K252818 · Dexcom, Inc. · QRX · Nov 18, 2025 · Clinical Chemistry
Device Facts
Record ID
K252818
Device Name
Dexcom Smart Basal
Applicant
Dexcom, Inc.
Product Code
QRX · Clinical Chemistry
Decision Date
Nov 18, 2025
Decision
SESE
Submission Type
Traditional
Regulation
21 CFR 862.1358
Device Class
Class 2
Attributes
Software as a Medical Device, Therapeutic
Indications for Use
The Dexcom Smart Basal insulin dose calculator is software intended for the management of type 2 diabetes in persons aged 18 years and older requiring long-acting insulin. This is not for use for patients on fast-acting insulin. The Dexcom Smart Basal insulin dose calculator calculates a dose of basal (long-acting) insulin using logged doses and glucose measurements from an integrated continuous glucose monitor (iCGM). The Dexcom Smart Basal insulin dose calculator requires a prescription and initial configuration by a healthcare provider.
Device Story
Dexcom Smart Basal is a cloud-based software system for type 2 diabetes management; requires prescription and HCP configuration. Inputs: iCGM glucose data, patient-logged basal insulin doses, and HCP-defined treatment parameters. Operation: algorithmic software processes inputs to generate daily basal insulin dose recommendations and titration episode status. Used by patients via mobile app and by providers via web interface. Output: basal dose recommendation; informs patient titration. Benefits: optimizes basal insulin dosing; improves time-in-range (TIR) for glycemic control.
Clinical Evidence
Prospective, multi-center, single-arm study (N=14) in adults with type 2 diabetes. Primary endpoint: change in TIR (70-180 mg/dL) over 35 days vs. 10-day baseline. Results: Mean TIR increased from 30.8% to 51.5% (p=0.0017). No severe hypoglycemia, DKA, or HHS reported. 96.5% of 401 system-generated dose recommendations were accepted by HCPs without modification. In-silico testing validated algorithm robustness against edge cases and CGM sensor bias (10-day vs. 15-day G7).
Technological Characteristics
Cloud-based software system; server-side titration algorithm. Components: Therapeutic Services (server), Therapeutic/Host Portals (web), Host App (mobile). Connectivity: Networked/Cloud-hosted. Software: Major level of concern; complies with IEC 62304. Cybersecurity: Complies with FDA guidance. Compatible with glargine U-100 insulin and Dexcom G7 CGM.
Indications for Use
Indicated for management of type 2 diabetes in adults aged 18+ requiring long-acting insulin. Contraindicated for patients on fast-acting insulin.
Regulatory Classification
Identification
An insulin therapy adjustment device is a device intended to incorporate biological inputs, including glucose measurement data from a continuous glucose monitor, to recommend insulin therapy adjustments as an aid in optimizing insulin therapy regimens for patients with diabetes mellitus.
Special Controls
(b) Classification: Class II (special controls). The special controls for this device are
- 1. Design verification and validation must include the following:
- (i) A complete description of the required data inputs, including timeframe over which data inputs must be collected and number of data points required for accurate recommendations.
- (ii) A complete description of the types of device outputs and insulin therapy adjustment recommendations, including how the recommendations are generated.
- (iii)Robust data demonstrating the clinical validity of the device outputs and insulin therapy recommendations.
- (iv)A robust assessment of all input data specifications, including accuracy requirements for continuous glucose monitors and other devices generating data inputs, to ensure accurate and reliable therapy adjustment recommendations. This assessment must include adequate clinical justification for each specification.
- (v) A detailed strategy to ensure secure and reliable means of data transmission to and from the device, including data integrity checks, accuracy checks, reliability checks, and security measures.
- (vi)Robust data demonstrating that users can understand and appropriately interpret recommendations generated by the device.
- (vii) An appropriate mitigation strategy to minimize the occurrence of dosing recommendation errors, and to mitigate the risk to patients of any residual dosing recommendation errors to a clinically acceptable level.
- 2. The device must not be intended for use in implementing automated insulin dosing.
- 3. Your 809.10(b) labeling must include:
- (i) The identification of specific insulin formulations that have been demonstrated to be compatible with use of the device.
- (ii) A detailed description of the specifications of compatible devices that provide acceptable input data (e.g., continuous glucose monitors, insulin pumps) used to provide accurate and reliable therapy adjustment recommendations.
- (iii) A detailed description of all types of required data (inputs) and dosing recommendations (outputs) that are provided by the device.
- (iv) A description of device limitations, and instructions to prevent possible disruption of accurate therapy adjustment recommendations (e.g., time zone changes due to travel).
