Search Results
Found 5 results
510(k) Data Aggregation
(291 days)
The Chloride test, as part of the epoc Blood Analysis System, is intended for use by trained medical professionals as an in vitro diagnostic device for the quantitative testing of samples of heparinized or un-anticoagulated arterial, venous or capillary whole blood in the laboratory or at the point of care. Chloride measurements from the epoc Blood Analysis System are used in the diagnosis and treatment of electrolyte and metabolic disorders.
The Creatinine test, as part of the epoc Blood Analysis System, is intended for use by trained medical professionals as an in vitro diagnostic device for the quantitative testing of samples of heparinized or un-anticoagulated arterial, venous or capillary whole blood in the laboratory or at the point of care. Creatinine measurements from the epoc Blood Analysis System are used in the diagnosis and treatment of certain renal diseases and in monitoring renal dialysis.
The epoc Blood Analysis System is an in vitro analytical system comprising a network of one or more epoc Readers designed to be used at the point of care (POC). The readers accept an epoc single use test card containing a group of sensors that perform diagnostic testing on whole blood. The blood test results are transmitted wirelessly to an epoc Host, which displays and stores the test results. The epoc System is intended for use by trained medical professionals as an in vitro diagnostic device for the quantitative testing of samples of whole blood. The test card panel configuration currently includes sensors for Sodium Na, Potassium K, Ionized Calcium iCa, pH, pCO2, pO2, Lactate, Glucose and Hematocrit Hct. This submission adds Chloride and Creatinine to this list of approved tests.
This medical device (epoc System) is an in vitro analytical system that provides diagnostic testing for various analytes in whole blood. This submission adds Chloride and Creatinine tests to its existing capabilities.
Here’s a breakdown of the acceptance criteria and supporting studies:
1. Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria are generally implied through the comparison with predicate devices and established standards like CLSI recommendations. The reported device performance is presented in various non-clinical and clinical studies.
Chloride Test
| Acceptance Criteria / Performance Metric | Predicate Device (i-STAT™ Chloride) Target / Standard | epoc Chloride Test Performance |
|---|---|---|
| Intended Use | Diagnosis and treatment of electrolyte and metabolic disorders, including cystic fibrosis, diabetic acidosis, and hydration disorders. | Diagnosis and treatment of electrolyte and metabolic disorders. |
| Where Used | Hospital, point of care | Hospital, point of care |
| Sample Type | Venous, arterial, and capillary whole blood | Venous, arterial and capillary whole blood (heparinized or un-anticoagulated) |
| Reportable Range | 65 - 140 mmol/L | 65 - 140 mmol/L (supported by linearity study, Section 5.5.2) |
| Detection Principle | Ion selective membrane potentiometry | Ion selective membrane potentiometry |
| Sample Volume | 100 μL | At least 92 μL |
| Aqueous Precision (Total %CV) | (Implicitly comparable to predicate) | Level 1: 0.5%, Level 3: 0.7% (Figure 5.3) Clinical Site (various users, all sites, control fluids): Level 1: 0.7%, Level 2: 0.6%, Level 3: 0.9% (Figure 5.14) |
| Whole Blood Precision (Avg. SD W-R / CV% W-R) | (Implicitly comparable to predicate) | Syringe: Normal: 0.63 / 0.6%, Spiked: 0.86 / 0.7% (Figure 5.10) Capillary: Normal: 0.70 / 0.7%, Spiked: 1.11 / 0.9% (Figure 5.10) |
| Method Comparison (vs. Predicate/Comparator) | (Expected high correlation, low bias) | vs. non-POC Systems: R² = 0.96, Mean Bias at 112 mM = -1.3 (Figure 5.5) vs. Abbott i-STAT: R² = 0.98, Mean Bias at 112 mM = -1.0 (Figure 5.5) Various Matrices vs. i-STAT: R² ranges from 0.97 to 0.99 for venous, arterial, capillary (Figure 5.15), Avg. Bias at decision levels from -0.9 to 0.0 (Figure 5.16) |
Creatinine Test
| Acceptance Criteria / Performance Metric | Predicate Device (Roche Cobas c 511/512 CREP2) Target / Standard | epoc Creatinine Test Performance |
|---|---|---|
| Intended Use | Quantitative determination of creatinine in human serum, plasma, and urine for diagnosis of renal diseases. | Diagnosis and treatment of certain renal diseases and in monitoring renal dialysis. |
| Where Used | Hospital, laboratory | Hospital, point of care |
| Sample Type | Serum, Plasma, Urine | Venous, arterial and capillary whole blood (heparinized or un-anticoagulated) |
| Reportable Range | 0.03 - 30 mg/dL | 0.3 - 15.0 mg/dL (supported by linearity study, Slope 1.00, Intercept 0.07, R² 0.99 for 0.25 - 15.5 mg/dL, Section 5.5.2) |
| Detection Principle | Enzymatic cascade reaction (creatininase, creatinase, sarcosine oxidase) leading to peroxidase-catalyzed chromogenic peroxide detection. | Enzymatic cascade reaction (creatininase, creatinase, sarcosine oxidase) leading to amperometric peroxide detection. |
| Sample Volume | 2-5 μL | At least 92 μL |
| Aqueous Precision (Total %CV) | (Implicitly comparable to predicate) | Level 1: 4.9%, Level 3: 4.1% (Figure 5.3) Clinical Site (various users, all sites, control fluids): Level 1: 6.8%, Level 2: 6.4%, Level 3: 6.3% (Figure 5.14) |
| Whole Blood Precision (Avg. SD W-R / CV% W-R) | (Implicitly comparable to predicate) | Syringe: Normal: 0.05 / 7.6%, Spiked: 0.06 / 3.9% (Figure 5.10) Capillary: Normal: 0.04 / 6.8%, Spiked: 0.06 / 3.9% (Figure 5.10) |
| Method Comparison (vs. Predicate) | (Expected high correlation, low bias) | vs. Roche Cobas 6000: R² = 0.99, Mean Bias at 1.25 mg/dL = -0.06 (Figure 5.6) Various Matrices vs. Roche Cobas: R² ranges from 0.99 for venous, arterial, capillary (Figure 5.15), Avg. Bias at decision levels from -0.04 to -0.08 (Figure 5.16) |
2. Sample Sizes Used for the Test Set and Data Provenance
Chloride Test:
- Method Comparison (Clinical Field Trials, Patient Samples):
- vs. non-POC Systems (Roche Cobas 6000, Siemens Advia): N = 96 (pooled venous samples, approximately equal numbers vs. each system). Data provenance not explicitly stated but implies clinical sites (hospitals).
