Jazz

K223659

AI Medical AG · cleared 2023-09-22 · product code LLZ · Radiology

Premarket evidence — what FDA accepted

Device typesamd
source quote (p.5)
“Jazz is a Software as a Medical Device (SaMD) consisting of a software intended to be a facilitating tool for the physicians, in the sense of a semi-automatic pipeline for the process of identifying, labeling and quantifying the volume of segmentable brain structures identified on MR images.”
Algorithmmachine learning algorithms
source quote (p.7)
“This device uses machine learning algorithms as part of its key functionality. Networks were trained using brain images, which were fully segregated from the test set, and using a ground truth which was set using gold standard human expert opinion.”
Adaptive (vs locked)No
source quote (p.7)
“Networks were trained using brain images, which were fully segregated from the test set, and using a ground truth which was set using gold standard human expert opinion. “Semi-automatic" refers to the possibility given to the physician to correct the segmentation of the software before saving.”
PCCPNo
Cybersecurity addressedFDA source did not state this

Validation studies (1)

Retrospective clinical

n=344 patients

endpoints: voxel-wise sensitivity; voxel-wise specificity; lesion-wise dice score; lesion-wise true positive rate; lesion-wise false negative rate; lesion-wise false discovery rate; anatomy localization score; coregistration score; identical number of lesions; total lesions volume; report generated in a process-reprocess experiment

Reported performance (4 observations)

sensitivity0.4CI at least
source quote (p.7)
“In one accuracy experiment, the acceptance criteria for device performance were set to be a voxel-wise sensitivity of at least 40%, a voxel-wise specificity of at least 95%, and a lesion-wise dice score of at least 0.5 for both the multiple sclerosis high sensitivity and high specificity models, as well as a lesion-wise true positive rate of at least 60% and a lesion-wise false negative rate of at most 40% for the high sensitivity model, as well as a lesion-wise false discovery rate of at most 50% for the high specificity model.”
specificity0.95CI at least
source quote (p.7)
“In one accuracy experiment, the acceptance criteria for device performance were set to be a voxel-wise sensitivity of at least 40%, a voxel-wise specificity of at least 95%, and a lesion-wise dice score of at least 0.5 for both the multiple sclerosis high sensitivity and high specificity models, as well as a lesion-wise true positive rate of at least 60% and a lesion-wise false negative rate of at most 40% for the high sensitivity model, as well as a lesion-wise false discovery rate of at most 50% for the high specificity model.”
diceas written: “lesion-wise dice score”0.5CI at least
source quote (p.7)
“In one accuracy experiment, the acceptance criteria for device performance were set to be a voxel-wise sensitivity of at least 40%, a voxel-wise specificity of at least 95%, and a lesion-wise dice score of at least 0.5 for both the multiple sclerosis high sensitivity and high specificity models, as well as a lesion-wise true positive rate of at least 60% and a lesion-wise false negative rate of at most 40% for the high sensitivity model, as well as a lesion-wise false discovery rate of at most 50% for the high specificity model.”
sensitivityas written: “lesion-wise true positive rate”0.6CI at least
source quote (p.7)
“In one accuracy experiment, the acceptance criteria for device performance were set to be a voxel-wise sensitivity of at least 40%, a voxel-wise specificity of at least 95%, and a lesion-wise dice score of at least 0.5 for both the multiple sclerosis high sensitivity and high specificity models, as well as a lesion-wise true positive rate of at least 60% and a lesion-wise false negative rate of at most 40% for the high sensitivity model, as well as a lesion-wise false discovery rate of at most 50% for the high specificity model.”

Each value carries its own analysis unit and task — never compare or pool across devices. Source: 510(k) summary PDF.

Predicate network

Postmarket — what happened after clearance

48
recalls in product code, 24mo
295
MAUDE reports in code, 12mo
+683%
vs code's own 3-yr baseline
26
drift signals on this device
  • adverse_event_inflection

    MAUDE adverse-event reports for product code LLZ: 295 in the 12 months ending 2026-06, vs a 37.7/12mo average over the prior 3 windows (+683%). Code-level count — reports are not attributed to this specific device.

    first seen 2026-07-08 · openFDA /device/event.json count=date_received product_code=LLZ

  • recall_reason_pattern

    Software/algorithm-related recall in product code LLZ (GE Medical Systems SCS, initiated 2026-05-08): "GE HealthCare has become aware of a context synchronization issue in AW Server 3.2 ext. 6.5. When a user selects a patient or exam in the AW Server Web Client worklist and launches" Recalling firm is another firm in the same product code.

    first seen 2026-07-08 · recall res_event_number:99042

  • recall_reason_pattern

    Software/algorithm-related recall in product code LLZ (PHILIPS MEDICAL SYSTEMS NEDERLAND B.V. Veenpluis 4-6 Best Netherlands, initiated 2026-03-05): "Potential that mis-ordered frames in Vue Motion during dynamic cine runs may cause images frames to display out of sequence." Recalling firm is another firm in the same product code.

    first seen 2026-07-08 · recall res_event_number:98520

  • recall_reason_pattern

    Software/algorithm-related recall in product code LLZ (GE Medical Systems, LLC, initiated 2026-01-30): "There is a potential cybersecurity vulnerability affecting certain versions of Centricity Universal Viewer. User login credentials may be exposed on the local client workstation, w" Recalling firm is another firm in the same product code.

    first seen 2026-07-08 · recall res_event_number:98428

  • recall_reason_pattern

    Software/algorithm-related recall in product code LLZ (Focalyx Technologies, LLC., initiated 2025-12-23): "Software device that is an accessory for image-guided interventional and diagnostic procedures involving prostate has accessories that may not function properly with Windows 10, wh" Recalling firm is another firm in the same product code.

    first seen 2026-07-08 · recall res_event_number:98137

  • recall_reason_pattern

    Software/algorithm-related recall in product code LLZ (DICOM Grid, Inc., initiated 2025-11-18): "Software intended to aid in diagnosing conditions, planning treatments, visualizing anatomical structures has a bug that, if all of following are met: Viewing images in InteleShare" Recalling firm is another firm in the same product code.

    first seen 2026-07-08 · recall res_event_number:97993

  • …and 20 more.

Recall and MAUDE counts are product-code-level (reports aren't reliably attributable to one device); a recall is shown as device-attributed only when the recall record itself lists this clearance number. Signals are descriptive observables with sources — never a judgment that the device is unsafe or drifting. Snapshot 2026-07-08.

Reimbursement — how devices like this got paid

Not yet tracked — no payment pathway indexed for this clearance (the reimbursement corpus is a growing seed set).

Applicable FDA guidance — what the submission is measured against

FDA guidance documents and guiding principles applicable to 510(k) AI/ML devices in the Radiology panel. A curated reference index, not legal or regulatory advice — each item states its own status, and a draft is never binding.

Applicability is derived from the device's FDA advisory panel and pathway — cross-cutting guidances apply to every AI/ML device; panel-specific ones are flagged. Titles, dates, and links verified against fda.gov as of July 2026.

Constat Precedent · public FDA/CMS data · descriptive decision-support, not regulatory or reimbursement advice. Share this page: constat.dev/precedent/device/K223659