Search Results

BioSieve™ Dx Marijuana Test Panel 20 is competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 20 ng/mL. The test provides only preliminary test results. A more specific alternative chemical method must be used in order to obtain a confirmed analytical result. GC/MS or LC/MS is the preferred confirmatory method. For in vitro diagnostic use only.

BioSieve™ Dx Marijuana Test Strip 20 is competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 20 ng/mL.

The test provides only preliminary test results. A more specific alternative chemical method must be used in order to obtain a confirmed analytical result. GC/MS or LC/MS is the preferred confirmatory method. For in vitro diagnostic use only.

BioSieve™ Dx Marijuana Test Panel 50 is competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 50 ng/mL. The test provides only preliminary test results. A more specific alternative chemical method must be used in order to obtain a confirmed analytical result. GC/MS or LC/MS is the preferred confirmatory method. For in vitro diagnostic use only.

BioSieve™ Dx Marijuana Test Strip 50 is competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 50 ng/mL.

BioSieve™ Marijuana Test Panel 50 is competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 50 ng/mL.

BioSieve™ Marijuana Test Strip 50 is competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 50 ng/mL.

Device Description

The BioSieve™ Marijuana Test Panel (Strip) and the BioSieve™ Dx Marijuana Test Panel (Strip) tests are immunochromatographic assays that use a lateral flow system for the qualitative detection of Marijuana in human urine. The products are single-use in vitro diagnostic devices. Each test kit contains a Test Device and a package insert. Each test device is sealed with a desiccant in an aluminum pouch.

AI/ML Overview

The provided document describes the BioSieve™ Marijuana Test Panel and Test Strip devices for qualitative detection of Marijuana in human urine. Here's a breakdown of the acceptance criteria and study information:

1. Table of Acceptance Criteria and Reported Device Performance

The document doesn't explicitly state "acceptance criteria" for the overall device performance in a consolidated table, but rather presents performance characteristics of the device. From the precision study, it can be inferred that the device is expected to correctly identify drug-free samples and samples significantly above the cutoff, while showing a degree of variability around the cutoff. The provided precision data and the results of the comparison studies (with LC/MS) serve as the device's reported performance against implied accuracy and consistency expectations.

Inferred Acceptance Criteria / Performance Goals (based on Precision Study and Comparison Studies):

Performance Characteristic	Acceptance Criteria (inferred)	Reported Device Performance (Summary)
Precision	Consistent results for samples well below and well above the cutoff. Some variability allowed around the cutoff concentrations (e.g., -25% to +25% of cutoff). Specifically, samples at -100%, -75%, -50% cut off should be negative, and samples at +50%, +75%, +100% cut off should be positive. Samples at -25% and +25% should show a mix of results, crossing the cutoff. This is implied by the study design and typical expectations for qualitative immunoassay cutoffs.	For 20 ng/mL Cut-off (Panel & Strip, across 3 lots):
-100%, -75%, -50% cutoff: 100% negative.
+50%, +75%, +100% cutoff: 100% positive.
Results around cutoff (-25%, cutoff, +25%) show expected mixed results (e.g., at cutoff, 32-37 positive, 23-28 negative).
	For 50 ng/mL Cut-off (Panel & Strip, across 3 lots):
-100%, -75%, -50% cutoff: 100% negative.
+50%, +75%, +100% cutoff: 100% positive.
Results around cutoff (-25%, cutoff, +25%) show expected mixed results (e.g., at cutoff, 22-38 negative, 22-38 positive).
Accuracy (Comparison to LC/MS)	High agreement with LC/MS, especially for samples far from the cutoff. Close agreement for samples near the cutoff, while acknowledging some inherent variability in qualitative tests at the cutoff.	For 20 ng/mL Cut-off (Strip & Panel):
High agreement for drug-free, low negative, and high positive samples. Mixed results for near-cutoff negative and positive samples, with some discordant results (false positives and false negatives relative to LC/MS at/near cutoff).
For 50 ng/mL Cut-off (Strip & Panel):
Similar high agreement for drug-free, low negative, and high positive samples, with expected mixed and discordant results around the cutoff.
Interference	No interference from common substances within specified concentrations.	No interference observed from 70+ substances at tested concentrations (e.g., 100µg/mL for most, 1% ethanol, 100 mg/dL albumin).
Specificity (Cross-Reactivity)	Predictable cross-reactivity with structurally similar compounds, with 100% cross-reactivity for the target analyte at the cutoff. Other cannabinoids show varying degrees of cross-reactivity (e.g., 1% for 11-Hydroxy-Δ9-Tetrahydrocannabinol).	Confirmed: 100% cross-reactivity for 11-nor-Δ9-THC-9-COOH and 11-Nor-Δ8-Tetrahydrocannabinol-9-COOH at respective cutoffs. Low cross-reactivity (0.5% - 1%) for Δ8-Tetrahydrocannabinol, Δ9-Tetrahydrocannabinol, and 11-Hydroxy-Δ9-Tetrahydrocannabinol. No cross-reactivity for Cannabinol and Cannabidiol at 100000 ng/mL.
Effect of Urine Specific Gravity & pH	No effect on accuracy within specified ranges.	No effect on accuracy/precision for specific gravity (1.000-1.035) and pH (4-9) at +/- 25% (or +/- 50%) cutoff levels.
Stability	Stable for a specified duration under defined storage conditions.	24 months at 4-30°C based on accelerated and real-time stability studies.
Lay-user Comprehension	Instructions for use are easy to understand and follow, leading to correct test interpretation.	All participants found instructions easy to understand. Flesch-Kincaid read-ability score of Grade Level 7. Overall high percentage of correct results (95-100%) by lay users, with minor exceptions at -25% and +25% cutoff.