- vs. Abbott i-STAT 300 (Predicate): N = 155 (patient samples, approximately equal numbers of venous, arterial, and capillary samples). Data provenance implies clinical sites (hospitals).
- Matrix Effects (Clinical Field Trials, Patient Samples):
- Venous: N = 49
- Arterial: N = 43
- Capillary: N = 63
- All: N = 155 (These are subsets of the Abbott i-STAT comparison data)
- Blood Precision (Clinical Sites, End Users):
- Chloride Blood Precision Site 1: 4 users, 10-10 replicates each for normal/spiked syringe samples. Total N around 80.
- Chloride Blood Precision Site 2: 8 users for syringe, 4 users for capillary. 10-11 replicates each. Total N around 220.
- Overall Blood Precision Summary:
- Normal Syringe: 120 tests (12 runs, 10 replicates)
- Spiked Syringe: 119 tests (12 runs, 10 replicates)
- Normal Capillary: 40 tests (4 runs, 10 replicates)
- Spiked Capillary: 40 tests (4 runs, 10 replicates)
- Anticoagulant Effect: 46 samples from a hospital, supplemented with 29 in-house samples.
- Provenance: Clinical field trials at two hospitals (patient samples), and in-house studies (aqueous precision, linearity, detection limit, analytical specificity, some anticoagulant effect evaluation). The data is a mix of prospective (patient samples collected in clinical field trials) and retrospective (in-house studies using prepared samples).
Creatinine Test:
- Method Comparison (Clinical Field Trials, Patient Samples):
- vs. Roche Cobas 6000 (Predicate): N = 144 (patient samples, approximately equal numbers of venous, arterial, and capillary samples). Data provenance implies clinical sites (hospitals).
- Matrix Effects (Clinical Field Trials, Patient Samples):
- Venous: N = 53
- Arterial: N = 42
- Capillary: N = 49
- All: N = 144 (These are subsets of the Roche Cobas comparison data)
- Blood Precision (Clinical Sites, End Users):
- Creatinine Blood Precision (multiple sites, multiple users): Each user performed 9-10 replicates for normal/spiked syringe and capillary samples. Total N is around 118 for syringe (normal and spiked), and around 30 for capillary (normal and spiked).
- Overall Blood Precision Summary:
- Normal Syringe: 118 tests (12 runs, 10 replicates)
- Spiked Syringe: 118 tests (12 runs, 10 replicates)
- Normal Capillary: 29 tests (3 runs, 10 replicates)
- Spiked Capillary: 30 tests (3 runs, 10 replicates)
- Anticoagulant Effect: 46 samples from a hospital, supplemented with 29 in-house samples.
- Provenance: Clinical field trials at a hospital site (patient samples), and in-house studies (aqueous precision, linearity, detection limit, analytical specificity, some anticoagulant effect evaluation). The data is a mix of prospective (patient samples collected in clinical field trials) and retrospective (in-house studies using prepared samples).
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
The document does not explicitly state the number of experts used or their specific qualifications (e.g., "radiologist with 10 years of experience") for establishing ground truth for the test set. Instead, the "ground truth" or reference values for the clinical method comparison studies were established by:
- Predicate Devices: i-Stat™ Model 300 Portable Clinical Analyzer (for Chloride) and Roche Cobas c 511/512 CREP2 Creatinine Plus ver. 2 assay (for Creatinine). These are legally marketed devices that provide accepted reference measurements.
- Comparative Instruments/Laboratory Methods: Other non-point-of-care systems (e.g., Roche Cobas 6000, Siemens Advia for Chloride) and a serum-based laboratory method (for Creatinine) at clinical sites.
- Traceability: Both Chloride and Creatinine concentration values assigned to controls and calibrator fluids are traceable to NIST standards.
The expertise lies in the established and validated methodologies of these predicate and comparative devices/laboratory methods, rather than individual expert adjudication for each test case.
4. Adjudication Method for the Test Set
No explicit "adjudication method" in the sense of expert review for discrete cases (like 2+1, 3+1) is described. For in vitro diagnostic devices like this, the performance is typically evaluated by comparing the device's measurements against established reference methods (predicate devices or laboratory analyzers) which are considered the "ground truth." The statistical analysis (regression, bias, R²) serves as the method to determine agreement.
5. If a Multi Reader Multi Case (MRMC) Comparative Effectiveness Study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance
An MRMC study, particularly in the context of human readers and AI assistance, is not applicable to this device. This is an automated in vitro diagnostic system that directly measures analytes in blood. There are no "human readers" interpreting images or data that AI would assist. The device itself performs the analysis, and the studies assess its accuracy, precision, and agreement with reference methods.
6. If a Standalone (i.e. algorithm only without human-in-the-loop performance) was done
Yes, the studies presented are essentially "standalone" performance evaluations of the device. The epoc Blood Analysis System is an automated system where the test card is inserted, blood is introduced, and analytical steps are performed automatically. The output is a direct measurement of analyte concentrations. The "human-in-the-loop" aspect primarily involves trained medical professionals collecting samples and operating the device, but not in interpreting raw data or making diagnostic decisions that the device's algorithm would assist. The performance data (precision, linearity, method comparison) reflects the device's inherent analytical capabilities.
7. The Type of Ground Truth Used (expert consensus, pathology, outcomes data, etc.)
The ground truth used for these studies is primarily:
- Reference Method Comparison: Measurements from legally marketed and established predicate devices (i-Stat™ for Chloride, Roche Cobas for Creatinine) and other validated laboratory methods (e.g., Roche Cobas 6000, Siemens Advia).
- Traceability to Standards: Calibration and control values are traceable to NIST (National Institute of Standards and Technology) standards (SRM 967 for Creatinine).
8. The Sample Size for the Training Set
The document does not explicitly mention a "training set" in the context of machine learning or AI algorithms as the primary component of the device's measurement principle. The device relies on electrochemical sensors (ion-selective membrane potentiometry for Chloride, enzymatic cascade with amperometric detection for Creatinine).
However, in the broader sense of device development and calibration:
- In-house aqueous precision study: N=240 for Chloride (L1, L3) and N=239/241 for Creatinine (L1, L3) are mentioned (Figure 5.3). While these are presented as evaluation data, similar-sized or larger datasets would likely be used during initial development and calibration.
- Linearity study: Involved nine blood samples prepared from two pools, evaluated against an in-house standard method.
- The development and optimization of the enzymatic reactions and sensor response curves would involve extensive testing with many samples during the device's R&D phase, which functionally serves a "training" purpose for the device's internal calibration and algorithms. This specific data is not detailed as a distinct "training set" with a quantifiable size in this 510(k) summary.
9. How the Ground Truth for the Training Set Was Established
As noted above, a distinct "training set" with specific ground truth establishment isn't explicitly detailed in the context of an AI/ML device. For a sensor-based diagnostic device like this, the "ground truth" for calibrating and optimizing the sensors (analogous to training) would be established through:
- NIST Traceability: Calibrator and control fluids are assigned values traceable to NIST standards. This is the ultimate ground truth for establishing the accuracy of the measurements.
- Reference Laboratory Methods: During development, the device would have been extensively correlated with established laboratory methods to ensure its measurements align with accepted clinical standards.
- Controlled Samples: Use of precisely prepared aqueous solutions, spiked blood samples, and pooled human serum with known concentrations, following guidelines like CLSI EP6-A and EP7-A2 for linearity, detection limits, and analytical specificity.
Ask a specific question about this device
(231 days)
The Lactate test, as part of the epoc Blood Analysis System, is intended for use by trained medical professionals as an in vitro diagnostic device for the quantitative testing of samples of heparinized or un-anticoagulated arterial, venous or capillary whole blood in the laboratory or at the point of care in hospitals, nursing homes or other clinical care institutions.
Lactate measurements from the epoc Blood Analysis System are used to evaluate the acid-base status and are used in the diagnosis and treatment of lactic acidosis (abnormally high acidity of the blood).
The epoc Lactate Test is being added as an additional sensor to the existing single use test card that is used with the epoc Blood Analysis System. This test card is inserted into the epoc Reader and all analytical steps are performed automatically. Patient and user information may be entered into the mobile computing device (epoc Host) during the automated analysis cycle.
The epoc Blood Analysis System is an in vitro analytical system comprising a network of one or more epoc Readers designed to be used at the point of care (POC). The readers accept an epoc single use test card containing a group of sensors that perform diagnostic testing on whole blood. The blood test results are transmitted wirelessly to an epoc Host, which displays and stores the test results.
Here's a summary of the acceptance criteria and study information for the epoc Lactate Test, based on the provided document:
1. Table of Acceptance Criteria and Reported Device Performance (Lactate only)
| Acceptance Criteria Category | Specific Metric (Lactate) | Acceptance Criteria (Implicit) | Reported Device Performance (Lactate) |
|---|---|---|---|
| Precision (Aqueous Controls) | Within-Device CV% | Not explicitly stated but expected to be low for clinical accuracy. | L1: 4.9% (WD), 6.3% (Total); L3: 3.1% (WD), 4.7% (Total) |
| Precision (Blood Samples - Site 1) | %CV | Not explicitly stated. | WB L1: 6.0% (Phlebotomist 1), 3.3% (Phlebotomist 2) |
| Precision (Aqueous Controls - Site 1) | %CV | Not explicitly stated. | L3: 3.3% (RN 1), 2.9% (Anesthesia Tech); L2: 1.8% (RN 2), 2.8% (Resp Therapist) |
| Linearity/Reportable Range | Test Range (mM) | 0.3 - 20 mmol/L (as specified in technology comparison) | 0.001 - 20.1 mM (appears to be "test range" based on the table, indicating it covers the specified range) |
| Traceability | To NIST standards | Device calibrated and QC materials traceable to NIST. | Explicitly states: "epoc System is calibrated is against methods traceable to NIST standards." and "Calibration verification uses commercially available calibration verification fluids whose concentration values are traceable to NIST standards." |
| Detection Limit | Statistically discernable from Limit of Blank | Test's low end of reportable range >= Limit of Detection | Low end of reportable range (0.30 mM) is >= Limit of detection and statistically discernable from Limit of Blank (0.21 mM). |
| Analytical Specificity (Interference Bias) | Unacceptable bias defined as >5% significant error. | Max 5% significant error. | Several interfering substances listed with specific bias values; most found to be "insignificant" at tested levels. |
| Method Comparison (vs. Predicate) | R² | Generally >0.95 for good correlation. | 0.9711 (overall), 0.9769 (venous), 0.9829 (arterial), 0.9653 (capillary) |
| Method Comparison (vs. Predicate) | Slope | Ideally close to 1. | 0.967 (overall) |
| Method Comparison (vs. Predicate) | Intercept | Ideally close to 0. | 0.132 (overall) |
| Method Comparison (vs. Predicate) - Bias | Average Bias for decision levels | Expected to be small, with narrow 95% CI. | At 2.2mM: 0.061 (all) with 95% CI ± 0.119; At 5.0mM: -0.031 (all) with 95% CI ± 0.084 |
| Effect of Anticoagulant (vs. Predicate) | R² | Generally >0.95 for good correlation. | 0.9916 |
| Effect of Anticoagulant (vs. Predicate) | Slope | Ideally close to 1. | 1.036 |
| Effect of Anticoagulant (vs. Predicate) | Intercept | Ideally close to 0. | -0.045 |
2. Sample Size and Data Provenance
- Test Set for Method Comparison:
- Sample Size: 373 patient samples for overall method comparison; broken down into 126 venous, 73 arterial, 174 capillary samples for matrix effects.
- Data Provenance: Field trials at several hospitals on "patient samples of whole blood at various locations." This indicates prospective, real-world data from multiple sites (likely within North America given the FDA submission).
- Test Set for Blood Precision:
- Sample Size: 15 samples per user at Site 1, and unspecified number at Site 2 (likely similar).
- Data Provenance: Field trials at two (2) hospitals on "volunteer samples of whole blood by potential end users." This indicates prospective, real-world data from multiple sites.
