Clinical patient samples; Clinical diagnosis records
The clinical performance study evaluated the device's sensitivity and specificity in differentiating Inflammatory Bowel Disease (IBD) from Irritable Bowel Syndrome (IBS) using patient stool samples and established clinical diagnoses.
Clinical performance; Diagnostic accuracy; IBD vs IBS differentiation
Clinical Evidence
Study Design
Population
Comparator
Key Endpoints
Retrospective clinical performance study
Patients evaluated for IBD and IBS; Sample Size: 138
Clinical diagnosis (IBD vs. Negative/IBS)
Sensitivity, Specificity, PPV, NPV
Indications for Use
Calprest® is a quantitative ELISA for detecting concentration of fecal calprotectin. Calprest® can be used as an in vitro diagnostic to aid in the diagnosis of Inflammatory Bowel Diseases (IBD, Crohn's disease and ulcerative colitis) and to differentiate IBD from Irritable Bowel Syndrome (IBS) in conjunction with other clinical and laboratory findings.
Device Story
Calprest® is an enzyme-linked immunosorbent assay (ELISA) system for quantitative detection of fecal calprotectin. The device uses a 96-well microtiter plate coated with polyclonal antibodies to capture calprotectin from diluted stool samples. An enzyme-conjugated antibody binds to the captured antigen, followed by a substrate reaction that produces a color change. The intensity of the color, measured at 405 nm, is proportional to the calprotectin concentration, which is calculated against provided standards. The test is performed in a laboratory setting by trained personnel. Results are used by clinicians in conjunction with other clinical and laboratory findings to aid in the diagnosis of Inflammatory Bowel Diseases (IBD) and to differentiate IBD from Irritable Bowel Syndrome (IBS).
Clinical Evidence
Clinical study of 138 samples (98 IBD, 40 non-IBD). When borderline (50–120 mg/kg) is considered abnormal: sensitivity 96.9% (95% CI: 91.3–99.4%), specificity 85.0% (95% CI: 70.2–94.3%). When borderline is considered normal: sensitivity 79.6% (95% CI: 70.3–87.1%), specificity 92.5% (95% CI: 79.6–98.4%). Method comparison with predicate (N=131) showed Deming regression y = 0.98x – 1.85.
Technological Characteristics
ELISA system using 96-well microtiter plates. Detection is colorimetric at 405 nm using alkaline phosphatase conjugate and pNPP substrate. Quantitative assay for fecal calprotectin. Requires sample dilution (1:2500) and specific incubation times (45-45-30 minutes at room temperature).
Indications for Use
Indicated for patients suspected of having Inflammatory Bowel Disease (IBD) or Irritable Bowel Syndrome (IBS) to aid in diagnosis and differentiation. Used as an in vitro diagnostic test for fecal calprotectin concentration.
Regulatory Classification
Identification
A fecal calprotectin immunological test system is an in vitro diagnostic device that consists of reagents used to quantitatively measure, by immunochemical techniques, fecal calprotectin in human stool specimens. The device is intended for in vitro diagnostic use as an aid in the diagnosis of inflammatory bowel diseases (IBD), specifically Crohn's disease and ulcerative colitis, and as an aid in differentiation of IBD from irritable bowel syndrome.
Special Controls
The device is classified as Class II under regulation 21 CFR 866.5180 with special controls. The special control guidance document, "Class II Special Controls Guidance Document: Fecal Calprotectin Immunological Test Systems" will be available shortly.
*Classification.* Class II (special controls). The special control for these devices is FDA's guidance document entitled “Class II Special Controls Guidance Document: Fecal Calprotectin Immunological Test Systems.” For the availability of this guidance document, see § 866.1(e).
Predicate Devices
PhiCal™ Test (k050007)
Submission Summary (Full Text)
{0}
1
# 510(k) SUBSTANTIAL EQUIVALENCE DETERMINATION DECISION SUMMARY
A. 510(k) Number:
k130945
B. Purpose for Submission:
New Device
C. Measurand:
Fecal calprotectin
D. Type of Test:
Quantitative, ELISA
E. Applicant:
Eurospital S.p.A.
F. Proprietary and Established Names:
Calprest®
G. Regulatory Information:
1. Regulation section:
21 CFR§866.5180 – Fecal calprotectin immunological test system
2. Classification:
Class II
3. Product code:
NXO, Calprotectin, Fecal
4. Panel:
Immunology (82)
{1}
H. Intended Use:
1. Intended use(s):
Calprest® is a quantitative ELISA for detecting concentration of fecal calprotectin. Calprest® can be used as an in vitro diagnostic to aid in the diagnosis of Inflammatory Bowel Diseases (IBD, Crohn's disease and ulcerative colitis) and to differentiate IBD from Irritable Bowel Syndrome (IBS) in conjunction with other clinical and laboratory findings.
2. Indication(s) for use:
Same as Intended Use
3. Special conditions for use statement(s):
Prescription use only
4. Special instrument requirements:
Microtiter plate reader (405 nm filter)
I. Device Description:
The Calprest® kit contains the following materials:
- Microtiter plate coated with rabbit anti-calprotectin antibodies (12 strips, 8 wells/strip)
- Alkaline phosphatase labeled rabbit anti-calprotectin antibody (1x15 mL)
- Substrate (1x15 mL)
- 20x Washing solution (1x50 mL)
- 10x Diluent solution (1x20 mL)
- 2.5x Extraction solution (2x50 mL)
- Calibrators (6x1.0 mL, 6 vial of calprotectin solution at concentration of 6.25, 12.5, 25, 50, 100, and 200 ng/mL)
- Control 1 (low) (1x1.0 mL, Approximate range = 20 – 40 ng/mL or 50 – 100 mg/kg)
- Control 2 (high) (1x1.0 mL, Approximate range = 40 – 80 ng/mL or 100 – 200 mg/kg)
J. Substantial Equivalence Information:
1. Predicate device name(s) and 510(K) number(s):
PhiCal™ Test (k050007)
2. Comparison with predicate:
{2}
| Similarities | | |
| --- | --- | --- |
| Item | Device Calprest® | Predicate PhiCal™ Test |
| Intended Use/Indication for Use | to aid in the diagnosis of Inflammatory Bowel Diseases (IBD, Crohn's disease and ulcerative colitis) and to differentiate IBD from Irritable Bowel Syndrome (IBS) in conjunction with other clinical and laboratory findings. | to aid in the diagnosis of inflammatory bowel diseases (IBD): Crohn's disease and ulcerative colitis, and to differentiate IBD from IBS; when used in conjunction with other diagnostic testing and the total clinical picture. |
| Analyte | Calprotectin | Same |
| Assay format | Quantitative | Same |
| Method | Colorimetric ELISA | Same |
| Sample type | fecal | Same |
| Specimen Requirement | 1.0 – 5.0 g stool | Same |
| Instrument | ELISA reader (405 nm) | Same |
| Capture Antibody | Rabbit anti-human calprotectin | Same |
| Detection Antibody | Alkaline phosphatase labeled rabbit anti-calprotectin | Same |
| Substrate | pNPP | Same |
| Control | 2 (high and low) | Same |
| Stability | Open vial: Conjugate: 2-8°C, 30 days Substrate: 2-8°C, 90 days Calibrators: 2-8°C, 30 days Controls: 2-8°C, 30 days | Same |
| Sample Transportation | Stool specimen should be received by the laboratory within 10 days of collection. Temperature during shipment should not exceed 37°C. Sample must be extracted within 10 days of collection. | Same |
| Differences | | |
| --- | --- | --- |
| Item | Device Calprest® | Predicate PhiCal™ Test |
| Calibrator | 6 levels: (6.25, 12.5, 25, 50, 100, and 200 ng/mL) | 5 levels: (6.25, 12.5, 25, 50, and 100 ng/mL) |
| Measuring range | 15.6 – 500 mg/kg | 15.6 – 250 mg/kg |
| Results | Normal: <15.6 – 50 mg/kg | Normal: <15.6 – 50 mg/kg |
{3}
| Differences | | |
| --- | --- | --- |
| Item | Device Calprest® | Predicate PhiCal™ Test |
| interpretation | Borderline: 50 – 100 mg/kg
Abnormal: >100 mg/kg | Borderline: 50 – 120 mg/kg
