2 of 90

DevTeams at 12 Sites

The University of Kansas Medical Center (KUMC)
Indiana / Regenstrief (IU)
University of Iowa Healthcare (UIOWA)
The Medical College of Wisconsin (MCW)
Marshfield Clinic (Wisconsin) (MCRF)
The University of Missouri (MU)
The University of Nebraska Medical Center (UNMC)
The University of Texas Health Sciences Center at San Antonio (UTHSCSA)
The University of Texas Southwestern Medical Center (UTSW)
The University of Utah (UTAH)
Allina Health
Intermountain Health Care

3 of 90

Tell us about you and your site

Names, Titles / Roles
Clinical Data Sources, Population
ETL Tools, Processes

To i2b2
To PCORNet (SAS datasets, SAS views)
Other? OMOP, REDCap, ...
Tools: RDBMS Version? SAS? Jenkins? Docker?
Hardware Sizing

Strengths of your approach

4 of 90

KUMC

People

PI: Russ Waitman, Director of Medical Informatics
Engineering Team Lead: Imran Hafeez
Software/Systems Engineers: Dan Connolly, Lav Patel, Daryl Budine
Investigators, Analysts, Honest Brokers: Sravani Chandaka, Maren Wennberg, Xing Song, Mei Liu
Project Management: Hillary Weedman, Steve Fennel, Brooklyn Winkel

5 of 90

HERON Architecture

6 of 90

KUMC: ordinary stuff

Bulk of data comes from Epic @ KU Hospital

Demographics, Diagnoses, Procedures, Lab results, Medications on ~500K patients

GE IDX clinic billing system

Phased out in favor of Epic resolute billing

NAACCR Tumor Registry: ~60K cases
Linux, python, R, Jenkins, mercurial, git
Oracle 12c on SuSE, REDHat Linux
SAS 9.x on Windows, Linux

Docker: experimenting

7 of 90

KUMC Strengths

Self-service: ~1000 queries/month by ~40ppl
REDCap for oversight workflow, data delivery
i2b2 is the data warehouse

Flowsheets (1B+ facts)�Microbiology, Alerts, …
UHC
Social Security Death Master File, NCDR, NTDS, …

60+ monthly releases going back to 2011

Share and Enjoy: kumc-bmi/heron on github

HardwareSizing: 2.2TB FusionIO solid state storage

8 of 90

KUMC terms�on Babel

9 of 90

KUMC: i2p-transform to CDM

Prototype: gpc-pcornet-cdm early 2015
Adapted as SCILHS/i2p-transform Dec 2016
kumc-bmi/i2p-transform

forked from SCILHS/i2p-transform Feb 2016

TODO: sync with ~70 upstream commits
TODO: migrate pre/post-processing kludges back into HERON ETL

multi-day run-time :-/

10 of 90

Intermountain Team

Principal Investigator: Kirk Knowlton, MD
Co-Investigator: Benjamin Horne, PhD
Stakeholder Engagement: Samuel Brown, MD
Project Manager: Gary Bishop
Patient Representative: Chris Benda
Informatics Support:

Peter Haug, MD
Susan Rea, PhD
Justin Mundt, MS
Bart Dodds

11 of 90

Intermountain Healthcare

Not for Profit, Integrated Healthcare Organization
Twenty-three Hospitals
A Medical Group with more than 1,600 Physicians� and Advanced Practice Clinicians
Approximately 180 Clinics
A Health Plans division (Select Health)
Cerner EHR
Internally Developed Electronic Data Warehouse

Our Mission: Helping people live the healthiest lives possible^®�

12 of 90

10/11/2018

Intermountain Patients by ZipCode 2006 – Present

13 of 90

EDW → AHR → CDM

Intermountain supports an Enterprise Data Warehouse (EDW) containing clinical and business data.
A cleansed and optimized extract called the Analytic Health Repository (AHR) was created for research beginning in 2009.
Our PCORnet CDM is largely an extract from the AHR.