- Test Set for Aqueous Precision:
- Sample Size: 15 samples per user per QC level at Site 1, and unspecified number at Site 2 (likely similar).
- Data Provenance: Field trials at two (2) hospitals on commercially available control fluids by potential end users.
- Test Set for Anticoagulant Effect:
- Sample Size: 60 samples (43 from hospital POC sites, 17 from in-house studies).
- Data Provenance: Patient samples from hospital POC sites and in-house studies.
3. Number of Experts and their Qualifications (for Test Set Ground Truth)
- The document describes the predicate device as the "gold standard" for comparison. The ground truth for the method comparison studies was established by the predicate device (i-Stat™ Lactate Test using i-Stat™ Model 300 Portable Clinical Analyzer).
- No specific number of human experts or their qualifications for establishing ground truth are mentioned, as the comparison is against another established device. However, the predicate device itself would have undergone its own validation with expert input.
4. Adjudication Method
- Given that the ground truth is established by a predicate analytical device, there is no human adjudication method (like 2+1 or 3+1) described or applicable in this context. The comparison is quantitative against readings from the predicate device.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- A formal MRMC comparative effectiveness study, as typically performed for diagnostic imaging devices involving human readers, was not conducted or described.
- The studies involved different "users" (phlebotomists, RNs, Anesthesia Techs, Resp Therapists) performing tests, which is a form of multi-reader study, but it's focused on device precision/reproducibility across different operators rather than assessing AI assistance for human diagnostic performance. Therefore, there is no effect size reported for human readers improving with/without AI assistance.
6. Standalone Performance Study
- Yes, a standalone performance study was done. The entire document describes the standalone performance of the epoc Lactate Test (algorithm and device combined) against a predicate device and established analytical standards. The reported device performance metrics in the tables (precision, linearity, method comparison slope, intercept, R², bias) are all measures of the device's standalone performance. There is no human-in-the-loop component for result interpretation.
7. Type of Ground Truth Used
- The primary ground truth for the clinical and non-clinical studies is:
- Readings from a legally marketed predicate device (i-Stat™ Lactate Test using i-Stat™ Model 300 Portable Clinical Analyzer) for method comparison studies.
- NIST traceable standards for calibration, quality control, and linearity studies.
- Pooled human serum and blood samples (spiked with known interferents or aged to increase lactate) for analytical specificity and blood precision studies.
8. Sample Size for the Training Set
- The document does not explicitly state the sample size for the training set for the epoc Lactate Test development. This type of submission (510(k) for an IVD) typically focuses on validation data rather than internal development/training data for the algorithm.
9. How the Ground Truth for the Training Set was Established
- As the training set size is not disclosed, the method for establishing its ground truth is also not explicitly described in this document. However, given the nature of the device (a quantitative sensor measurement system), the training/development likely involved:
- Controlled reference materials with known lactate concentrations.
- Comparison to reference laboratory methods known to be accurate and traceable to NIST standards.
Ask a specific question about this device
(195 days)
The epoc Blood Analysis System is intended for use by trained medical professionals as an in vitro diagnostic device for the quantitative testing of samples of heparinized or unanticoagulated arterial, venous or capillary whole blood in the laboratory or at the point of care in hospitals, nursing homes or other clinical care institutions.
Care-Fill Capillary Tubes are intended for use with the epoc Blood Analysis system and are used for the collection and dispensing of capillary blood samples with epoc Test Cards.
The Blood Gas Electrolyte (BGE) test card panel configuration includes sensors for Sodium -Na, Potassium - K, Calcium - iCa, pH, pCO2, pO2 and Hematocrit - Hct.
The Blood Gas Electrolyte (BGEM) test card panel configuration includes sensors for Sodium - Na, Potassium - K, Calcium - iCa, pH, pCO2, pO2, Hematocrit - Hct and Glucose -Glu.
Measurement of sodium is used in diagnosis and treatment of diseases involving electrolyte imbalance.
Measurement of potassium is used in diagnosis and treatment of diseases involving electrolyte imbalance.
Measurement of Ionized Calcium is used in diagnosis and treatment of parathyroid disease, a variety of bone diseases, chronic renal disease and tetany.
Measurement of pH, pCO2, pO2 (blood gases) is used in the diagnosis and treatment of lifethreatening acid-base disturbances.
Measurement of Hct distinguishes normal from abnormal states of blood volume, such as anemia and erythrocytosis.
Glucose measurements are used in the diagnosis and treatment of carbohydrate metabolism disorders including diabetes mellitus, idiopathic hypoglycemia, and of pancreatic islet cell tumors.
The epoc Blood Analysis System consists of three (3) components:
- epoc Test Card: single use blood test card with sensors, fluidic channel, and on-board calibrator.
- epoc Card Reader: raw-signal acquisition peripheral with card orifice, mechanical actuation assembly, bar code scanner, electrical contact array, thermal subsystem, and circuits for signal processing and wireless transmission.
- epoc Host: dedicated-use Personal Digital Assistant (PDA) computing device with custom software for displaying test results.
The epoc Care-Fill Capillary Tube is intended for use only with epoc Blood Analysis System for the collection and dispensing of capillary blood samples.
Acceptance Criteria and Study Details for epoc® Blood Analysis System for Capillary Samples
The epoc® Blood Analysis System sought clearance to use capillary blood specimens and to remove the limiting labeling regarding the glucose test using neonatal samples. The acceptance criteria were implicitly established by demonstrating substantial equivalence to the predicate device, the i-STAT® Model 300 Portable Clinical Analyzer, primarily through method comparison studies and precision studies. The core of the acceptance criteria is the observed bias between the epoc system and the i-STAT system, and the precision (SD and %CV) of the epoc system.
1. Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria are not explicitly defined as numerical thresholds in the provided document. Instead, the study aims to demonstrate that the epoc system's performance, when using capillary blood, is "substantially equivalent" to the predicate device (i-STAT Model 300). This is assessed by comparing the biases and precision against the predicate device in clinical and non-clinical settings.
For both method comparison studies, the key performance indicator is the average(Yii-Xii), which represents the bias between the epoc system (Y) and the predicate i-STAT system (X). For precision, SD and %CV are used.
