Ann: Voluntary Suspension, page-328

Hospital records and trial records are kept separate, so the "patient file" for the purposes of healing is independent of the "trial subject data".

The process will go broadly like this:

1. Hospital patient agrees to enrol in trial.
2. Subject is given a unique subject ID and the hospital patient's ID is recorded against the subject ID.
3. Subject's cough sound and symptoms are recorded against the subject ID.  Chances are that ResappDx will collect this data because it has the front-line quality assessment features.  But the diagnosis will not be given to the operator or the subject.
4. Resapp is given a copy of the data for QA purposes.
5. Subject is assessed as a patient by the hospital and appropriate tests booked for the suspected condition.
6. Patient is treated by the hospital, hopefully cured and sent home.
7. At some point the patient's test results are provided to the trial's clinical adjudication team who make a decision about what the subject's condition was and recorded it against the subject ID.  This data is kept secure and secret.
8. At some point the subject will be diagnosed by ResappDx and the diagnosis saved against the subject ID.  The diagnosis may be made in step 3 above, but saved elsewhere and not disclosed to the operator or the subject.  The diagnosis may be made later by a test team dedicated to that operation, possibly even after all the hospital test results are adjudicated and Resapp shares are already in trading halt.  This test team will absolutely not be Resapp.  The result is a set of subject IDs each with a diagnosis.
9. Nobody who has access to the data from step 4 is allowed to have access to the data from step 8.  The two sets are kept (normally) in different locations and maintained by entirely separate teams.
10. Once ALL test results from step 4 and step 8 are in hand the researchers tell Resapp that they are ready to start matching the two sets of data.  This is when the trading halt begins.  At this point nobody has compared results for even a single patient.
11. Once trading is halted the process of matching ResappDx diagnoses to the adjudication team's decision for each subject.  Accuracy in matching is critical.  Each match will result in one of three outcomes: A) ResappDx and adjudicators agree that subject had the same condition (positive agreement), B) ResappDx and adjudicators agree that subject did not have a given condition (negative agreement), C) ResappDx and adjudicators disagree.
12. Once all the matches are made the total agreements (positive and negative) and disagreements are tallied for each condition.  Percentage agreements are calculated (positive and negative) and 95% confidence intervals are calculated using standard statistical methods.
13. All counts, totals, percentages are intervals are critically evaluated, validated, verified, cross-checked and confirmed.
14. An announcement is crafted passing judgement on the results, and possibly outlining near term next steps.
15. Announcement is released and 30 minutes later (or at open) we're all back in action... stampeding in... or out... as warranted.