14 of 90

EDW: A Source of Aggregated Data

Data Sources

Finance

Internal

State/Federal

External

Lab

Claims

Pharmacy

EMR

Data Access

EDW

Primary Care

Data Mart

Other

Data Marts

Data Mart

Case Mix

Data Mart

Claims

Data Mart

Analytic Processes

Direct SQL

Desktop ODBC tools

Web applications

Statistical Analysis

Reporting Tools

OLAP Tools

OTHERS

Data Mining

15 of 90

Using EDW Data

EDW

Primary Care

Data Mart

Other

Data Marts

Data Mart

Case Mix

Data Mart

Claims

Data Mart

Financial

Analysis

Mandated

Reporting

Quality

Management

Business

Planning

Clinical

Research

Planning

Research

Data Analysis

Genetic

Epidemiology

Etc.

Data Mining/

Machine Learning

Process

Tracking

16 of 90

Analytic Health Repository

L3 – Analytic

Clinical constructs (Registries, Scores, Quality measures)
Population-based views of disease progression, predictors & outcomes
Flexible, transparent and consistent
Gateway into Specialized Population-based Views

L2 – Building Blocks

Sources integrated and terminology standardized
Cleansed and Quality enforced
Exceptions reported
Complete data sets over time

L1 – Enterprise Data Warehouse

Contains source system data
Integrated using Data Bus Architecture
Common data attributes with standard names and data types

Enterprise ETL

Modeling Engines

Concepts based on models. These represent high level clinical or business categories that are implemented using a variety of knowledge-based tools (rule engines, ML, ontologies, etc.).

The base data extracted directly from original data sources. Individual tables incomplete, using original coding and structure, with errors unaddressed.

Data restructured and optimized for clinically relevant searching. Carefully screened for errors, converted to national and international coding standards, complete for relevant data over time.

17 of 90

Analytic Health Repository -> CDM Database

L3 – Analytic

Clinical constructs (Registries, Scores, Quality measures)
Population-based views of disease progression, predictors & outcomes
Flexible, transparent and consistent
Gateway into Specialized Population-based Views

L2 – Building Blocks

Sources integrated and terminology standardized
Cleansed and Quality enforced
Exceptions reported
Complete data sets over time

L1 – Enterprise Data Warehouse

Contains source system data
Integrated using Data Bus Architecture
Common data attributes with standard names and data types

Enterprise ETL

Modeling Engines

PCORnet CDM

Database

18 of 90

10/11/2018

Number of Patients

Demographic Table: 2,830,321

Care Prvdr. Visits, Jul 17-Jun18: 1,469,350

Intermountain Healthcare CDM Vital Statistics

Care Facilities in Utah, S. Idaho

Hospitals 23

Clinics 180

Diagnostic Services, Home Care, Rehab,

Skilled Nursing, Telehealth

Primary Children Medical Center

Number of Care Providers

Employed MD, PA, APRN, other Prof.: 2300

Affiliated practitioners: 3500

Programs of Interest

Select Health Plans claims ~25% patients

Tissue biorepository:

specimens > 30 yrs, ~5 million

Linkages to Utah Population Data Base

CV Family History data base

Active Inmtn Ca Registry

19 of 90

IU / Regenstrief

Team:

PI-...Tim Imler (Umberto T.)

Engineering Team-Tony French, Jeff Stroup

Data Analyst-Ross Hayden

Project Manager-Dan Hood

20 of 90

IU/Regenstrief

Data Sources

-Data fed from Indiana Network for Patient Care (INPC, Indiana Health Information Exchange). Specifically, IU Health and Eskenazi Hospital systems.

-Certain GPC/PCORI deliverables have required data pulls directly from IU Health database

-NAACCR Tumor Registry

-Supplemental Px’s and Dx’s

21 of 90

IU/Regenstrief

ETL Tools and Processes

-Oracle Views for De-identification

-Oracle chains, stored procedures and views for ETL from RMRS → I2B2

-SCILHS i2p-transform for i2b2 → PCORNET CDM V3

-SAS Datasets for PCORNet (Failed attempt at using Oracle Views)

Tools:

Oracle 11g, SAS 9.4, RHEL 6/7, Java, Atlassian Tool Suite, Git,

Docker (investigative), Ansible (investigative)

22 of 90

IU/Regenstrief

23 of 90

GPC partners

Leadership: Department of Epidemiology, College of Public Health
Clinical: University of Iowa Health Care
Informatics: Institute for Clinical and Translational Science (ICTS)

24 of 90

People

Betsy Chrischilles Principal Investigator
Ryan Carnahan Investigator
Mary Schroeder Investigator
Michael Wright Dev Lead
Gi-Yung Ryu Data Architect
Lucas Van Tol Technical Support
Brad McDowell Research Specialist
Brian Gryzlak Project Manager
Boyd Knosp GPC DROC Representative
Ashlee Wilson GPC DROC Representative (alt)

25 of 90

UIOWA strengths

Clinical Research Data Warehouse (CRDW) provides a normalized foundation to:

build other architectures (e.g. i2b2, CDM);
increase efficiencies in fulfilling data requests (e.g. TriNetX, PCORnet); and
feed custom datamarts (e.g. ORIEN, mother-child datamart, cancer datamart)

Prospective data collection

REDCap use rate is high, including MyCap
TeleForm for creating machine-readable forms and databases to contain the data
Iowa Personal Health Record for online consenting, collecting patient-reported data and research management. Semi-

automated interface with CRDW, REDCap, TeleForm,

statewide cancer registry

26 of 90

Data flow supporting GPC/PCORnet requests

27 of 90

Data environment

MS SQL Server shop
Clinical Research Data Warehouse (CRDW) DB Server

Serves as i2b2 staging
Hardware: Dell R720xd
Processor: 2 x 6 Xeon E5-2640 (12 cores total, 24 with hyperthreading)
Memory: 32GB RAM
Storage: ~20TB (mix of RAID1/5/6)

i2b2 Production DB Server

Same as CRDW but ~5TB of storage

28 of 90

Marshfield - People

PI: Jeff VanWormer, PhD

Robert Greenlee, PhD (Co-I)

PM: Judith Hase
Technical leads: Steff Roush, Lynda Kubacki-Meyer
Programmers

Steff Roush - i2b2 ETL, data mapping, QA, SAS
Eric LaRose - i2b2 ETL, CDM transform, SQL
Erica Scotty - PCORnet queries, SAS

29 of 90

Marshfield - Data Sources, etc.

Internally developed EHR - 20+ years capture;

Also: insurance plan, hospital (Cerner, Soarian), dental clinics, registries

Patient population: ~2.5M historic; ~400k unique/year

Central and Northern Wisconsin

Research data delivered as “expert mediated” model

12 Research Programmer/Analysts
Netezza Data Warehouse, SAS 9.4

Software development stack is Microsoft

Windows, MS SQL server, C#, Azure DevOps

30 of 90

Marshfield - ETL Tools, Processes

Data Warehouse → Combined Tables

Process to de-duplicate sources
Netezza, SAS
Timing: Weekly

Combined Tables + supplemental data → i2b2

Netezza, SAS, C#; mapping IDs, formatting
Timing: Every 1-2 months, or as needed

i2b2 [SCILHS i2p-transform] → PCORnet CDM

MS SQL; mapping variables, formatting
Timing: After i2b2 build, based on Data Characterization schedule
*Limited Datasets* - date-shift on demand model
Considering moving away from SCILHS for CDM builds

31 of 90

Marshfield - i2b2 SQL Server Specs

Software: SQL Server 2012 SP1 (v11.0.3000.0)

RAM: 8.0 GB (Max SQL Server memory: 6.144 MB)

CPU: 4 Cores (Intel Xeon CPU E5-2690 v3 @ 2.60 GHz)

Disk:

C: (50 GB) – OS Only
D: (1.5 TB) – Data drive 1
E: (60 GB) – Log file location
F: (800 GB) – Data drive 2
G: (700 GB) – Data drive 3
H: (1 TB) – Data drive 4
I: (300 GB) - Data drive (tempDB only)
J: (700 GB) - Data drive 5

Notes:

We run two instances of i2b2 (live and down), which is why we have multiple data drives.
Extra space is needed for data loads and building indexes

Without a load in progress, free disk space: 39%, 100%, 13%, 8%, 12%, 100%, 16% (drives D-J).
At times during a load, the log files need to be extended over to some of the data drives

32 of 90

UTHSCSA

The Team

Dr. Meredith Zozus (NEW -- YAY!!!) - Division Chief - PI
Alex Bokov - Faculty
Olivia Ellsmore (Suarez) - IRB Liaison and Project Manager
Laura Manuel - Software Engineer- Senior- Lead Programmer
Eric Moffett- App Sys Prog Sys Analyst - Associate - Programmer
*Coming soon* - Honest Broker

Informatics Leads for Family Weight and Health Survey (Obesity)

33 of 90

UTHSCSA

Clinical Data Sources

UT-Med Clinic Data
UHS Hospital/Clinic Data

ETL Tools, Processes

Inhouse code Integrate UHS Sunrise (Allscripts) and UTMed- Clarity to I2B2.