Implicit Acceptance Criteria (interpreted from "Substantially Equivalent" and comparison to predicate):
- Bias (epoc vs. i-STAT): The measured biases should be clinically acceptable and comparable to or better than previously cleared devices and clinical standards. The document presents the observed biases without stating explicit thresholds for acceptance a priori.
- Precision (epoc system): The standard deviation (SD) and coefficient of variation (%CV) for each analyte should be within clinically acceptable ranges and consistent with expected performance for point-of-care blood gas and electrolyte analyzers.
Reported Device Performance (from studies):
The following table summarizes the reported performance of the epoc system when using capillary samples, compared to the i-STAT system, and its precision.
| Parameter | Performance Metric | Reported Value (Method Comparison: epoc/CareFill vs i-STAT/CliniTube, Clinical Study) | Reported Value (Precision Study: epoc/CareFill, Field Trial) - Example Range |
|---|---|---|---|
| pH | Average Bias (Y-X) | -0.02 | N/A (Bias not measured against a predicate in precision study) |
| SD (%CV) - Example from Clinical Precision Study | N/A | 0.006-0.013 (0.1%-0.2%) | |
| pCO2 | Average Bias (Y-X) | 1.5 | N/A |
| SD (%CV) - Example from Clinical Precision Study | N/A | 0.5-1.2 (1.6%-2.8%) | |
| pO2 | Average Bias (Y-X) | 2.3 | N/A |
| SD (%CV) - Example from Clinical Precision Study | N/A | 1.2-9.1 (1.6%-7.4%) | |
| Na | Average Bias (Y-X) | -2.5 | N/A |
| SD (%CV) - Example from Clinical Precision Study | N/A | 0.5-1.5 (0.4%-1.0%) | |
| K | Average Bias (Y-X) | -0.2 | N/A |
| SD (%CV) - Example from Clinical Precision Study | N/A | 0.04-0.24 (1.4%-3.5%) | |
| iCa | Average Bias (Y-X) | -0.041 | N/A |
| SD (%CV) - Example from Clinical Precision Study | N/A | 0.008-0.028 (0.8%-2.5%) | |
| Glu | Average Bias (Y-X) | 0.53 (for all capillary) | N/A |
| Average Bias (Y-X) - Neonatal Capillary | 1.8 (at Decision Level 1), -5.2 (at Decision Level 2) | N/A | |
| SD (%CV) - Example from Clinical Precision Study | N/A | 1.5-8.6 (2.9%-3.9%) | |
| Hct | Average Bias (Y-X) | -4.5 | N/A |
| SD (%CV) - Example from Clinical Precision Study | N/A | 0.3-1.4 (1.4%-2.9%) |
2. Sample Sizes and Data Provenance
Equivalence of Care-Fill vs. Syringe (Non-Clinical Study):
- Sample Size: N = 42 for all analytes.
- Data Provenance: Retrospective, "in house" experiments. The origin of the blood samples (e.g., human, animal, spiked, etc.) and country are not explicitly stated, but the context implies laboratory-controlled samples, potentially modified to extend analyte ranges.
In-house Method Comparison (Capillary samples, epoc vs. i-STAT):
- Sample Size: N = 51 for pH/pCO2, N = 52 for pO2/Na/K/Ca/Glu/Hct.
- Data Provenance: Retrospective, "in-house" study using capillary blood samples. Origin and country are not specified.
Clinical Precision Study (Care-Fill capillary tubes):
- Sample Size: N = 10 replicates per study, across 6 precision studies (3 pools of blood, 2 POC locations, 6 different operators). So, 60 measurements per analyte in total, distributed.
- Data Provenance: Prospective, patient samples, collected at "2 POC locations" (Nursery and NICU), implying a hospital setting. Country not explicitly stated but implied to be where epoc is manufactured/marketed (Canada/USA).
Clinical Method Comparison (Capillary samples, epoc vs. i-STAT):
- Sample Size: N = 47 for pH/iCa/Hct, N = 48 for pCO2/pO2/Na/K/Glu. For neonatal glucose specifically, N = 36.
- Data Provenance: Prospective, patient samples of whole blood (12 adult capillary, 36 neonatal capillary). Performed "at a hospital" in "four (4) locations: NICU, Wellbaby Nursery and two (2) different outpatient drawing areas." Country not explicitly stated, but implied as per above.
3. Number of Experts and Qualifications for Ground Truth
The concept of "experts" and "ground truth" as typically used in AI/image analysis studies (e.g., radiologists interpreting images) is not directly applicable here. This submission is for a medical device that measures physiological parameters.
- Ground Truth Establishment: The ground truth or reference standard for comparison in these studies is the predicate device, the i-STAT Model 300 Portable Clinical Analyzer. The i-STAT system itself is a cleared device already accepted as an accurate measurement tool for these parameters. There is no mention of human experts establishing a separate "ground truth" or reference, beyond the inherent accuracy of the predicate device.
4. Adjudication Method
Not applicable. As explained in point 3, the ground truth is established by the predicate device, not by human interpretation or consensus that would require an adjudication method.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
Not applicable. This is not a study involving human readers interpreting cases (e.g., medical images) with and without AI assistance. It is a device performance study comparing a new device (epoc) to a predicate device (i-STAT) for quantitative measurements. Therefore, there is no "effect size of how much human readers improve with AI vs without AI assistance."
6. Standalone Performance Study
Yes, a standalone performance study was conducted.
- Equivalence of Care-Fill vs. Syringe (Non-Clinical Study): This study directly compared two delivery methods on the epoc system, not against an external reference, to ensure the new capillary tube delivery method did not alter results.
- Clinical Precision Study: This study evaluated the precision (repeatability) of the epoc system itself when using capillary samples collected via the Care-Fill tubes, without direct comparison to a predicate device for each measurement. It assessed the algorithm's consistency and reliability in real-world use.
7. Type of Ground Truth Used
The ground truth used for the comparative studies (method comparison) was the measurements obtained from the predicate device, the i-STAT Model 300 Portable Clinical Analyzer. This is considered "reference method" ground truth, where a previously validated and cleared device serves as the standard.
