a complete handoff. See Dataset to De-Identification to external sharing.

Recipes 13 through 16 use Data Curation with a custom registry to structure medication discontinuation reasons, implanted-device details, cancer recurrence events, and family history - facts that live only in free-text notes and have no native OMOP field. See the Advanced Data Curation section above.