2. Sample Size Used for the Test Set and Data Provenance

Precision Study Test Set:
- For each of the four device types (20 ng/mL Strip, 20 ng/mL Panel, 50 ng/mL Strip, 50 ng/mL Panel), 9 concentrations were tested (ranging from -100% cutoff to +100% cutoff).
- For each concentration, tests were performed two runs per day at each site for each of 3 lots for 10 days.
- This equates to: 9 concentrations * 2 runs/day * 3 sites * 3 lots * 10 days = 1620 tests per device type across all concentrations.
- Data Provenance: The document does not specify the country of origin for the samples or the study sites. It implies prospective testing as samples were "prepared by spiking 11-Nor-△9-THC-9-COOH in drug-free urine samples" and confirmed by LC/MS.
Comparison Studies Test Set (Algorithm Only/Standalone):
- For each of the four device types (20 ng/mL Strip, 20 ng/mL Panel, 50 ng/mL Strip, 50 ng/mL Panel), 80 unaltered urine samples were used.
- These 80 samples consisted of 40 negative and 40 positive samples, categorized further into: Drug-Free, Low Negative, Near Cutoff Negative, Near Cutoff Positive, and High Positive.
- Data Provenance: The document does not specify the country of origin of these "unaltered urine samples." It implies these are retrospective samples as they were blind-labeled and compared to LC/MS results.
Lay-user Study Test Set:
- 280 lay persons participated.
- Urine samples were prepared at 7 concentrations (-100%, -75%, -50%, -25%, +25%, +50%, +75% of 50 ng/mL cutoff).
- Each concentration used 20 samples.
- Each participant was given 1 blind-labeled sample. This means a total of 7 concentrations * 20 samples/concentration = 140 samples were tested across the 280 lay persons (since each person tested only one sample).
- Data Provenance: The document does not specify the country of origin. The samples were "prepared by spiking 11-Nor-△9-THC-9-COOH into drug free-pooled urine specimens," indicating prospective preparation.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts

Precision and Comparison Studies: The ground truth for spiked samples (-100% cutoff to +100% cutoff for precision, and the various categories for comparison studies) was established by LC/MS (Liquid Chromatography-Mass Spectrometry). This is a highly accurate and widely accepted analytical method. The document does not mention human experts establishing this ground truth; the instrument itself provides the "truth."
Lay-user Study: The ground truth for the lay-user study samples was also established by LC/MS confirming the drug concentrations after spiking.

4. Adjudication Method for the Test Set

Precision Study: The results are presented as counts of positive/negative readings. No explicit adjudication method (like 2+1) is mentioned, as is common for analytical precision studies where each test's outcome is recorded.
Comparison Studies: The individual operators' results (positive/negative) were directly compared to the LC/MS results. Discordant results are individually listed against the LC/MS result. There is no mention of an adjudication process among the operators.
Lay-user Study: The lay users' obtained results (positive/negative) were compared against the LC/MS confirmed drug concentration of the sample they tested. No adjudication is mentioned.

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 MRMC comparative effectiveness study was done, as this device is a standalone in-vitro diagnostic test kit (Strip/Panel), not an AI-assisted diagnostic tool for human readers. Therefore, there is no AI component, and no improvement effect size for human readers.

6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done

Yes, standalone performance was evaluated by human observers interpreting the test results visually.
- The "Comparison Studies" section evaluates the device's performance by having six operators visually interpret the results of the strips/panels and compare them against LC/MS ground truth. This effectively determines the device's accuracy in a standalone setting (device + human interpretation) for laboratory professional users.
- The "Lay-user study" also evaluates the standalone performance by 280 lay persons visually interpreting the results.

7. The type of ground truth used

The primary ground truth used for establishing drug concentrations and confirming results was LC/MS (Liquid Chromatography-Mass Spectrometry). This is considered a highly definitive analytical method for confirming drug presence and concentration.

8. The sample size for the training set

The document does not mention a "training set" in the context of machine learning or AI. This is a traditional in-vitro diagnostic device (immunoassay) and therefore does not involve machine learning models that require training data. All samples described are used for performance validation and testing.

9. How the ground truth for the training set was established

As there is no mention of a "training set" for a machine learning model, this question is not applicable to the described device. The samples used for performance evaluation (precision, comparison, lay-user studies) had their ground truth established by LC/MS confirmation of spiked drug concentrations or by direct LC/MS reference for unaltered urine samples.

Ask a Question

Ask a specific question about this device

K Number

K192301

Device Name

BIOEASY Marijuana Test Dip Card, BIOEASY Marijuana Test Strip

Manufacturer

Shenzhen Bioeasy Biotechnology Co.,Ltd.

Date Cleared

2019-09-20

(28 days)

Product Code

Regulation Number

Type

Panel

Reference & Predicate Devices

K182530

Intended Use

BIOEASY Marijuana Test Dip Card is competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 50 ng/mL.

BIOEASY Marijuana Test Strip is competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 50 ng/mL.

Device Description

The BIOEASY Marijuana Test Dip Card and the BIOEASY Marijuana Test Strip tests are immunochromatographic assays that use a lateral flow system for the qualitative detection of Marijuana in human urine. The products are single-use in vitro diagnostic devices. Each test kit contains a Test Device and a package insert. Each test device is sealed with a desiccant in an aluminum pouch.

AI/ML Overview

This document describes the performance characteristics and acceptance criteria for the BIOEASY Marijuana Test Dip Card and BIOEASY Marijuana Test Strip.

1. Table of Acceptance Criteria and Reported Device Performance

The acceptance criteria for qualitative immunoassay devices like these are typically based on achieving a certain level of agreement with a reference method (LC/MS in this case) across different concentration ranges relative to the cutoff, and demonstrating reliable performance under various conditions.

Since the document does not explicitly state "acceptance criteria" numerical targets, these are inferred from the presented data, aiming for high agreement, especially at concentrations away from the cutoff, and demonstrating appropriate performance near the cutoff.