8. Sample Size for the Training Set
The document does not explicitly mention a separate "training set" as would be typical for machine learning algorithms. The epoc system is described as having "on-board calibrator" and "custom software that displays the test results" and "software to control the test and calculate analytical values from raw sensor signals." This implies a rule-based or empirically calibrated system rather than a machine learning model that requires a distinct training phase with a labeled dataset in the contemporary sense. The calibration and development likely involved extensive in-house testing and engineering, but these are not referred to as a "training set."
9. How the Ground Truth for the Training Set Was Established
Given that a distinct "training set" in the context of machine learning is not mentioned (see point 8), the concept of establishing ground truth for it also does not directly apply. The calibration and performance optimization of the epoc system's algorithms/software would have been established through a combination of:
- Reference materials/standards: Calibrators are mentioned as "on-board" the test card.
- Extensive laboratory testing: Comparison against established laboratory methods and reference analyzers during the development and validation phases.
- Empirical data collection: Using various blood samples (e.g., with known concentrations, or compared to highly accurate laboratory instruments) to optimize the sensor responses and calculation algorithms.
Ask a specific question about this device
(160 days)
The Glucose test, as part of the epoc Blood Analysis System is intended for use by trained medical professionals as an in vitro diagnostic device for the quantitative testing of samples of heparinized or un-anticoagulated arterial or venous whole blood in the laboratory or at the point of care in hospitals, nursing homes or other clinical care institutions.
Glucose measurements from the epoc Blood Analysis System are used in the diagnosis and treatment of carbohydrate metabolism disorders including diabetes mellitus, and idiopathic hypoglycemia, and of pancreatic islet cell tumors.
The EPOC glucose test is being added as an additional sensor to the existing single use test card that is used with the EPOC Blood Analysis System. This test card is inserted into the EPOC Reader and all analytical steps are performed automatically. Patient and user information may be entered into the mobile computing device (EPOC Host) during the automated analysis cycle.
The EPOC Blood Analysis System is an in vitro analytical system comprising a network of one or more EPOC Readers designed to be used at the point of care (POC). The readers accept an EPOC single use test card containing a group of sensors that perform diagnostic testing on whole blood. The blood test results are transmitted wirelessly to an EPOC Host, which displays and stores the test results.
Here's a detailed breakdown of the acceptance criteria and the study that proves the device meets them, based on the provided text:
Device: EPOC Glucose Test (part of the EPOC Blood Analysis System)
1. Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria for the EPOC Glucose Test are not explicitly stated as quantitative targets in the document. Instead, the studies demonstrate the device's performance characteristics (precision, linearity, hematocrit effect, and analytical specificity) and then conclude equivalence to the predicate device. The performance data itself acts as the evidence to satisfy implicit acceptance criteria generally expected for such a device (i.e., that it performs reliably and comparably to a legally marketed device).
Therefore, the table below summarizes the reported performance characteristics from the provided studies. The acceptance criteria are inferred as demonstrating comparable or acceptable performance for each metric.
| Performance Metric | Acceptance Criteria (Inferred) | Reported Device Performance |
|---|---|---|
| Aqueous Precision (CV%) | Low variability (e.g., comparable to industry standards) | L1: 2.30% CVL3: 2.30% CV |
| Blood Precision (CV%) - Glucose 20 | ≤ 4.8% CV (based on predicate/industry standards) | 5.4% CV (Total) at 20 mg/dL (Note: One lot met 4.8% CV) |
| Blood Precision (CV%) - Glucose 120 | ≤ 4% CV (based on predicate/industry standards) | 2.4% CV (Total) at 120 mg/dL |
| Blood Precision (CV%) - Glucose 200 | ≤ 4% CV (based on predicate/industry standards) | 3.9% CV (Total) at 200 mg/dL |
| Blood Precision (CV%) - Glucose 300 | ≤ 6% CV (based on predicate/industry standards) | 4.2% CV (Total) at 300 mg/dL |
| Blood Precision (CV%) - Glucose 500 | ≤ 6% CV (based on predicate/industry standards) | 3.2% CV (Total) at 500 mg/dL |
| Linearity (Slope) | Close to 1.0 | 0.9996 |
| Linearity (Intercept) | Close to 0.0 | 0.64 |
| R-squared (Method Comparison) | High correlation (e.g., > 0.95 or 0.98) | R2: 0.999 (overall vs. predicate) |
| Hematocrit Effect (Bias) | Minimal bias across different Hct levels | Varied from -18.9 mg/dL to +20.0 mg/dL, with most biases being smaller than these extremes across different glucose and Hct levels. (Specific criteria for acceptable bias are not explicitly stated, but the values are presented as acceptable by the manufacturer.) |
| Analytical Specificity (Interference Bias) | Interferent bias < Total Allowable Error (TE) | Most interferents showed a bias as a fraction of TE to be < 1.0, with a few exceptions (e.g., L-Cysteine at 1.5 mM, NaFluoride at 100 mM, Mannose at 5 mM, CaOxalate at 78 mM, Thiocyanate at 6.9 mM, Uric Acid at 1.5 mM) exceeding this threshold. The document presents these results as acceptable. |
| Method Comparison (vs. Predicate) | Equivalent performance to the i-STAT Model 300 | Slope ≈ 1.022, Intercept ≈ 2.338, R² ≈ 0.999 (overall) |
| Anticoagulant Effect | Similar performance for heparinized vs. unheparinized blood | Slope: 0.994 (heparinized), 1.019 (unheparinized) |
| Venous vs. Arterial Blood | Similar performance for arterial vs. venous blood | Slope: 0.991 (arterial), 1.028 (venous) |
| Altitude Effect | Equivalent performance at high altitude | Slope: 1.031, R²: 0.9976 (overall vs. ABL800Flex) |
2. Sample Sizes Used for the Test Set and Data Provenance
-
Aqueous Precision: Not specified for each level, but "a twenty day precision study performed on 4 lots using aqueous controls at two levels L1 and L3."
-
Blood Precision:
- For each of five glucose concentrations (20, 120, 200, 300, 500 mg/dL): "over 100 cards/blood sample on 50 different readers." (e.g., 102 for 20 mg/dL, 98 for 120 mg/dL, 101 for 200 mg/dL, 105 for 300 mg/dL, 103 for 500 mg/dL).
-
Linearity/Reportable Range: "A total of nine blood samples were prepared."