Abnormal: >120 mg/kg |
| Stability (Working solution) | Washing buffer: 20-25°C, 30 days
Dilution buffer: 2-8°C, 30 days | Washing buffer: 20-25°C, 7 days
Dilution buffer: 2-8°C, 7 days |
| Sample storage | Stored at 2-8°C for up to 4 days before testing. If not tested immediately, freeze stored samples at -20°C. | Samples should be stored at 2-8°C for up to 11 days, or for 1 year at –20°C. |
## K. Standard/Guidance Document Referenced (if applicable):
CLSI EP05-A2 “Evaluation of Precision Performance of Quantitative Measurement Methods; Approved Guideline-Second Edition”
CLSI EP06-A “Evaluation of the Linearity of Quantitative Measurement Procedures: A Statistical Approach; Approved Guideline”
CLSI EP7-A2 “Interference Testing in Clinical Chemistry; Approved Guideline – Second Edition”
CLSI EP17-A “Protocols for Determination of Limits of Detection and Limits of Quantitation; Approved Guideline”
Guidance for Industry and FDA Staff – Class II Special Controls Guidance Document: Fecal Calprotectin Immunological Test Systems
## L. Test Principle:
The test is performed on stool samples, collected without preservatives. The samples should be tested within 4 days, or frozen at -20°C if not immediately tested. An extract is prepared by combining the stool sample with extraction buffer at a ratio of 1:50 (e.g., 0.1 gm of stool added with 4.9 mL of extraction buffer) and mixing for 30 minutes. Following centrifugation, 20μL of the supernatant is diluted 1:50 with dilution buffer (final dilution 1:2500).
Calprest® is a colorimetric enzyme-linked immunosorbent assay (ELISA). The assay uses a polyclonal rabbit antibody against calprotectin as the capture antibody. Calprotectin present in the diluted sample is bound by the antibody adsorbed onto the surface of the microtiter plate. Alkaline phosphatase conjugated antibodies (rabbit IgG antibodies) bind to the captured calprotectin. The enzyme catalyzes the conversion of the substrate to a colored product and the optical density (OD) of the sample is read on an ELISA plate reader. The intensity of the color is proportional to the amount of conjugate bound, and thus to the amount of captured calprotectin. The concentration of calprotectin in the extracted sample is
{4}
interpreted from a standard curve generated from the 6 calibrators and converted to mg of calprotectin per kg of stool using a conversion of factor supplied by package insert.
## M. Performance Characteristics (if/when applicable):
### 1. Analytical performance:
#### a. Precision/Reproducibility:
Intra-assay precision was conducted by extracting 8 different stool samples. Each extract sample was tested with 10 replicates in a single assay run. The mean, standard deviation (SD) and the coefficient of variation (CV) were calculated. The results met the acceptance criteria with the %CV of each sample <15%.
| | #1 | #2 | #3 | #4 | #5 | #6 | #7 | #8 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Mean (mg/kg) | 438.7 | 355.2 | 352.9 | 311.7 | 199.9 | 125.7 | 57.8 | 27.7 |
| SD (mg/kg) | 33.7 | 43.0 | 43.7 | 38.8 | 6.6 | 9.8 | 2.9 | 1.0 |
| % CV | 7.7 | 12.1 | 12.4 | 12.4 | 3.3 | 7.8 | 5.0 | 3.7 |
Inter-assay precision was conducted by extracting 8 different stool samples. Each extract was tested using one lot of the test kit with 5 replicates per run, 1 run per day for 6 days with 2 operators. The mean, standard deviation (SD) and the coefficient of variation (CV) were calculated. The results met the acceptance criteria with %CV of each sample <15%.
