All patients.

Yes, we use surrogate ID's to link all data around a patient. Yes, we keep a x-walk so that re-identification is possible with IRB approval.

It does not contain Cause of Death. It only contains date of death. It is only available to link records downstream.

Both provider identities and patient PHI are scrubbed.

Free for most users. If the user needs to create new datasets the cost will be ~$150.

Dates are shifted +-365 days. The date shift for each patient is consistent for all that patient's data.

The UCSF ARS (Academic Research Systems) team is working with ZSFG to work through the details. The current thinking is initially the ARS analysts will have access to ZSFG Epic to perform data requests. Down the road we may do some kind of integration.

We have a daily process that keeps the cross-walk updated. This is so that the different data sets can all be kept in sync (e.g. De-ID CDW, InfoCommons).

The new UCSF Bakar Computational Health Sciences Institute jacket

The Open Proposals Imaging Data proposal is going through the UCSF IT Governance process.

No

De-ID CDW will be a superset of RDB. Over time RDB will be phased out. Based on needs we may create a view and extracts from he De-ID CDW that closely match the RDB flat files. We are also evaluating tools to replace the RDB cohort explorer interface.

Yes, UCSF has deployed modules of the EPIC Research application. For specific studies please contact the AER.

The APeX Enabled Research unit and services are still in development. If there is an interest, we can investigate.

DD&T is a governance process. AER is for consultations and in some cases build/technical/integration/product services but we do coordinate closely. Many of the projects that receive AER consultations will also go through the governance process of DD&T.

Flowsheet data is in Clarity tables.

There may be groups that have that expertise; the ARS (Academic Research Systems) team's expertise is de-identification.

Currently SlicerDicer is not officially supported for research.

Currently SlicerDicer is not officially supported for research. SlicerDicer looks across all data when displaying counts. When drilling down to specific patient-level data you can only see your own patients.

No.

MyResearch, which gives free storage for up to 10 GB and it's 24 cents per GB per month above that.

Support for design of questionnaires for research studies can be requested at https://ucsf.service-now.com/ess/consulting_planning.do

The data files available in Information Commons are identical to the Research Data Browser Flat files. Information Commons provides a fast computing paradigm suitable for large files. In other words, your programs will run much-much faster on Information Commons than your standard issue computer.

All presentations are now available on data.ucsf.edu. Some are publicly available and others that may have sensitive information are behind the MyAccess Login. If you need to log in to view any files, please do not share or forward the files.

Information Commons does not currently support SAS. However, data can be extracted and made available for use on compliant environment where SAS is installed.

Info Commons sample code is available in the Info Commons Wiki. Additional code samples will be added as they become available.

Currently there are no size limits on user-uploaded data sets. Info Commons team will be monitoring usage of the platform in beta and may introduce recharges for large data set.

The data download process is available in Information Commons, the data can be transferred to a compliant environment that has SAS installed.

Currently, there is no plan to implement PatientExploreR on identified data.

Information Commons team will look into implementing auto-scaling as the system usage grows.

PatientExploreR points to de-identified RDB, which is the same as de-identified structured EHR data currently available.

Export of data from PatientExploreR is only allowed to a compliant environment. Please consult with UCSF Privacy if your local machine is compliant to store P3 level data. More information

See this TipSheet for assigning PROMIS questionnaires.

The frequency for PROMIS was on Track 3 Presentation 1 slide 9. You can run the MyChart Questionnaire Monitoring Report to see the frequency for other MyChart questionnaires (see slide 29).

There will be multiple reminders sent at this time.

Currently no mandates, but a couple of clinics have added it to their routine workflow. We are exploring opportunities to assign questionnaires automatically.

There is the ability to request custom questionnaires in APeX so if you'd prefer the PHQ9 that is a route. This is a clinical decision. PHQ9 is also available as a patient questionnaire.

No IRB approval necessary for clinical care. Research using PROs would still fall under IRB.

Research staff can run a report to see the results.

Yes, they could submit a ticket to do this now via Reporting Workbench.

Not everyone will have reporting workbench access, but the majority of users do have access at some level. There are many different security points and groups that determine what level of access they will have. Anyone can request access to reporting workbench. The reports that are displayed to them depend on which role and department they are in.

Not at this time.

Research staff can run a report to see the results.

Caboodle is updated nightly from Clarity. It does have other data sources that vary (e.g. death registry data is updated monthly). The de-id CDW is refreshed monthly from Caboodle.

We discussed a number of different things in the track so yes the AER is the best place to get started. Request AER consultation

Yes. You can request an account here.

Yes, there are ways to do this. If you have a project that's looking to do this, you’ll need to submit a request to the DD&T (Digital Diagnostics & Therapeutics) committee. If you have general inquiries, email them to [email protected].

We hosted EPIC for a FHIR deep dive in November and will be setting up ongoing sessions in the coming months. Subscribe the the #ucsf-fhir channel on UC Tech Slack.

Yes, it's possible, but depending on the app, there’s probably some integration work required. Email [email protected] if you have a specific use case in mind.

Subscribe the the #ucsf-fhir channel on UCTech Slack.

DD&T has the same priorities as UCSF Health.

No, you must be on a DD&T approved project. However, the HSPC sandbox is open to anyone.

It depends on the method used and requirements of the app. It could be any of these.

Yes they do. The base CDW data dictionary comes from Epic. UCSF then updates it to reflect the specifics at UCSF.

Data submitted to registries should be considered external data sharing and be treated as such. All approvals and security need to be in place and documented regardless of Registry entity. Additionally, data sharing agreement is critical as we are hearing of Registries commercializing and re-sharing data.

Data from external sources is not covered by the DTSF work. However, UCSF personnel need to follow all relevant policy, especially data/device security policies as they download any data or applications from external sources. All downloads should be from trusted, verified sources. We encourage working closely with IT security to minimize the very real risk of malicious cyber activity associated with such downloads.

The scientific principle applies here. The same verification methods that apply to any (data) model one wants to use should be used, no matter if the model parameters are learned from labeled or unlabeled data or derived using rules, common sense or geniality. This usually means to check the model for consistency with orthogonal knowledge, to check if the model explains existing data, and if it predicts additional data. One key characteristic is also simplicity, i.e. number of parameters needed to explain the data in relation to the complexity of the data. Recommended reading would be Bayesian and Shannon theories.

Still working on this answer--

Still working on this answer--

Still working on this answer--

Still working on this answer--

Still working on this answer--