-
Hematocrit Effect: Not explicitly stated for each measurement, but "six glucose level blood linearity studies performed at four different hematocrit levels." These are detailed in a large table.
-
Analytical Specificity: Not explicitly stated (number of replicates for each interferent tested).
-
Method Comparison with Predicate Device:
- Overall: N = 160 patient samples.
- Anticoagulant Effect: N = 58 patient samples (29 heparinized, 29 unheparinized).
- Venous versus Arterial Blood: N = 214 patient samples (100 arterial, 114 venous).
- Altitude Effect: N = 81 patient samples.
-
Data Provenance:
- In-house: Aqueous precision, blood precision, linearity/reportable range, hematocrit effect, analytical specificity.
- Field trials: Method comparison with predicate device, anticoagulant effect, venous versus arterial blood (at several hospitals/POC locations).
- Altitude Effect: Performed at an altitude of over 2000m (~6600 ft).
- Retrospective/Prospective: The text does not explicitly state "retrospective" or "prospective" for the clinical/field trials, but "patient samples" in method comparison studies typically implies prospective collection for the study.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications
The document does not mention the use of "experts" to establish ground truth in the context of diagnostic interpretation. Instead, for quantitative analytical devices like this, ground truth is established using:
- Reference instruments: For precision, linearity, and hematocrit studies, internal reference instruments (e.g., YSI, ISTAT, ABL) with traceability to NIST standards were used.
- Predicate device: For method comparison, the i-Stat™ Model 300 Portable Clinical Analyzer served as the reference.
- Other reference instrument: For the altitude study, the ABL800 Flex Radiometer whole blood instrument was used as the reference.
Therefore, the "ground truth" is established by these reference methods, not by human experts interpreting results.
4. Adjudication Method for the Test Set
Not applicable. This device is a quantitative diagnostic test for glucose. Ground truth is established by chemical reference methods and comparison to predicate devices, not by interpretation that requires adjudication (e.g., 2+1, 3+1).
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance
Not applicable. This is a standalone quantitative glucose measurement device, not an AI-assisted diagnostic imaging or interpretation tool. There are no "human readers" in the context of MRMC studies for this type of device.
6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was done
Yes, the entire evaluation is for the standalone performance of the EPOC Glucose Test. The studies described (precision, linearity, method comparison, etc.) assess the accuracy and reliability of the device's glucose measurements directly, without involving human interpretation or integration into a diagnostic workflow for performance assessment. The device provides a quantitative numerical result.
7. The Type of Ground Truth Used
The ground truth for the device's performance evaluation was established using:
- Reference instruments/systems:
- For in-house studies (precision, linearity, hematocrit), the device measurements were compared against "in-house reference instruments with traceability to NIST standards" (e.g., YSI, ISTAT, ABL).
- For calibration and quality control, the system uses materials with traceability to NIST standards (National Institute of Standards and Technology).
- Predicate device: For the primary method comparison studies, the i-Stat™ Model 300 Portable Clinical Analyzer was used as the comparative "ground truth."
- Other legally marketed device: For the altitude study, the ABL800 Flex Radiometer was used as the comparison "ground truth."
8. The Sample Size for the Training Set
The document does not explicitly mention a distinct "training set" in the context of an algorithm's development. This device likely relies on established electrochemical principles, sensor design, and calibration algorithms rather than a machine learning model that requires a separate training data set of patient samples. The "training" in this context would refer to the development and refinement of the sensor and software using internal laboratory samples and calibrators, leading to the final product validated in the studies mentioned.
9. How the Ground Truth for the Training Set Was Established
As noted above, a distinct "training set" for an algorithm in the machine learning sense is not mentioned. However, the development and calibration of the EPOC Glucose Test, which could be considered akin to "training" for a traditional analytical device, involved:
- On-board calibration material: Prepared gravimetrically and assayed on reference systems calibrated with traceability to NIST standards.
- Calibration verification fluids: Commercially available with concentration values traceable to NIST standards.
- Quality control materials: Commercially available fluids with concentrations traceable to NIST standards.
These NIST-traceable reference materials and systems would have been crucial for establishing the accuracy and performance characteristics of the glucose sensor and measurement system during its development and manufacturing.
Ask a specific question about this device
(111 days)
The EPOC Blood Analysis System is intended for use by trained medical professionals as an in vitro diagnostic device for the quantitative testing of samples of whole blood in the laboratory or at the point of care in hospitals, nursing homes or other clinical care institutions.
The Blood Gas Electrolyte (BGE) test card panel configuration includes sensors for Sodium - Na, Potassium - K. ionized Calcium - iCa, pH, pCO2, pO2 and Hematocrit -Hct.
Measurement of Sodium and Potassium are used in diagnosis and treatment diseases involving electrolyte imbalance. Measurement of lonized Calcium is used in diagnosis and treatment of parathyroid disease, a variety of bone diseases, chronic renal disease and tetany. Measurement of ph pCO2, p02 (blood gases) is used in the diagnosis and treatment of life-threatening acid-base disturbances. Measurement Hct distinguish normal from abnormal states of blood volume, such as anemia and erythrocvtosis.
The EPOC Blood Analysis System consists of three (3) components:
- EPOC Test Card: The single use blood test card comprises a port for introduction of a blood sample to an array of sensors on a sensor module. The sensor module is mounted proximal to a fluidic channel contained in a credit-card sized housing. The card has an on-board calibrator contained in a sealed reservoir fluidically connected to the senor array through a valve.
- EPOC Card Reader: The reader is a minimally featured raw-signal acquisition peripheral. The reader comprises a card orifice for accepting a test card, and a mechanical actuation assembly for engaging the test card after it is inserted into the card orifice. Within the reader's card orifice there is a bar code scanner, an electrical contact array for contacting the card's sensor module, and a thermal subsystem for heating the card's measurement region to 37°C during the test. The reader also comprises circuits for amplifying, digitizing and converting the raw sensor signals to a wireless transmittable Bluetooth™ format.
- EPOC Host: The host is a dedicated use Personal Digital Assistant (PDA) computing device with custom software that displays the test results. The reader and host computer together constitute all of the subsystems generally found in a traditional analyzer that operates on unit-use sensors and reagents.
Here's an analysis of the provided text, focusing on the acceptance criteria and study information for the EPOC™ Blood Analysis System:
1. Table of Acceptance Criteria and Reported Device Performance
The document does not explicitly state formal "acceptance criteria" for each parameter in the same way a regulatory body might define them (e.g., "bias must be less than X," or "CV must be less than Y"). However, the non-clinical and clinical test results implicitly serve as the demonstrated "performance" against which substantial equivalence is claimed to a predicate device. For the purpose of this analysis, I will synthesize the linearity data as a primary indicator of performance across relevant ranges.
| Parameter | Accepted Study Performance (Linearity Slope & R²) | Performance against Predicate (Slope & R) |
|---|---|---|
| pH | Slope: 1.021, R²: 0.998 | Slope: 0.966, R: 0.991 |
| pCO2 | Slope: 1.058, R²: 0.998 | Slope: 1.041, R: 0.990 |
| pO2 | Slope: 1.022, R²: 0.999 | Slope: 1.053, R: 0.978 |
| Na | Slope: 0.973, R²: 0.999 | Slope: 1.077, R: 0.953 |
| K | Slope: 1.006, R²: 0.999 | Slope: 1.013, R: 0.993 |
| iCa | Slope: 1.017, R²: 0.998 | Slope: 1.021, R: 0.985 |
| Hct | Slope: 1.005, R²: 0.999 | Slope: 1.066, R: 0.987 |
Note on "Acceptance Criteria": The document claims the device performs effectively based on the non-clinical data and that its clinical performance is equivalent to the predicate device. The strong linearity and high R-squared values for the in-house linearity study (indicating a close fit to a linear model) and the clinical method comparison study (comparing the device to the predicate) would be the basis for these conclusions. Specific numerical acceptance cutoffs are not provided in this summary.
2. Sample Sizes Used for the Test Set and Data Provenance
- Non-Clinical (Precision Study - Aqueous Controls):
- Sample Size: Not explicitly stated as a number of unique samples, but refers to "n=20 replicates on each of 2 controls per day over 20 days" for the 20-day precision study with aqueous controls. This implies 800 measurements for blood gases and electrolytes (20 days * 2 controls * 20 replicates) and 400 measurements for hematocrit (20 days * 2 controls * 10 replicates for each level A and B, assuming 10 replicates per control per day).
- Data Provenance: In-house laboratory.
- Non-Clinical (Precision Study - Whole Blood Field Trials):
- Sample Size: 10 replicates of different whole blood patient samples for each operator at each site. There were 7 operators across 3 sites.
- Data Provenance: Three point-of-care sites (hospitals, nursing homes, or other clinical care institutions), located in Canada (judging by the company address).
- Non-Clinical (Linearity Study):
- Sample Size: Not explicitly stated, but performed "in-house." The results are presented for a "Test range" rather than a number of distinct samples.
- Data Provenance: In-house laboratory.
- Non-Clinical (Interference Studies):
- Sample Size: Not explicitly stated.
- Data Provenance: Not explicitly stated, but implied to be in-house.
- Clinical (Method Comparison Study):
- Sample Size:
- pH: 149
- pCO2: 143
- pO2: 142
- K: 146
- Na: 156
- iCa: 156
- Hct: 142
- Data Provenance: Patient samples of whole blood from a hospital in a field trial. Locations included the intensive care unit, cardiac intensive care unit, hematology/oncology department, and the central lab. Sample types included arterial, venous, and mixed venous/arterial. The country of origin is not explicitly stated in this section, but the company is Canadian, suggesting Canadian sites.
- Sample Size:
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
- This information is not provided in the document. For device performance studies like these, the "ground truth" is typically established by measurements from a reference method (often a central lab analyzer or a predicate device, as used here). The expertise would lie in the operation and validation of these reference methods, rather than clinical consensus.
4. Adjudication Method for the Test Set
- This concept is not directly applicable to the type of device performance studies described (analytical accuracy and precision studies). Adjudication usually pertains to human expert review of clinical cases, particularly in imaging or diagnostic accuracy studies where there's subjectivity. In this case, results are quantitative measurements compared to a reference standard or predicate device.
5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was Done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance
- No, an MRMC comparative effectiveness study was not done. This type of study applies to diagnostic devices where human interpretation is involved, often with AI assistance. The EPOC Blood Analysis System is an in vitro diagnostic device providing quantitative measurements, not an interpretive aid for human readers.
6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was Done
- Yes, the performance studies presented (precision, linearity, interference, and method comparison) are essentially standalone performance studies of the device's ability to accurately measure the target analytes. The device generates quantitative results without requiring human interpretation of raw signals; human-in-the-loop performance is not a relevant concept for this type of automated analyzer. The "operators" in the field trial precision study are performing the sample introduction and use of the device, not interpreting results in a subjective manner.
7. The Type of Ground Truth Used
- Non-Clinical (Precision, Linearity, Interference): The ground truth was established using in-house standard methods with traceability to NIST standards (for linearity) and aqueous controls or prepared samples with known concentrations.
- Clinical (Method Comparison): The ground truth was established by comparing the EPOC system's measurements to those obtained from the predicate device, the i-Stat™ Model 300 Portable Clinical Analyzer. This is a common approach for demonstrating substantial equivalence for in vitro diagnostic devices.
8. The Sample Size for the Training Set
- Not explicitly stated. The document describes performance testing, but there is no explicit mention of a "training set" in the context of machine learning or AI models. This device is an automated sensor-based system, not a machine learning algorithm that requires a distinct training phase in the same way. The development and calibration would involve internal testing, but not a formally defined "training set" as understood in AI studies. The "calibration materials" and "quality control materials" mentioned contribute to the device's operational robustnes and accuracy.
9. How the Ground Truth for the Training Set Was Established
- As noted above, a "training set" as in machine learning is not applicable here. However, the reference materials used for calibration (which could be considered analogous in a broad sense to establishing a "truth" for the device's internal algorithms) are:
- On-board calibration material: Prepared gravimetrically and assayed on reference systems calibrated with traceability to NIST standards.
- Calibration verification fluids: Commercially available, traceable to NIST standards.
- Quality control materials: Commercially available, traceable to NIST standards.
Ask a specific question about this device
Page 1 of 1