FDA 510(k) Search - By Innolitics (SaMD and AI Experts)

HemosIL ReadiPlasTin is an in vitro diagnostic thromboplastin reagent, based on recombinant human tissue factor, for the quantitative determination, in human citrated plasma, of Prothrombin Time (PT) and Fibrinogen, on the ACL TOP Family of analyzers. The product is intended to be used for the evaluation of the extrinsic coagulation pathway and the monitoring of Oral Vitamin K Antagonist Therapy.

Device Description

The thromboplastin reagent included in the ReadiPlasTin kit is a liquid liposomal preparation that contains recombinant human tissue factor (RTF) in a synthetic phospholipid blend combined with calcium chloride, buffer and a preservative. In the PT test, the addition of tissue thromboplastin to the patient plasma, in the presence of calcium ions, initiates the activation of the extrinsic pathway. This results in the conversion of fibrinogen to fibrin, with the formation of a solid gel. The fibrinogen is then quantitated by relating the absorbance to a calibrator.

ReadiPlasTin can be used with II's Normal control, Routine controls 1-3, Low abnormal control, high abnormal control, ISI calibrate and INR validate.

AI/ML Overview

Here's a breakdown of the requested information based on the provided text for the HemosIL ReadiPlasTin device:

1. Table of Acceptance Criteria and Reported Device Performance

Performance Characteristic	Acceptance Criteria (Implicit)	Reported Device Performance
Precision (PT - Total %CV)	Not explicitly stated but expected to be low for clinical use.	Normal Control: 1.1 - 1.4% CV
		Low Abnormal: 1.2 - 2.9% CV
		High Abnormal: 1.4 - 1.9% CV
Precision (Fibrinogen - Total %CV)	Not explicitly stated but expected to be low for clinical use.	Normal Control: 2.0 - 2.2% CV
		Low Abnormal: 1.8 - 2.4% CV
		Low Fibrinogen: 3.1 - 3.5% CV
Fibrinogen Linearity	r ≥ 0.95	r ≥ 0.95 (met)
Fibrinogen Analytical Range	Not explicitly stated but supported by linearity.	60-700 mg/dL
Extrinsic Factor Linearity (II, V, VII, X)	r ≥ 0.95	r ≥ 0.95 (met)
Interfering Substances (PT)	Insensitive up to specified levels.	Heparin: 1.0 IU/mL
		LMW: 1.4 IU/mL
		Hemoglobin: 500 mg/dL
		Triglycerides: 1000 mg/dL
		Bilirubin: 50 mg/dL
Interfering Substances (Fibrinogen)	Insensitive up to specified levels.	Heparin: 1.5 IU/mL
		LMW: 1.7 IU/mL
		Hemoglobin: 500 mg/dL
		Triglycerides: 600 mg/dL
		Bilirubin: 50 mg/dL
Method Comparison (PT sec)	High correlation and close agreement with predicate.	ACL TOP 700: Slope 0.913, r 0.9989
		ACL TOP 500: Slope 0.903, r 0.9981
		Field Site 1: Slope 0.909, r 0.9966
		Field Site 2: Slope 0.970, r 0.9972
		Field Site 3: Slope 0.889, r 0.9955
Method Comparison (PT INR)	High correlation and close agreement with predicate.	ACL TOP 700: Slope 0.979, r 0.9975
		ACL TOP 500: Slope 0.937, r 0.9955
		Field Site 1: Slope 0.938, r 0.9947
		Field Site 2: Slope 0.928, r 0.9838
		Field Site 3: Slope 0.914, r 0.9945
Method Comparison (Fibrinogen)	High correlation and close agreement with predicate.	ACL TOP 700: Slope 0.946, r 0.9955
		ACL TOP 500: Slope 0.975, r 0.9945
		Field Site 1: Slope 0.947, r 0.9964
		Field Site 2: Slope 0.971, r 0.9956
		Field Site 3: Slope 0.998, r 0.9931
Fresh vs. Frozen Samples	Comparable performance expected.	Demonstrated comparable performance.
Normal Range Study	Not explicitly stated as acceptance criteria, but a reference range was established.	PT (sec): 10.2 - 12.9 sec
		Fibrinogen (mg/dL): 282 - 553 mg/dL

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

The text describes several studies, each with specific sample sizes and implicit provenance:

Precision/Reproducibility:
- Sample Size: 3 lots of reagent, 3 instruments (2 ACL TOP 700, 1 ACL TOP 500 CTS), 3 operators. For each control level (Normal, Low Abnormal, High Abnormal for PT; Normal, Low Abnormal, Low Fibrinogen for Fibrinogen), N=80 per instrument per lot (20 days, 2 runs/day, 2 replicates/run), totaling 240 measurements per control level per lot across all instruments.
- Data Provenance: Implied to be prospective laboratory testing conducted by the manufacturer ("utilizing 3 lots of reagent on 3 representative members of the ACL TOP Family by 3 independent operators"). Country of origin is not explicitly stated but assumed to be where the company is based (Bedford, MA, USA).
Fibrinogen Linearity:
- Sample Size: Not explicitly stated how many samples were used, but the study was performed in accordance with CLSI EP6-A.
- Data Provenance: Implied to be retrospective (prepared linearity samples) or prospective laboratory testing. Country of origin not specified.
Extrinsic Factor Linearity:
- Sample Size: Not explicitly stated how many samples were used for factors II, V, VII, and X.
- Data Provenance: Implied retrospective (prepared factor-deficient plasmas) or prospective laboratory testing. Country of origin not specified.
Analytical Specificity (Interfering Substances):
- Sample Size: Not explicitly stated how many samples were tested for each interfering substance, but the study was performed according to CLSI EP7-A.
- Data Provenance: Implied retrospective (spiked samples) or prospective laboratory testing. Country of origin not specified.
Method Comparison with Predicate Device:
- Sample Size:
  - In-house: "100+ samples (Normal & Abnormal)". Specifically, n=230 for PT(sec) on ACL TOP 700, n=65 for PT (INR) on ACL TOP 700, n=241 for Fibrinogen on ACL TOP 700. Similar numbers for ACL TOP 500 CTS (n=238 for PT(sec), n=65 for PT (INR), n=242 for Fibrinogen).
  - US Field Studies: Site 1: 254 for PT(sec), 76 for PT(INR), 237 for Fibrinogen. Site 2: 312 for PT(sec), 89 for PT(INR), 284 for Fibrinogen. Site 3: 135 for PT(sec), 71 for PT(INR), 129 for Fibrinogen.
- Data Provenance: The "in-house method comparison" is likely prospective or retrospective from internal lab samples. The "US field studies" are prospective patient samples from clinical sites in the US.
Matrix Comparison (Fresh vs. Frozen):
- Sample Size: n=245 (140 normal, 105 abnormal patient samples).
- Data Provenance: Prospective patient samples. Country of origin not specified, but likely associated with the manufacturer or field sites.
Normal Range Study:
- Sample Size: n=199 healthy subjects.
- Data Provenance: Prospective collection from healthy subjects. Country of origin not specified, but likely associated with the manufacturer or field sites.

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

No "experts" in the traditional sense (e.g., radiologists interpreting images) are mentioned in this submission. This is a submission for an in vitro diagnostic reagent and system, where the "ground truth" is typically established by reference methods, accuracy against known standards, or comparison with a cleared predicate device.
The "predicate device" (HemosIL RecombiPlasTin 2G) served as the primary comparator for establishing substantial equivalence, meaning its results were considered the "reference" or "ground truth" for comparison in the method comparison studies. The predicate device itself would have undergone its own rigorous validation.
For the Normal Range Study, the "ground truth" was simply the biochemical values obtained from a healthy population.

4. Adjudication Method for the Test Set

No adjudication method is described. This type of IVD submission does not typically involve human adjudication of results in the way image analysis or clinical endpoint studies might. The results are quantitative (PT in seconds/INR, Fibrinogen in mg/dL), and performance is assessed statistically by comparing the device's numerical outputs against predicate device outputs or established reference values.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done

No, an MRMC comparative effectiveness study was not done. This type of study is primarily relevant for diagnostic imaging interpretation or other scenarios where multiple human readers are tasked with making a subjective assessment, and the AI's impact on their performance is being evaluated. This submission is for an automated in-vitro diagnostic assay.

6. If a Standalone (Algorithm Only Without Human-in-the-Loop Performance) Was Done

Yes, the primary performance studies are standalone algorithm/device performance. This device, the HemosIL ReadiPlasTin reagent on the ACL TOP Family analyzers, is an automated system. Its performance (precision, linearity, method comparison, etc.) is measured directly by the instrument, without human intervention in the result generation process itself. Humans operate the instrument and interpret the results, but the analytical performance described is "standalone" in the sense that it reflects the device's inherent capability to produce accurate measurements.

7. The Type of Ground Truth Used

The "ground truth" for the performance studies primarily used:
- Predicate Device Data: For the method comparison studies, the results from the HemosIL RecombiPlasTin 2G served as the comparative "truth" to demonstrate substantial equivalence.
- Reference Standards/Calibrators: Traceability is described against "ISI certified House Standard" which is calibrated against "international reference preparation according to the WHO recommendations" for PT/INR, indicating a form of metrological traceability for quantitative accuracy.
- Known Reference Samples: Clinical samples (normal and abnormal patient samples) were used to cover the operating range.
- Statistical Methodology: For precision, linearity, and interference studies, performance metrics were compared against statistical acceptance criteria (e.g., %CV, correlation coefficient r) which are based on established clinical laboratory standards (CLSI guidelines).
- Healthy Population Values: For the normal range study, the "ground truth" was derived from a statistically significant sample of healthy individuals.

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. This is a traditional in-vitro diagnostic device (a reagent) used with an existing analyzer. The device itself is not an AI algorithm that "learns" from data.
However, the traceability and calibration processes for the reagent involve internal "House Standards" which are themselves developed and validated using extensive data. Each batch of ReadiPlasTin is calibrated against these standards. While not a "training set" in the AI sense, this represents the historical data and standards used to ensure accurate performance of each reagent lot.

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

As noted above, there isn't a "training set" in an AI/ML context. The calibration and standardization process serves a similar function:
- House Standards: These internal reference materials are "calibrated against the corresponding international reference preparation according to the WHO recommendations." This means that the ultimate "ground truth" for the device's accuracy is established through international consensus standards and reference measurement procedures, which are the gold standard for many IVD assays.
- The "consensus values from > 200 laboratories" for INR reference values also indicate a broadly established and verified "ground truth" derived from a large number of expert laboratories.

Summary

Regulation Number and Section

§ 864.7750 Prothrombin time test.

(a)
Identification. A prothrombin time test is a device used as a general screening procedure for the detection of possible clotting factor deficiencies in the extrinsic coagulation pathway, which involves the reaction between coagulation factors III and VII, and to monitor patients receiving coumarin therapy (the administration of one of the coumarin anticoagulants in the treatment of venous thrombosis or pulmonary embolism).(b)
Classification. Class II (performance standards).