Performance Characteristic	Acceptance Criteria (Inferred)	Reported Device Performance (BIOEASY Marijuana Test Dip Card / Strip)
Precision	High agreement (ideally 100% Negative or 100% Positive) for samples far from the cutoff (-100%, -75%, -50% cut off; +50%, +75%, +100% cut off). Acceptable distribution of results (both positive and negative) for samples at the cutoff (50 ng/mL), reflecting the qualitative nature and inherent variability around the cutoff.	Dip Card:
-100% to -25% Cutoff: 100% Negative (200/200 total per lot).
+25% to +100% Cutoff: 100% Positive (200/200 total per lot).
At Cutoff: Mixture of Positive/Negative (25-/25+, 23-/27+, 24-/26+ for Lot 1, 2, 3 respectively).
Strip:
-100% to -25% Cutoff: 100% Negative (200/200 total per lot).
+25% to +100% Cutoff: 100% Positive (200/200 total per lot).
At Cutoff: Mixture of Positive/Negative (19-/31+, 28-/22+, 23-/27+ for Lot 1, 2, 3 respectively).
Specificity	No interference from common physiological and pathological substances found in urine. Limited or no cross-reactivity with structurally similar compounds at high concentrations, or clear explanation of cross-reactivity at specified concentrations.	Interference: No interference observed at specified concentrations for a wide range of common substances (e.g., albumin, ethanol, acetaminophen, ibuprofen).
Cross-Reactivity:
100% to 11-nor-Δ9-THC-9-COOH, 11-Hydroxy-Δ9-Tetrahydrocannabinol, 11-Nor-Δ8-Tetrahydrocannabinol-9-COOH, (-)-11-nor-9-carboxy-Δ9-THC at 50 ng/mL.
Lower cross-reactivity for Cannabinol (0.25%), Δ8-Tetrahydrocannabinol (0.33%), Δ9-Tetrahydrocannabinol (0.33%) at significantly higher concentrations (15000-20000 ng/mL).
Cannabidiol (100000 ng/mL).
11-Nor-Δ9-THC-carboxy glucuronide (66.7% at 75 ng/mL).
Effect of Urine SG & pH	Reliable performance (correct result relative to cutoff) across a range of urine specific gravity (SG) and pH values.	Specific Gravity: Results were all positive for samples at and above +25% Cut-Off and all negative for samples at and below -25% Cut-Off for SG range of 1.000 to 1.035.
pH: Results were all positive for samples at and above +25% Cut-Off and all negative for samples at and below -25% Cut-Off for pH range of 4 to 9.
Method Comparison (Analytical Accuracy)	High agreement with LC/MS (gold standard) for samples categorized as "Negative," "Low Negative," "High Positive." A reasonable number of discordant results are expected at "Near Cutoff" concentrations, reflecting the inherent variability of qualitative assays around the cutoff.	Strip Format:
Negative, Low Negative, High Positive LC/MS categories show high agreement across viewers (e.g., Viewer A: 6/6 Negatives, 14/14 Low Negatives, 18/18 High Positives).
Near Cutoff Negative and Positive categories show some discordance (e.g., Viewer A: 1 Positive out of 20 Near Cutoff Negatives, 1 Negative out of 22 Near Cutoff Positives).
Dip Card Format: Similar high agreement for clear negative/positive samples and expected discordance near cutoff.
Lay-User Study Performance	High percentage of correct results when used by lay persons, demonstrating ease of use and interpretability of results as per the Instructions For Use (IFU). Acceptable performance around the cutoff. User surveys confirm ease of understanding IFU.	Strip & Dip Card:
-100%, -75%, -50% Cutoff: 100% correct results.
-25% Cutoff: 95% (Strip) and 90% (Dip Card) correct results (1-2 false positives out of 20).
+25% Cutoff: 90% correct results (2 false negatives out of 20).
+50%, +75% Cutoff: 100% correct results.
Lay-user surveys: All users indicated instructions were easily followed (Grade Level 7 Flesch-Kincaid).
Stability	Devices remain stable and perform within acceptance criteria over the claimed shelf-life under specified storage conditions.	Stable at 4-30 ℃ for 24 months based on accelerated stability at 45 ℃ and real-time stability at 4 ℃ and 30 ℃.

2. Sample Sizes Used for the Test Set and Data Provenance

Precision Studies:
- For each of the 8 concentration points (ranging from -100% of cutoff to +100% of cutoff) relative to the 50 ng/mL cutoff, 50 tests were performed per concentration point for each of the 3 lots.
- Total tests for Precision: 8 concentrations x 50 tests/concentration x 3 lots = 1200 tests per device format (Dip Card and Strip).
- Data Provenance: Samples were "prepared by spiking 11-Nor-△9-THC-9-COOH in negative samples." The document does not specify the country of origin but implies laboratory-prepared samples. It's a prospective study on prepared samples.
Interference Studies:
- Tested at 25% below and 25% above cutoff with various interfering substances.
- "Three batches of each device" were used. The number of samples for each interfering substance is not explicitly stated but implied to be sufficient for testing.
Specificity Studies (Cross-Reactivity):
- Various drug metabolites and other components were tested.
- "Three batches of each device format" were used. The number of tests per compound is not explicitly stated.
Effect of Urine Specific Gravity and pH:
- Tested at 25% below and 25% above cutoff with varying SG and pH.
- "Three lots of each device format" were used. The number of samples per condition is not explicitly stated.
Method Comparison Studies (Analytical Accuracy):
- 80 "unaltered clinical samples" were used for each device format (40 negative and 40 positive).
- Data Provenance: "Clinical samples." No country of origin is specified. The study is retrospective in the sense that samples were collected and then tested.
Lay-User Study:
- 280 lay persons participated.
- 7 concentration points were tested (negative, +/-75%, +/-50%, +/-25% of the cutoff).
- For each concentration point, 20 samples were tested.
- Total tests: 7 concentrations x 20 samples/concentration = 140 samples per device format.
- Data Provenance: Urine samples were "prepared by spiking 11-Nor-△9-THC-9-COOH into drug free-pooled urine specimens." The location of the lay-user study is described as "three intended user sites," but no specific country is mentioned. This is a prospective study on prepared samples with lay users.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts

Analytical Ground Truth:
- For the precision, interference, specificity, and lay-user studies, the ground truth for spiked samples was established by LC/MS (Liquid Chromatography/Mass Spectrometry), which is a highly accurate and specific analytical method.
- For the method comparison study, "LC/MS results" were used as the gold standard for classifying clinical samples.
- No human experts were explicitly "establishing" ground truth in terms of diagnostic interpretation; rather, the ground truth was based on the objective, quantitative results of LC/MS.

4. Adjudication Method for the Test Set

Analytical Studies (Precision, Interference, Specificity, SG/pH): No adjudication method is mentioned or needed as the results are quantitative and read as positive/negative based on the device's qualitative nature. The preparation and blinding for precision studies suggest a robust method.
Method Comparison Study: The document states, "Operators ran 80 (40 negative and 40 positive) unaltered clinical samples. The samples were blind labeled and compared to LC/MS results." "Viewer A, B, C" for the strip and "Viewer D, E, F" for the dip card were the "laboratory assistants." The data are presented per viewer, implying individual results were compared to LC/MS. There is no mention of an adjudication process for discordant results among the viewers or between the device results and LC/MS. The discordant results are simply listed.
Lay-User Study: No adjudication. The results are presented as the "No. of Positive" and "No. of Negative" given by the lay persons, compared to the known spiked concentration verified by LC/MS.

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 document describes the validation of a qualitative in-vitro diagnostic device (Marijuana Test Dip Card/Strip) for detecting Marijuana in urine directly by human observation of lines on a test strip/card. There is no AI component involved, so no study on AI assistance or human reader improvement with AI can be conducted or reported.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) Was Done

Not applicable. This device is a manual, visually read immunoassay. There is no algorithm or automated reading component. The "standalone" performance is simply the device's ability to produce a correct result when used as intended and read by a human. The "Method Comparison Studies" and "Precision Studies" essentially demonstrate the standalone performance of the device itself (though still read by human operators).

7. The Type of Ground Truth Used

The primary ground truth used for all performance studies (precision, method comparison, lay-user study) was analytical confirmation by LC/MS (Liquid Chromatography/Mass Spectrometry). This is considered a gold standard for quantitative drug concentration measurement.
For interference and specificity, the ground truth was based on the known presence/absence and concentration of the spiked substances.

8. The Sample Size for the Training Set

This document describes performance validation studies for a medical device (Marijuana Test Dip Card/Strip). These types of devices are not typically "trained" in the machine learning sense. Therefore, there is no "training set" in the context of an AI/ML model for this device. The data presented are for demonstrating the device's analytical and clinical performance.

9. How the Ground Truth for the Training Set Was Established

As there is no AI/ML component and thus no "training set," this question is not applicable. The ground truth for the test samples (used for validation) was established using LC/MS.

Ask a Question

Ask a specific question about this device

K Number

K191924

Device Name

SAFECARE® THC Urine Strip Test

Manufacturer

Safecare Biotech (Hangzhou) Co., Ltd.

Date Cleared

2019-08-16

(29 days)

Product Code

Regulation Number

Type

Panel

Reference & Predicate Devices

N/A

Intended Use

SAFECARE® THC Urine Strip Test is a competitive binding, lateral flow immunochromatographic assay for qualitative detection of Marijuana in human urine at the cutoff concentrations of 50 ng/mL.

The test provides only preliminary test results. A more specific alternative chemical must be used in order to obtain a confirmed analytical result. GC/MS or LC/MS is the preferred confirmatory method.

The test is intended for over-the-counter use.

