1. Impact on break through
2. NC by Andrea
2.1. NC 1
2.1.1. Non-Conformity 1: SHA (Symphony Health Analytics) Data Universe dBase is not built under ISO13485 QMS and the supplier is not a qualified Medical Device supplier as per ISO13485/QSR820. No access / visibility to key quality related aspects of the provided dBase (like data cleaning, data collection, data reconciliation) => Not compliant to current regulatory requirements.
2.1.1.1. Concern
2.1.1.1.1. Data Quality
2.1.1.1.2. Data integrity
2.1.1.2. Mitigation
2.1.1.2.1. Vendor qualification process
2.1.1.2.2. UCB responsibilities
2.1.1.2.3. To do
2.2. NC 2
2.2.1. Current SHA Data Universe dBase is missing precise patient identification and proximity patient selection => Not compliant to current regulatory requirements.
2.2.1.1. Concerns
2.2.1.1.1. Match accuracy
2.2.1.1.2. Proximity selection
3. What ifs?
3.1. Able to qualify as Vendor
3.1.1. Both NCs will be resolved
3.2. SHA unable to qualify as Vendor?
3.2.1. Stop project
3.2.2. Pivot without compromising patient value
3.2.2.1. Technology options
3.2.2.1.1. Treat SHA as training dataset, clinical validation against EMR data
4. Items to close (Followup of 2021/10/15 disc)
4.1. 1. Targeted Clinical condition: 3 or more ASMs in 3 years
4.1.1. What is the condition
4.1.1.1. What is the ground truth?
4.1.1.2. How do you measure/establish ground truth?
4.1.1.2.1. What is the reference standard?
4.1.2. Closure: What is the evidence package needed to substantiate the claims in our intended use?
4.2. Is Open claims a sufficient data source to measure 3+ over 3 years
4.2.1. Is claims sufficient to establish clinical association (clinical condition)
4.2.2. Is claims data sufficient to conduct clinical validation (retrospective clinical trial)
4.3. Backup study
4.3.1. What are we trying to accomplish
4.3.2. Why are we emphasizing EMR?
4.4. Clarify the working of device to H&L and Carlos
4.5. Emphasize FDAs position of not needing generalizability study for our case (#8)
4.5.1. Is this also applicable for Clinical Validation?
4.6. FDA did not explicitly reject clinical validation using Real world data
4.6.1. Insufficient information
4.6.1.1. Definition of clinical condition (1.a, 1.d)
4.6.1.1.1. We revised Intended use
4.6.1.2. Confounding factors and their impact on diagnosis of clinical condition (1.b, 2.iii)
4.6.1.2.1. What is the definition of the new clinical condition, and are there confounding factors associated with it?
4.6.1.3. Measuring clinical outcome (ground truth determination, 2.vi)
4.6.2. Algorithm design
4.6.2.1. Accepted all proposals
4.6.3. Closure: