Clinical Approval of Software IDE Decisions - 2026-08-27 and 2026-08-31¶
| Field | Value |
|---|---|
| Evidence type | Controlled correspondence decision record |
| Respondent | Camille Powe, MD |
| Initial response date/time | 2026-08-27 10:30 AM EDT (14:30 UTC) |
| Follow-up response date/time | 2026-08-31 8:00 AM EDT (12:00 UTC) |
| Email subject | Re: A few final software decisions for the IDE package |
| Source retention | Original email retained in project correspondence outside the software repository |
| Product impact | Documentation and submission-scope disposition only; no software behavior change |
Approval Statement¶
The 2026-08-27 response approved the seven emailed software/IDE statements and clarified the official study-data and safety-report sources. The 2026-08-31 follow-up instructed the study team to make all eight software-consistency changes recommended in the 2026-08-28 review and return tracked protocol and consent updates for review. The follow-up therefore records clinical agreement with the change direction; it does not establish that the controlled protocol, consent, or labeling has already been revised or approved for submission.
Approved Decisions¶
-
Algorithm start input: The algorithm requires a Dexcom reading when it is first started. A fingerstick cannot replace that first reading.
-
CGM gaps: During a connection outage, the app does not synthesize missing historical Dexcom readings. It records them as unavailable and accounts for insulin delivered during the outage.
-
Clinical Settings access: Study staff enter Clinical Settings during initial setup. Later edits require a single-use clinical unlock code. This is an investigational study control, not a commercial hospital user-account system.
-
BionicScout scope: Scout supports monitoring and troubleshooting, but the phone continues dosing without it. Scout is not the official source of study outcomes or safety reporting.
-
Physical-use claim: The package will describe extensive automated testing and formative real-Pod use supported by retained records. It will not claim a separate formal hardware-validation study.
-
Human-factors claim: The package will describe engineering UI testing, formative real-device testing, and clinical review. It will state that no separate formal participant usability study was conducted.
-
Fingerstick protocol wording: The protocol should describe the app's distinct safeguarded fingerstick process rather than state that an entered fingerstick is treated as Dexcom CGM data.
The approved claim boundary means STP-HW-001 is not retained as a separate
formal hardware-validation claim. The approved description of app usability,
formative real-device testing, and clinical review likewise does not retain supplemental
STP-ALERT-001 as a separate operator-observed formal study claim. Existing
automated alert verification and formative real-Pod records remain supporting
evidence.
Approved Clinical-Document Change Directions¶
The 2026-08-31 follow-up approves implementation of these eight recommendations from the software consistency review:
- describe fingerstick BG as a distinct safeguarded input mode, including the related meal-availability and participant-disclosure language during CGM interruption;
- describe the Standard and Pregnancy permanent targets, Pregnancy Temporary Targets, automatic return, unchanged Pregnancy safety/backup basis, and the Standard-to-Pregnancy 90% meal-setting proposal with clinical review before save;
- align the protocol with the 50 through 500 lb inclusive app-entry range and the algorithm's 20 through 230 kg final safety range;
- replace Phase 3 "factory reset" wording with the Same-Participant Reset workflow and reserve New Participant Reset for phone reassignment;
- describe G7 replacement as finishing startup and providing readings rather than requiring routine calibration;
- use the approved official-source hierarchy below and retain the recovery ZIP and Scout as supporting sources;
- describe the evidence as risk-based software verification and engineering representative-use testing under simulated and real-device conditions, without claiming a separate formal hardware-validation or summative participant-usability study; and
- complete the software-labeling set with the designated build, approved IFU revision, study-phone label where applicable, and in-app investigational-use banner specimen.
Clarified Official Study Sources¶
| Study information | Official source |
|---|---|
| CGM outcome data | Dexcom Clarity |
| Insulin-delivery outcome data | BionicLoop app exported CSV file |
| Adverse events and other safety information | Electronic case-report forms entered by study staff |
Tara Bresnahan, RN separately proposed that the controlled study documents state that adverse events and other required safety events are entered directly into the EDC by study staff; source information may come from participant report or assessment and medical-record review; and the investigator determines causality and relatedness to the study device and protocol. This is retained as document-update input pending incorporation and approval in the controlled protocol, data plan, or safety-reporting procedure.
The protected device-local recovery ZIP remains the primary software session-reconstruction record for the covered interval. BionicScout remains supportive monitoring and corroborating evidence. These software-evidence roles do not replace the official clinical study sources above.
Remaining Administrative Action¶
The correspondence did not identify the person authorized to give final sponsor or submission acceptance. The study team must still designate that role and retain final controlled approvals. Revised protocol, consent, and labeling files must implement the approved directions and receive their applicable document-control and FDA/IRB dispositions.