Device Description

SAFECARE® THC Urine Strip Test devices are immunochromatographic assays for the qualitative detection of 11-nor-A9-THC-9 COOH (target analyte) in human urine. The product is a single-use in vitro diagnostic device. It contains a Test Device and a package insert. Each test device is sealed with a desiccant in an aluminum pouch.

AI/ML Overview

The provided document outlines the acceptance criteria and performance data for the SAFECARE® THC Urine Strip Test, which is a qualitative lateral flow immunoassay for detecting Marijuana in human urine.

Here's the breakdown of the requested information:

1. Table of Acceptance Criteria and Reported Device Performance

The document does not explicitly state formal "acceptance criteria" for precision or accuracy using specific numerical thresholds (e.g., >95% agreement). Instead, it presents the results of various performance studies. The overall acceptance is inferred from the device being deemed "substantially equivalent" to a predicate device.

However, based on the performance characteristics, we can infer performance goals for accuracy at different concentrations relative to the cutoff.

Acceptance Criteria Category	Specific Metric (Inferred)	Acceptance Threshold (Inferred/Observed)	Reported Device Performance	Comments on Performance
Analytical Performance
Precision	Agreement at various	High agreement expected, especially
	concentrations	at concentrations further from cutoff.
	-100% Cut-off	100% Negative (0% Positive)	Lot 1, 2, 3: 50-/0+	Meets expectation
	-75% Cut-off	100% Negative (0% Positive)	Lot 1, 2, 3: 50-/0+	Meets expectation
	-50% Cut-off	100% Negative (0% Positive)	Lot 1, 2, 3: 50-/0+	Meets expectation
	-25% Cut-off	100% Negative (0% Positive)	Lot 1, 2, 3: 50-/0+	Meets expectation
	Cut-off (50 ng/mL)	~50% Positive, ~50% Negative	Lot 1, 2: 25-/25+; Lot 3:	Meets expectation
			26-/24+
	+25% Cut-off	100% Positive (0% Negative)	Lot 1, 2, 3: 50+/0-	Meets expectation
	+50% Cut-off	100% Positive (0% Negative)	Lot 1, 2, 3: 50+/0-	Meets expectation
	+75% Cut-off	100% Positive (0% Negative)	Lot 1, 2, 3: 50+/0-	Meets expectation
	+100% Cut-off	100% Positive (0% Negative)	Lot 1, 2, 3: 50+/0-	Meets expectation
Interference	No interference from	No false positives or negatives due to	Compounds at 100µg/mL	No interference reported
	common substances	specified interfering substances when	(list provided) showed no
		drug is absent or present above cutoff	interference.
Specificity	Cross-reactivity with	Low or no cross-reactivity with	Provided specific cross-	Demonstrated specificity
	related compounds	non-target compounds.	reactivity percentages for
			related cannabinoids.
Urine Specific Gravity	Correct results across	Correct classification (negative for	Samples from 1.000 to 1.035	Device performs reliably
	range of SG	-25% cutoff, positive for +25% cutoff).	with THC at +/-25% cutoff	across range of urine SG
			showed expected results.
Urine pH	Correct results across	Correct classification (negative for	Samples from pH 4 to 9 with	Device performs reliably
	range of pH	-25% cutoff, positive for +25% cutoff).	THC at +/-25% cutoff	across range of urine pH
			showed expected results.
Comparison Studies	Agreement with LC/MS	High agreement between device results
	for clinical samples	and LC/MS, particularly for samples
		not near the cutoff.
	Overall Agreement	Not explicitly quantified, but inferred	Viewer A: 78/80 (97.5%)	High agreement for
		from discordant results.	Viewer B: 78/80 (97.5%)	human readers
			Viewer C: 79/80 (98.75%)
Lay-User Study	Ease of use	Instructions are easily understood.	All lay users indicated	Instructions are clear
			instructions were easy to
	Accuracy of self-testing	High percentage of correct results,
		decreasing near the cutoff.
	-100% Cutoff	100% agreement	100% Correct	Excellent
	-75% Cutoff	100% agreement	100% Correct	Excellent
	-50% Cutoff	100% agreement	100% Correct	Excellent
	-25% Cutoff	>90% agreement, with expected	90% Correct	Acceptable, slight dip
		increase in discordance.		near cutoff.
	+25% Cutoff	>90% agreement, with expected	95% Correct	Acceptable, slight dip
		increase in discordance.		near cutoff.
	+50% Cutoff	100% agreement	100% Correct	Excellent
	+75% Cutoff	100% agreement	100% Correct	Excellent

