Koios DS

K212616

Koios Medical, Inc. · cleared 2021-12-16 · product code POK · Radiology

Premarket evidence — what FDA accepted

Device typesamd
source quote (p.3)
“Koios DS is an artificial intelligence (AI)/machine learning (ML)-based computer-aided diagnosis (CADx) software device intended for use as an adjunct to diagnostic ultrasound examinations of lesions or nodules suspicious for breast or thyroid cancer.”
Algorithmartificial intelligence (AI)/machine learning (ML)-based computer-aided diagnosis (CADx) software device; two distinct AI/ML engines using computer vision and machine learning techniques; deep-learning derived cancer risk assessment
source quote (p.3)
“Koios DS is an artificial intelligence (AI)/machine learning (ML)-based computer-aided diagnosis (CADx) software device intended for use as an adjunct to diagnostic ultrasound examinations of lesions or nodules suspicious for breast or thyroid cancer. Koios DS contains two distinct AI/ML engines to characterize breast lesions and thyroid nodules. Each system uses computer vision and machine learning techniques embedded within an engine capable of reading, interpreting, analyzing, and generating findings from ultrasound data. The underlying breast and thyroid engines draw upon knowledge learned from a large database of known cases, tying image features to their eventual diagnosis, to form a predictive model.”
Adaptive (vs locked)No
PCCPFDA source did not state this
Cybersecurity addressedFDA source did not state this

Validation studies (4)

Retrospective clinical

n=650 cases · 2 site(s)

endpoints: AUC; Sensitivity; Specificity; Inter-Reader Variability; Impact on Interpretation Time

Retrospective clinical

n=900 patients

endpoints: AUC; Sensitivity; Specificity; inter-operator variability; intra-operator variability

Bench

n=900 cases

endpoints: AUC; Sensitivity; Specificity; Positive Predictive Value (PPV); Negative Predictive Value (NPV); Positive Likelihood Ratio (PLR); Negative Likelihood Ratio (NLR); sensitivity to shifts in ROI and transducer frequency

Bench

n=500 cases

endpoints: AUC; Sensitivity; Specificity

Reported performance (16 observations)

sensitivity0.97CI [0.96, 0.99]
source quote (p.15)
“System performance on the 900 cases reported an AUC of 92.9%, with a Sensitivity of 0.97 [0.96, 0.99] and a Specificity of 0.61 [0.57, 0.66].”
specificity0.61CI [0.57, 0.66]
source quote (p.15)
“System performance on the 900 cases reported an AUC of 92.9%, with a Sensitivity of 0.97 [0.96, 0.99] and a Specificity of 0.61 [0.57, 0.66].”
aurocas written: “auc”0.929
source quote (p.15)
“System performance on the 900 cases reported an AUC of 92.9%, with a Sensitivity of 0.97 [0.96, 0.99] and a Specificity of 0.61 [0.57, 0.66].”
aurocas written: “AUC (subject device)”0.929CI (0.913, 0.945 95% CI)
source quote (p.15)
“The subject device's updated breast classification engine was compared to the predicate device on the same 900 case validation set and demonstrated a statistically significant shift in AUC to 0.929 (0.913, 0.945 95% CI) from 0.882 (0.857, 0.907 95% CI).”
aurocas written: “AUC (additional test)”0.93CI [0.914, 0.946 95% CI]
source quote (p.15)
“This additional test generated a resulting AUC of 0.930 [0.914, 0.946 95% CI], demonstrating there is no degradation in performance attributable to dataset drift.”
aurocas written: “Thyroid Engine AUC”0.798
source quote (p.16)
“When applied to diagnoses made using ACR TI-RADS guidelines, the Al Adapter and descriptor predictors achieved an AUC of 79.8%, demonstrating a significant increase over the average physician AUC.”
sensitivityas written: “Thyroid Engine Sensitivity (biopsy)”0.644CI [0.545, 0.744]
source quote (p.16)
“When recommending biopsy, the system's sensitivity is 0.644 [0.545, 0.744] and specificity is 0.612 [0.566, 0.658].”
specificityas written: “Thyroid Engine Specificity (biopsy)”0.612CI [0.566, 0.658]
source quote (p.16)
“When recommending biopsy, the system's sensitivity is 0.644 [0.545, 0.744] and specificity is 0.612 [0.566, 0.658].”
sensitivityas written: “Thyroid Engine Sensitivity (follow-up)”0.879CI [0.812, 0.946]
source quote (p.16)
“When recommending follow-up, the system's sensitivity and specificity are 0.879 [0.812, 0.946] and 0.495 [0.446, 0.544], respectively.”
specificityas written: “Thyroid Engine Specificity (follow-up)”0.495CI [0.446, 0.544]
source quote (p.16)
“When recommending follow-up, the system's sensitivity and specificity are 0.879 [0.812, 0.946] and 0.495 [0.446, 0.544], respectively.”
aurocas written: “Thyroid Clinical AUC improvement (overall)”0.083CI (0.066, 0.099 95% CI)
source quote (p.15)
“The results of the subject device's clinical study evaluating its impact on the diagnostic performance of thyroid lesion classification successfully met all primary endpoints demonstrating a 0.083 (0.066, 0.099 95% CI) improvement in parametric AUC on the overall dataset along with a stratified analysis of United States (US)-based readers on US-based cases demonstrating an improvement of 0.074 (0.051, 0.098 95% CI) in parametric AUC.”
aurocas written: “Thyroid Clinical AUC improvement (US-based readers)”0.074CI (0.051, 0.098 95% CI)
source quote (p.15)
“The results of the subject device's clinical study evaluating its impact on the diagnostic performance of thyroid lesion classification successfully met all primary endpoints demonstrating a 0.083 (0.066, 0.099 95% CI) improvement in parametric AUC on the overall dataset along with a stratified analysis of United States (US)-based readers on US-based cases demonstrating an improvement of 0.074 (0.051, 0.098 95% CI) in parametric AUC.”
aurocas written: “Change in average AUC with Koios DS (US readers, US data) (parametric)”0.074CI [0.051, 0.098]
source quote (p.26)
“+0.074 [0.051, 0.098] (parametric)”
sensitivityas written: “Change in average Sensitivity of FNA with Koios DS (US readers, US data)”0.058CI [0.017, 0.098]
source quote (p.26)
“+ 0.058 [0.017, 0.098] (sensitivity)”
specificityas written: “Change in average Specificity of FNA with Koios DS (US readers, US data)”0.13CI [0.110, 0.151]
source quote (p.26)
“+ 0.130 [0.110, 0.151] (specificity)”
agreement_kappaas written: “Inter-Reader Variability (Relative Change %) (US readers, US data)”37.4
source quote (p.27)
“37.4% (US readers, US data)”

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

0
recalls in product code, 24mo
0
MAUDE reports in code, 12mo
—
vs code's own 3-yr baseline
1
drift signals on this device
  • re_clearance

    The FDA AI/ML device list shows a newer 510(k) K242130 (decision 2024-11-15) from Koios Medical, Inc. for a matching device line ("Koios DS") — a new clearance for the same line is a change event.

    first seen 2026-07-08 · k_number:K242130

Recall and MAUDE counts are product-code-level (reports aren't reliably attributable to one device). 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).

RIGOR™ Precedent · public FDA/CMS data · descriptive decision-support, not regulatory or reimbursement advice. Share this page: radar.healthai.com/precedent/device/K212616