Coming soon. UHS Clarity

Inhouse code -> Clarity/Sunrise to CDM

34 of 90

UTHSCSA Server Specs

Replacing SQL Server and CRC soon.

Further upgrades to New_stage also in the mix.

<3 new_stage

35 of 90

University of Nebraska Medical Center

Clinical partner, Nebraska Medicine.

-Two hospitals and multiple clinics

-Level I trauma center

-Newly opened Buffet Cancer Center.

37 of 90

UNMC People

PI: James McClay, MD
Project Manager: Carol Geary, PhD
Informatics Standards: James Campbell, Scott Campbell
ETL programming: Yeshwanth Narayana, Jay Pedersen
Architecture & System Design: Ashok Mudgapalli, Research Information Technology Office (RITO) Director
RITO Technical Staff

39 of 90

UNMC Architecture

Epic -> Epic Clarity-> Research copy of Epic Clarity (research clarity).
Research Clarity is MS SQL
Utilize Oracle Gateway to interface between MS SQL and Oracle
I2b2 instances: Linux, Oracle 11g
Postgis on Postgres

40 of 90

UNMC Strengths

Excellent relations with Clinical Partner
IDeA Clinical Translational Research Site
Vocabulary Standards
Encoding anatomic pathology and microbiology into i2b2

41 of 90

Developing ONC Standards Metadata for i2b2 Interoperability

Collaborating with NLM to develop historically complete metadata for RXNORM/NDC
Coordinating extension to SCILHS metadata (labs,meds,diagnoses) with Partners to include GPC partner labs and meds
Collaborating with Ontologies workgroup Transmart i2b2 foundation

42 of 90

Extending LOINC and SNOMED CT for Precision Medicine

Developing description-logic supported ontology for lab/pathology/clinical observation results (LOINC and SNOMED CT observables)
Developing results reporting for molecular pathology and genomics
Developing HL7 V2 interface from pathology/sequencing labs to Epic for real-time cancer data

43 of 90

UTSW People

PI: Lindsay Cowell, Associate Professor, Division of Informatics

Technical Team: Phillip Reeder, Jennifer Cai

Honest Brokers: Teresa Bosler, Shiby Antony

Project Management: Shiby Antony

44 of 90

UTSW Architecture

45 of 90

UTSW Data Sources, Tools

Data Sources

Bulk of data comes from UTSW EPIC (Clarity)

Demographics, Diagnoses, Procedures, Lab results, Medications ~1.3B observations on ~5.5M patients

Billing System (Siemens, Epic Resolute, IDX [data prior to 2009])
Volunteer Patient Registry
NAACCR Tumor Registry: ~60K cases

Tools

Linux
Oracle 11g
SAS 9.4 on Windows

46 of 90

UTSW i2b2 Terminology

47 of 90

UTSW Strengths

UT Southwestern Clinical Research Data Warehouse: access to standardized research data in addition to the i2b2: Bio-specimen data, Volunteer Patient Registry data, and Clinical Research Study data.
REDCap for data requests, delivery.
Import of clinical genomic data into i2b2.
Cancer gene connect - addition of sequencing data to i2b2 using Sequence Ontology.
REDCap FHIR integration with EPIC.

48 of 90

University of Missouri (MU)

49 of 90

MU Team

Abu Mosa, PhD

Primary Investigator

Honest Broker

Vasanthi Mandhadi

Project Manager

Honest Broker

Informatics Team

Todd McNeeley

Informatics Team

Marshall Gorski

Informatics Team

Cory Gassner

Informatics Team

Kamruz Zaman Rana

Informatics Team

Noelle Al-Khashti

Informatics Team

Jeff Ordway

Patient Advisor

Lynne Lawrence

Patient Engagement Officer

Lori Wilcox

IRB Representative

Jenelle Greaning

IRB Representative

William Stephens

Patient Advisor

50 of 90

MU Data Sources

Cerner Millennium EMR
IDX Billing
Cerner PathNet
Cerner PharmNet
Cancer Registry - Ellis Fischel Cancer Center = ~11K
SSDMF
ACS
Total # of Patients = ~635,245

51 of 90

MU i2b2/CDM ETL

53 of 90

MU Tools

SAS 9.4
MS SQL
Pentaho Data Integration (ETL)