2. Sample Size Used for the Test Set and Data Provenance

Precision Study:
- Sample Size: 9 different concentrations, with 50 tests performed for each concentration per lot (2 runs per day for 25 days). Since 3 lots were tested, this amounts to 9 concentrations * 50 tests/lot * 3 lots = 1350 tests in total for the precision study itself. The samples were prepared by spiking drug in "negative samples." The provenance is not explicitly stated but implies laboratory-prepared samples.
Comparison Studies (Clinical Samples):
- Sample Size: 80 unaltered clinical samples (40 negative and 40 positive).
- Data Provenance: The document states "unaltered clinical samples." The country of origin for these clinical samples is not specified. It is a retrospective analysis as the samples were collected and then tested.
Lay-User Study:
- Sample Size: 140 lay persons tested individual samples. There were 7 concentration levels, with 20 samples per concentration. So, 7 concentrations * 20 samples/concentration = 140 samples in total.
- Data Provenance: The samples were "spiked drug THC into drug free-pooled urine specimens." The country of origin is not specified. It is a prospective study in the sense that the lay users tested the devices with prepared samples, but the samples themselves were laboratory-prepared.

3. Number of Experts Used to Establish the Ground Truth for the Test Set and Their Qualifications

Precision Study: The ground truth was established by laboratory preparation of samples with known concentrations confirmed by LC/MS. No human experts are explicitly mentioned for ground truth.
Comparison Studies (Clinical Samples): The ground truth was established by LC/MS results. "LC/MS is the preferred confirmatory method." The number of LC/MS operators or analysts is not specified, nor are their qualifications.
Lay-User Study: The ground truth for the prepared urine samples was established by LC/MS. The number of LC/MS operators or analysts is not specified, nor are their qualifications.

4. Adjudication Method for the Test Set

Precision Study: Not applicable, as results were quantitative (known concentrations) rather than subjective interpretations needing adjudication.
Comparison Studies (Clinical Samples): The document mentions "three different laboratory assistants" performing the tests. Their results were compared to LC/MS. There is no explicit adjudication method stated for discrepancies between human readers or between human readers and LC/MS. The discordant results table simply lists them.
Lay-User Study: Not explicitly stated. Each lay person provided their own result. Their individual results were then compared to the LC/MS confirmed concentration.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done, and 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 device is a standalone diagnostic strip test intended for visual interpretation, not an AI-assisted diagnostic tool. Therefore, there is no AI component or human readers improving with/without AI assistance.

6. If a Standalone (i.e. algorithm only without human-in-the-loop performance) Was Done

Yes, in essence, the "device performance" in all studies represents a standalone performance relative to the ground truth (LC/MS). The device itself (the strip test) produces a visual result. While a human reads the result, the performance studies assess the accuracy of the device's output (presence/absence of a line) against known concentrations.
For the "Comparison Studies," the performance of the "Safecare THC Urine Strip Test" is measured for each viewer by comparing their reading of the device against LC/MS. This measures the combined device-human performance.
For the "Lay-User Study," it also measures the combined device-lay user performance.

7. The Type of Ground Truth Used

For all performance studies (Precision, Comparison, Lay-User), the ground truth was primarily established by LC/MS (Liquid Chromatography-Mass Spectrometry). This is a highly accurate and quantitative laboratory method.
For the precision study, it specifies "THC drug concentration was confirmed by LC/MS."
For the comparison studies, it states "The samples were blind labeled and compared to LC/MS results. LC/MS is the preferred confirmatory method."
For the lay-user study, it states "The concentrations of the samples were confirmed by LC/MS."

8. The Sample Size for the Training Set

Based on the provided document, this is a 510(k) submission for a diagnostic device. Such submissions typically focus on analytical and clinical performance validation rather than explicitly detailing a "training set" for an algorithm.
There is no information provided regarding a "training set" as this is not an AI/machine learning device. The immunoassay device relies on chemical reactions, not on data training.

9. How the Ground Truth for the Training Set Was Established

As there is no mention of a "training set" or an algorithm that requires training, this question is not applicable based on the provided document.

Ask a Question

Ask a specific question about this device

Page 1 of 1