| | #1 | #2 | #3 | #4 | #5 | #6 | #7 | #8 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Mean (mg/kg) | 359.9 | 209.8 | 185.5 | 117.0 | 70.4 | 48.8 | 40.9 | 20.6 |
| SD (mg/kg) | 27.5 | 17.8 | 17.8 | 9.3 | 6.8 | 4.5 | 3.4 | 2.6 |
| % CV | 7.7 | 8.5 | 9.6 | 8.0 | 9.6 | 9.3 | 8.2 | 12.4 |
Lot-to-lot reproducibility was done by extracting six different stool samples. Each sample was tested in 5 replicates using 3 different lots of the test kit. The results are shown in the following table.
| | #1 | #2 | #3 | #4 | #5 | #6 |
| --- | --- | --- | --- | --- | --- | --- |
| Mean (mg/kg) | 107.7 | 63.1 | 185.1 | 43.8 | 36.0 | 344.9 |
| SD (mg/kg) | 5.78 | 3.68 | 14.55 | 1.36 | 3.00 | 42.62 |
| % CV | 5.4 | 5.8 | 7.9 | 3.1 | 8.3 | 12.4 |
Extraction reproducibility was evaluated by using 3 stool samples with calprotectin concentrations of 28.4, 41.1 and 79.4 mg/kg. Each stool sample was extracted 10 times and each stool extract was tested in duplicate. The %CVs for the 3 stool samples are 13.6%, 8.9% and 7.0%.
#### b. Linearity/assay reportable range:
{5}
Linearity: The reportable range of the Calprest® was evaluated with both aqueous and matrix linearity. For the matrix linearity study, serially diluted samples with calprotectin concentrations ranging from 20.4 to 492.8 mg/kg were prepared by diluting the high positive extracted stool sample pool in the low concentration extracted stool sample pool. Each dilution was tested in triplicate. The percent (%) recovery ranged from 87.0% to 113.2% for tested samples. The linear regression analysis gives the following equation:
$$
y = 1.013x - 3.777, R^2 = 0.999
$$
For the aqueous linearity study, samples were prepared by diluting the calprotectin reference materials with sample diluent to generate 7 dilutions with concentrations ranging from 8.6 mg/kg to 550 mg/kg. Each dilution was tested in triplicate. The % recovery for samples ranged from 95.2% to 109.3%. The linear regression analysis gives the following equation:
$$
y = 0.985x + 2.714, R^2 = 0.999
$$
The assay was shown to give a linear response over the claimed reportable range (15.6 – 500 mg/kg).
Accuracy/Recovery study: Extracts from 7 stool samples were each spiked with the kit calibrator material (S3) that has a calprotectin concentration target of 25 ng/mL. The baseline extract for each sample was "spiked" with sample diluent to compensate for volume adjustments made to the calprotectin-spiked extracts. Each of the spiked extracts was then assayed in triplicate and results are shown in table below.
| Sample | #1 | #2 | #3 | #4 | #5 | #6 | #7 |
| --- | --- | --- | --- | --- | --- | --- | --- |
| Baseline (mg/kg) | 18.3 | 47.5 | 59.4 | 66.5 | 115.2 | 232.5 | 421.9 |
| Spike Value (mg/kg) | 31.2 | 31.2 | 31.2 | 31.2 | 31.2 | 31.2 | 31.2 |
| Theoretical (Base+Spike) (mg/kg) | 49.6 | 78.8 | 90.6 | 97.8 | 146.4 | 263.7 | 452.1 |
| Observed (Base+Spike) (mg/kg) | 49.8 | 81.1 | 100.1 | 108.7 | 166.0 | 271.3 | 475.0 |
| % Recovery | 100.5 | 103.0 | 110.5 | 111.1 | 113.4 | 102.8 | 112.6 |
c. Traceability, Stability, Expected values (controls, calibrators, or methods):
Traceability: There is no international reference material for calprotectin. The calibrators and controls are manufactured using an approved reference lot which is traceable to the internal reference material cleared in k050007.
The calibrator S6, prepared from the master stock solution, was used to make the calibrators and controls. New lots of the calibrator set and controls were value assigned and validated by a comparison against an approved lot of calibrators and controls in three different assay runs and further testing with a panel of 12 internal quality control samples. The target values for calibrators are 6.25, 12.5, 25, 50, 100 and 200 ng/mL, and the target values for controls range from 20 to 40 ng/mL for the
{6}
Low Control and 40 to 80 ng/mL for the High Control.
## Stability:
**Kit stability:** The real time stability was performed using three lots of Calprest® kits (including calibrators and controls) stored under the recommended temperature at 2 – 8°C. Data were collected at point 0, 6, 12, 18 and 19 months. The results support stability of the kits under the recommended storage of 2 – 8°C for up to 18 months.
**Open vial stability:** The study was done to evaluate the reagent stability after opening. Three kit lots of Calprest® were stored at 2 – 8°C after first opening. Data were collected at 0, 30 and 60 days. The results support that the reagents are stable once opened for up to 30 days when stored at 2 – 8°C.
**Working solutions stability:** The study was done to evaluate the stability of working solutions including dilution buffer, washing buffer and extraction solution. The 1X working solution for each buffer was prepared and stored at 2 – 8°C. Data were collected at different time points. The results support up to 30 days stability for 1X dilution buffer and extraction buffer and 3 months stability for 1X washing buffer.
**Sample stability:** Five (5) stool samples which cover the measuring range of Calprest® were stored at 2-8°C for 2, 4, and 7 days. The acceptance criteria for % recovery of calprotectin in the samples tested at each time point should be within ±5% of the initial reading taken at day 0. The results support that stability of stool sample stored at 2-8°C up to 4 days.
**Analyte (Calprotectin) stability:** The study was done to determine the stability of extracted stool samples when stored -20°C. A total of 10 stool samples (1 with concentration above 100 mg/kg, 2 around 100 mg/kg, 5 around 50 mg/kg, and 2 around 20 mg/kg) were tested. Data were collected at time 0 and 3 months. The acceptance criteria for % recovery of calprotectin at 3 months should be within ±15% of the initial reading. All samples met the acceptance criteria.
## d. Detection limit:
The limit of blank (LoB) was determined by assaying 5 blank samples 12 times. The LoB value was estimated to be 3.04 mg/kg. The limit of detection (LoD) was determined by assaying 5 extracted stool samples with low calprotectin level 12 times. The LoD value was calculated as the LoB + 1.645 × SD of the replicates for the low level samples and was found to be 3.98 mg/kg.
The lower limit of the reportable assay range for Calprest® is 15.6 mg/kg which corresponds to the lowest calibrator with a concentration of 6.25 ng/mL after conversion to mg of calprotectin per kg stool sample according to the assay's instructions for use.
7
{7}
# e. Analytical specificity:
Interference studies were performed according to EP7-A2 using 5 pooled stool samples, each containing a mixture of 3 different stool samples. The pooled samples included 2 high positives (with calprotectin concentrations around $460\mathrm{mg / kg}$ and 270 $\mathrm{mg / kg}$ ), 1 low positive $(145\mathrm{mg / kg})$ , 1 sample within borderline range $(68\mathrm{mg / kg})$ and 1 negative $(30\mathrm{mg / kg})$ . For non-interference to be claimed, the $\%$ recovery of the spiked sample should be within $95 - 105\%$ of the neat sample. Stool samples were tested for potential interference by:
Microorganisms: The data demonstrated that Calprest® was not affected by Escherichia coli, Klebsiella pneumoniae, Salmonella spp., Shigella spp., and Yersinia spp. (at cell count of $1.5 \times 10^{7}$ cfu/ml) in stool samples.
Drug and Nutrients: The data demonstrated that Calprest® was not affected by the following oral pharmaceuticals and nutritional supplements: Vancomycin (0.67 mg/50 mg stool); Ciprofloxacin HCL (0.50 mg/50 mg stool); Prevacid (0.02 mg/50 mg stool); Azathioprine (0.07 mg/50 mg stool); Prednisone (0.01 mg/50 mg stool); Pentasa (1.33 mg/50 mg stool); Asacol (1.33 mg/50 mg stool); multiple vitamin (Vit D: 1.1 UI/50 mg stool, Vit A: 8.0 UI/50 mg stool), Vit C: 0.05 mg/50 mg stool) and Vitamin E (0.10 mg/50 mg stool).
Hemoglobin: The results showed that the addition of hemoglobin into the stool samples at level of $5.83\mathrm{mg} / 50\mathrm{mg}$ stool did not interfere with Calprest®.
# f. Assay cut-off:
The assay cut-off for Calprest® is as follows:
| Calprotectin (mg/kg) | Interpretation | Suggested follow-up |
| --- | --- | --- |
| <15.6 – 50 | Normal | None |
| 50 – 120 | Borderline | Re-evaluate at 4-6 weeks |
| >120 | Abnormal | Repeat as clinically indicated. |
# 2. Comparison studies:
# a. Method comparison with predicate device:
A total of 131 clinical samples including 70 samples from patients diagnosed with Crohn's disease, Crohn's disease in remission, ulcerative colitis (UC), UC in remission, Diverticulitis, or intermediate colitis (IC), and 61 samples from patients with irritable bowel syndrome (IBS), anemia, chronic diarrhea, recurrent abdominal pain (RAP), celiac disease, and other conditions were assayed using Calprest® and the predicate device PhiCal in accordance with the instructions for use in the package inserts. The values of the samples tested covered the measuring range of both devices. Deming regression analyses are based on the balance of the paired results,
{8}
and the data are as follows:
| N= | Sample range (Calprest®) (mg/kg) | Regression analysis |
| --- | --- | --- |
| 131 | 15.6 – 266.50 | y = 0.98x – 1.85
Slope (95% CI): 0.96 – 1.01
Intercept (95% CI): -3.96 – 0.96 |
b. Matrix comparison:
Not applicable
3. Clinical studies:
a. Clinical Sensitivity and specificity
A total of 138 samples were tested with Calprest® assay: 98 were from patients diagnosed with Crohn's disease, UC and intermediate colitis (IC), and 40 were from patients with IBS, chronic diarrhea, RAP and celiac disease. The IBD patients were diagnosed by clinical findings and/or confirmed with colonoscopy.
The results are summarized in the following tables:
| | Clinical Diagnosis of IBD | | | |
| --- | --- | --- | --- | --- |
| | | Positive | Negative | Total |
| Calprest® | Abnormal (>120 mg/kg) | 85 | 3 | 88 |
| | Borderline (50 – 120 mg/kg) | 10 | 3 | 13 |
| | Normal (<50 mg/kg) | 3 | 34 | 37 |
| | Total | 98 | 40 | 138 |
| Borderline considered as abnormal:
Sensitivity: 96.9% (95% CI: 91.3 – 99.4%)
Specificity: 85.0% (95% CI: 70.2 – 94.3%)
PPV: 94.1% (95% CI: 87.5 – 97.8%)
NPV: 91.9% (95% CI: 78.1 – 98.3%) | | | | |
| Borderline considered as normal:
Sensitivity: 79.6% (95%CI: 70.3 – 87.1%)
Specificity: 92.5% (95% CI: 79.6 – 98.4%)
PPV: 96.3% (95% CI: 89.6 – 99.2%)
NPV: 64.9% (95% CI: 51.1 – 77.1%) | | | | |
b. Other clinical supportive data (when a. and b. are not applicable):
Not applicable
{9}
4. Clinical cut-off:
Not applicable.
5. Expected values/Reference range:
The expected value in the normal healthy population is generally <50 mg/kg according to study done by Røseth et al. (1992)*.
Ninety (90) samples from normal healthy people and 42 samples from patients with CD, UC, IBS, and IC were tested with Calprest® to validate the reference range value. The results showed that 85 out of 90 samples (92.2%) from normal healthy people tested negative and the remaining 5 samples fell in the range of 51.8 – 60.4 mg/kg. Of the 42 samples from patients with CD, UC, IBS and IC, 36 (85.7%) tested positive by the assay.
* Røseth AG, Fagerhol MK, Aadland E, Schonsby H. (1992). Assessment of the neutrophil-dominating protein calprotectin in feces. A methodology study. Scand J. Gastroenterol. 27(9): 793-798.
N. Proposed Labeling:
The labeling is sufficient and it satisfies the requirements of 21 CFR Part 809.10.
O. Conclusion:
The submitted information in this premarket notification is complete and supports a substantial equivalence decision.
10
Predicate graph will load when search results are available.
Embedding visualization will load when search results are available.
PDF viewer will load when search results are available.
Loading panels...
Select an item from Submissions
Click any panel, subpart, regulation, product code, or device to see details here.
Section Matches
Results will appear here.
Product Code Matches
Results will appear here.
Special Control Matches
Results will appear here.
Loading collections...
Loading
My Alerts
You will receive email notifications based on the filters and frequency you set for each alert.
Sort by:
Create Alert
Search Filters
Agent Token
Create a read-only bearer token for Claude, ChatGPT, or other agents that can call HTTP APIs.
Copy this now. It will not be shown again.
Connected apps
Apps you authorized through browser sign-in. Disconnecting revokes their access immediately.
Learn the FDA Browser
Two short videos show you everything — or skip straight to the written tutorial if you'd rather read. You can reopen this any time from the Tutorial button in the top bar.
Part 1 — Search, results, and everyday workflows 16 min
Part 2 — Embeddings: the galaxy map 3 min
1. Search: exact and fuzzy
Type a phrase like "coronary artery calcification" into the search box. You get two kinds of results. Exact results match the literal phrase — prefix searches work ("coronary artery calcificati") but suffix searches do not. Fuzzy results match on the meaning and intent of your phrase rather than the exact words, and are sorted by relevance score. Hover over the Exact or Fuzzy badge on any row to see exactly why it matched.
Use the checkboxes above the results to narrow: SaMD keeps only software-only devices, AI / ML keeps only devices with AI.
Exact vs. fuzzy search: what's the difference?
Exact matches on the literal phrase (prefix search works, suffix does not). Fuzzy matches on the meaning and intent of the phrase rather than the exact words. Hover over the badge on any row to see why it matched.
You search "coronary artery calcification" and want only software devices with AI. What two filters do you apply?
Narrow by SaMD (software-only devices), then narrow by AI/ML (devices with AI).
2. The results table
Scroll right in the results table. The intended use is extracted for you — no need to open the PDF. The device story gives a high-level snapshot of what the device does and how it's used. The AI Performance sub-table shows each output name, acceptance criteria, observed values, and development/test dataset descriptions — the same format Innolitics uses for regulatory strategy outputs, and the fastest high-level fingerprint of an AI device. It is AI-generated but has been very reliable in practice.
Where do you find a device's intended use without opening the PDF?
Scroll right in the search results table. The intended use column is extracted for you; no need to dig into the 510(k) summary PDF.
What does the AI Performance sub-table show, and why is it useful?
Output name, acceptance criteria, observed values, development dataset description, and test dataset description. It's the same format we use for regulatory strategy output and Fast 510(k) input, and the fastest high-level fingerprint of an AI device. AI-generated but reliable in practice.
3. Judging fuzzy relevance
Fuzzy results trail off in relevance as you scroll. Use three signals to decide how far down to go: the fuzzy badge explanations, the intended use column, and whether your target output (e.g., Cobb angle) still appears in the AI Performance sub-table. Once it stops appearing, you're past the relevant zone. A top hit with a low score (~0.4) and a stretched explanation is a hint the closest predicates are far away — the project may be headed for De Novo. Note the fuzzy search is a pattern match: it doesn't handle negation ("not") well, and hardware devices can appear — filter by SaMD/AI ML to cut them.
How do you judge how far down fuzzy search results to go?
Use the relevancy signals: the fuzzy badge explanations, the intended use column, and whether the target output (e.g., Cobb angle) still appears in the AI Performance sub-table. Once it stops appearing, results are trailing off in relevancy.
4. Device detail page: chat and citations
Click a device name to open its detail page: device facts on the left, a chat window on the right. Ask something like "Describe the training data". The answer carries little citation bubbles — click one to jump to the highlighted passage in the source PDF, so you can verify every AI answer against the document. There's also a Download PDF button for sharing.
How do you verify an AI chat answer on the device detail page?
Click the citation bubbles to jump to the relevant highlight in the source document.
Reading rule for every project: how many summaries do you read in full?
At least the three most relevant 510(k) or De Novo summaries, in full. After that, use targeted chat questions to confirm your memory quickly. The tool supports this professional habit — it doesn't replace it.
5. Side-by-side comparison
Select multiple rows in the results table (aim for under ~10), then open the PDF Viewer tab. Ask one question — it goes to all selected devices in parallel, each with citations. This is the fastest way to compare and contrast devices: training data, PCCP scope, how they handled adding new scanners, and so on.
What does the side-by-side PDF viewer mode do?
Select multiple devices, open the PDF viewer tab, and ask one question (e.g., "Describe the training data"). It queries all selected devices simultaneously with citations, so you can compare and contrast quickly.
6. Collections
With rows selected, go to the Collections tab and create a labeled collection (e.g., "Cobb Angle Project"). Reload that selection any time — before a client call, pull up the collection and ask questions across all of its devices at once.
How do you save a set of selected devices for later use?
Select the rows, go to the Collections tab, and create a labeled collection (e.g., "Cobb Angle Project"). You can reload the selection anytime and carry it into the PDF viewer and other tabs that support selections.
7. Product codes and the regulations tree
Click a product code in the results to jump to it in the regulations tree — identification text, sibling product codes, and devices you can open in a PDF viewer on the right. Click a regulation number to see its identification, special controls, and related product codes. You can also search by product code or regulation number at the top of the tree. Always read the special controls if any exist for your device — it broadens your search and sharpens pre-kickoff research.
What can you do from the regulations tree view?
Browse product codes and regulation numbers, read the identification text and special controls, browse sibling product codes, open device PDFs on the right, and search by product code or regulation number at the top of the tree.
8. Chart view
Click Show Chart and segment by regulation number (or product code) to see which regulations dominate your result set. Clicking a regulation takes you into the regulations tree. Great for spotting that most matches are, say, hardware laparoscopic devices — a cue to go back and filter.
How do you see which regulations dominate a search result set?
Click "Show Chart" and segment by Regulation Number. Clicking a regulation takes you to the regulations tree.
9. The predicate graph
Open the Predicates tab for a family-tree view of predicate relationships. Click a node to trace its parents and children; selections from search carry over pre-selected. Commonly predicated devices are worth reading — a lot of people predicated them for a reason. The visual lineage is also handy on client calls, e.g. to show how a predicate family evolved and justify why your predicate still holds.
In the predicate graph, why are commonly predicated devices worth reading?
A lot of people predicated them for a reason. Clicking a node traces parents and children, and selections from search carry over pre-selected.
10. Embeddings: the galaxy map
The Embeddings tab plots every matching document in a 2-D "galaxy map" where semantically similar devices cluster together. Hover or click clusters to explore, and let AI label the clusters for you. Embeddings beat product codes for grouping: two devices can carry different product codes (LLZ vs. QIH) yet do the same thing — the embedding captures the meaning of the intended use and device story. This is also exactly how retrieval-augmented generation (RAG) works under the hood, and it makes a great visual on client calls.
Try it yourself
Head to the search page and work through a few of these AI/ML fuzzy searches to build intuition: perivascular fat on CT · aortic valve calcification opportunistic screening on noncontrast CT · breast cancer prediction on digital pathology slides · autism detection · gestational age prediction · a hearing aid that can also detect a pulse · foundation model based analysis of ECG · large language models · penetration test. Watch how the relevance scores, intended use, and AI Performance tables tell you when results stop being meaningful.