Java
ECMAScript

SSIS (ETL)
ActiveBatch - cross platform job scheduling
REDCap - approved for storing identified data

54 of 90

MU Hardware

KC (Cerner)

Oracle on Linux

24 core
192 Gb RAM

SAS on Windows Server (MU SOM)

12 core
64 Gb RAM

SAS on Linux Server (MU research computing)

HPC3
4 nodes
96 cores

Sql Server on Windows (Tiger Institute)

16 core
256 Gb RAM
8Tb storage

55 of 90

MU Strengths

Large established user-base leveraging i2b2 for research
MU-iCATS - partner of WashU CTSA
A newly established Center for Biomedical Informatics (CBMI)
MU Center for Patient Center Outcomes Research (AHRQ Funded)
Recurring workshops on the use of informatics tools and applications
Centralized and integrated data governance for research data requests and research data brokers
Strategic support for i2b2 at the EMR vendor (Cerner)
Tiger Institute for Health Innovation - joint MU/Cerner partnership including a research focus
Piloting de-identification and feature extraction tasks using NLP pipeline on pathology and radiology reports
Collaboration with MU EECS department for automation of research data request and governance process using AI Chatbot and blockchain

56 of 90

MU Areas to Improve

Leverage GPC for funding opportunities
Free-text notes not available in i2b2
Resolving investigative checks in EDC

57 of 90

Medical College of Wisconsin

58 of 90

Medical College of Wisconsin

Main academic campus & practice in Milwaukee
2 satellite academic campuses in northern Wisconsin
Provides professional services to Froedtert Health System and Children’s Hospital of Wisconsin
Staffs 5 hospitals and dozens of clinics statewide
Provides primary, tertiary and Level 1 trauma care

59 of 90

MCW Dev Team

PI: Brad Taylor, CRIO
PM/Honest Broker: Kris Osinski, Business Analyst
Software Engineers

Alex Stoddard - ETL, CDM lead
Andrew Vallejos & George Kowalski - i2b2, ACT, PopMedNet, SHRINE, Notes De-id pipeline leads
Weihong Jin - ETL/Honest Broker

60 of 90

MCW Data Sources/Population

Epic EHR

Froedtert (2b facts on ~1.2m patients)
Children’s (200m facts on ~600k patients)

GE/IDX physician billing (1999+), converted to Epic 12/1/2018
Froedtert Health System hospital billing (Epic 2012+, Affinity legacy)
NAACCR tumor registry
MCW Tissue Bank biospecimens (OnCore)
Foundation Medicine genetics data
*New* Mosaiq radiation treatment dosimetry data elements

61 of 90

MCW ETL/DW Architecture

Notes:

i2b2 & CDM are built from same source data now

Converting from Oracle db to PostgreSQL 11 in 2018

Added discrete data from Foundation Medicine genetic testing result reports

62 of 90

MCW Strengths/Opportunities

Strengths

Well-curated CDM
TriNetX Query Tool/Trial Connect/Research Network
Unique self-service Honest Broker data extract process
Expanded i2b2 to adopt ACT/SHRINE ontology/network
Participating in All of Us Program with OMOP data model

Roadblocks/Opportunities for Improvement

Getting access to enriching 3rd party/legacy data is difficult
Adding notes & flowsheet data back into i2b2 - high de-id volume
Improving visibility across CTSA research community
Expanding our exposure in the CTSA RIC/TIC/TIN

63 of 90

Univ of Utah

Colin Moynier - REDCap
Ainsley Huffman - project manager
Reid Holbrook - IT
Molly Conroy
Rachel Hess - PI

64 of 90

•2 Hospitals

•12 Community Clinics

•5+ Specialty Centers

•1 Electronic Health Record System (+3 historical)

•200+ Ancillary Systems

•1 Enterprise Data Warehouse

66 of 90

•Current CDM Status

67 of 90

Appendix: Former Sites

68 of 90

U. C. Davis (guests)

Bill Riedl
SHRINE

Config mgmt code - source

69 of 90

Children’s Mercy Hospital (CMH)

Team

Mark Hoffman - PhD, PI
Warren Teachout - Director of Decision Support, Honest Broker, PM
Rita Fothergill - Software Architect, Programmer
Sierra Martin - Programmer (Python, RedCap)
Cerner Team: Claire Maples, Aaron Meyer, Paul Albright, Gary Gasperino, Tina McKaig

Clinical Data Sources, Population

Cerner Millennium Applications - clinical data, registration, pharmacy, lab, radiology

Total # of patients: ~500,000

GE Billing system/Meditech historical billing (from 2008)
NAACCR Tumor registry (from 1995)

Total # of patients: ~3500 patients

70 of 90

CMH Tools, Processes

Processes

Cerner Millennium -> HealthFacts -> i2b2

Refresh every 2 weeks

i2b2 [SCILHS i2p-transform] -> PCORnet CDM

Tools

Oracle 11g
SAS 9.4
RedCap
R, Python, SQL
Windows, C#, TFS, Git
MS SQL server

71 of 90

CMH - Health Facts Workflow

72 of 90

CMH Strengths and Areas to Improve

Strengths

Non-human subjects designation for de-identified node of i2b2 (means researchers can carry out retrospective data studies without additional IRB approval)
Internal DROC in place to quickly expedite GPC or PCORNet de-identified data requests
Cerner supports and enhances i2b2 focusing on PCORNet and GPC requirements

Areas to Improve

Availability of orders, especially medications
Ensuring drug mappings from EMR to i2b2 and the CDM are complete
Ability to easily incorporate registries

73 of 90

UMN

PIs:

Genevieve Melton-Meaux, MD, PhD, FACS, FASCRS, FACMI,

Constantin F. Aliferis MD, PhD, FACMI

Analysts:

Ahmad Abusalah, MSc, PhD (Director of Clinical Informatics)

Gretchen Sieger (Lead Analyst)

Sonya Grillo (Clinical Data Expert)

Kathleen McKay, PhD (PM and HB)

Project Managers:

Kathleen McKay, PhD

Programmers:

Tim Meyer

Luke Bicknese

Duy Duong

Andrew Hangsleben

74 of 90

UMN

Tools:

Oracle (Clinical Data Warehouse)

SQL Server (i2b2 & PCORnet CDM)

Pentaho Data-Integration (Kettle)

Jenkins Continuous Integration

i2b2 & SHRINE

Primary Data Sources:

EMR Data from Fairview Health Services

Claims Data from University of Minnesota Physicians

Tumor Registry from 4 hospitals

Death Records (MN)

Enrichment:

Data Modeling

Terminology Management

Code Mapping

Geocoding

Patient Matching

Ontology

Free-Text Notes Indexing

75 of 90

UMN

Clinical Data Management and Integration System

Metadata Management

Schema Management

Table Management

Row Counts & Load Information

Columns

Stats & Profiling

Relationships

Feeds Management

Define Destination, Source Query, Type of Load

Data Request Management

Patient Sets

Queries

Users (Integration with Active Directory)

De-Identification

76 of 90

WISC

People

PI’s

Mark Drezner
Umberto Tachinardi

Other Useful folks

Sarah Esmond
Laura Ladick

Geeks

Tom Mish
Don Steger
Debbie Yoshihara
Yiqiang Song

79 of 90

HackathonFour Dev Recap

Goals:

SNOW SHRINE

Toward interactive federated query

Unstructured Notes De-identification

From multiple sites

GROUSE

CMS claims data integration
Aka “RESDAC”

Roadmap, plans, peer-to-peer learning...

80 of 90

Shrine flow

George logs into shrine web client (tomcat) blueberry. (in DMZ) gets a session-id X123

Blueberry authenticates, authorizes (qualified investigator, trained, …) using PM cell in i2b2.mcw.edu (internal). George is on project P67

George issues query (in shrine ontology) from his web browser to blueberry using X123 (submit)
Blueberry forwards to hub (snow-hub.wisc.edu)

Certs, IP addresses have been exchanged between blueberry and snow-hub

Hub fwds to adapter at each site, e.g. KUMC
KUMC adapter (blobfish in DMZ) maps paths to KUMC ontology
Blobfish fwds to shrine-hive.kumc.edu (internal) (IU has an alternative: different usernames; UMN uses different projects)
Shrine-hive computes counts as usual; returns to blobfish
Blobfish returns counts to snow-hub… blueberry… George’s browser

81 of 90

SHRINE status poll

Moved to meeting notes

82 of 90

SNOW SHRINE

Keith / WISC gave a presentation

Hacked with Lav @KUMC

83 of 90