*Classification.* Class II (special controls). The special controls for this device are:(1) Design verification and validation must include the following:
(i) A complete description of the required data inputs, including timeframe over which data inputs must be collected and number of data points required for accurate recommendations;
(ii) A complete description of the types of device outputs and insulin therapy adjustment recommendations, including how the recommendations are generated;
(iii) Robust data demonstrating the clinical validity of the device outputs and insulin therapy recommendations;
(iv) A robust assessment of all input data specifications, including accuracy requirements for continuous glucose monitors and other devices generating data inputs, to ensure accurate and reliable therapy adjustment recommendations. This assessment must include adequate clinical justification for each specification;
(v) A detailed strategy to ensure secure and reliable means of data transmission to and from the device, including data integrity checks, accuracy checks, reliability checks, and security measures;
(vi) Robust data demonstrating that users can understand and appropriately interpret recommendations generated by the device; and
(vii) An appropriate mitigation strategy to minimize the occurrence of dosing recommendation errors, and to mitigate the risk to patients of any residual dosing recommendation errors to a clinically acceptable level.
(2) The device must not be intended for use in implementing automated insulin dosing.
(3) Your 21 CFR 809.10(b) labeling must include:
(i) The identification of specific insulin formulations that have been demonstrated to be compatible with use of the device;
(ii) A detailed description of the specifications of compatible devices that provide acceptable input data (e.g., continuous glucose monitors, insulin pumps) used to provide accurate and reliable therapy adjustment recommendations;
(iii) A detailed description of all types of required data (inputs) and dosing recommendations (outputs) that are provided by the device; and
(iv) A description of device limitations, and instructions to prevent possible disruption of accurate therapy adjustment recommendations (e.g., time zone changes due to travel).
{0}
FDA
U.S. FOOD & DRUG
ADMINISTRATION
# 510(k) SUBSTANTIAL EQUIVALENCE DETERMINATION DECISION SUMMARY
## I Background Information:
A 510(k) Number
K252818
B Applicant
Dexcom, Inc.
C Proprietary and Established Names
Dexcom Smart Basal
D Regulatory Information
| Product Code(s) | Classification | Regulation Section | Panel |
| --- | --- | --- | --- |
| QRX | Class II | 21 CFR 862.1358 – Insulin Therapy Adjustment Device | CH – Clinical Chemistry |
A Purpose for Submission:
New Device
## II Intended Use/Indications for Use:
A Intended Use(s):
See Indications for Use below.
B Indication(s) for Use:
The Dexcom Smart Basal insulin dose calculator is software intended for the management of type 2 diabetes in persons aged 18 years and older requiring long-acting insulin. This is not for use for patients on fast-acting insulin. The Dexcom Smart Basal insulin dose calculator calculates a dose of basal (long-acting) insulin using logged doses and glucose measurements from an integrated continuous glucose monitor (iCGM). The Dexcom Smart Basal insulin dose calculator requires a prescription and initial configuration by a healthcare provider.
C Special Conditions for Use Statement(s):
Food and Drug Administration
10903 New Hampshire Avenue
Silver Spring, MD 20993-0002
www.fda.gov
{1}
Rx – For Prescription Use Only
The Dexcom Smart Basal system is not intended to be used with automated insulin dosing (AID) systems or fast-acting insulin.
The Dexcom Smart Basal system is compatible with glargine U-100 insulin only. Only use glargine U-100, as it is the only insulin type tested and confirmed to be safe for use with the Dexcom Smart Basal system.
The Dexcom Smart Basal system is not recommended for use during pregnancy. The Dexcom Smart Basal system has not been tested with individuals who are pregnant.
## III Device Description
The Dexcom Smart Basal System is a basal insulin (long-acting insulin) dose calculator intended for use by individuals with type 2 diabetes. It requires a prescription and initial configuration by a healthcare provider (HCP). The Smart Basal System is a software system designed to optimize and recommend basal insulin doses for Dexcom CGM App Users (patients) using their CGM data, insulin dose logs, and HCP-entered patient-specific treatment parameters. The Dexcom Smart Basal System is comprised of applications that initiate and manage the basal titration process for Dexcom CGM App Users with Type 2 diabetes who may benefit from long-term insulin treatment.
The Smart Basal System consists of cloud-based software components that function as a CGM-informed basal insulin dose calculator (subject device). The Smart Basal feature can be accessed via a module within a compatible CGM System mobile application for the patient to interact with (referred to as Host App), and a web-based user interface for the HCP (referred to as Host Portal).
The Dexcom Smart Basal System contains the following subject software components: Therapeutic Services and Therapeutic Portal. In addition, the Dexcom Smart Basal System utilizes a compatible Host App and Host Portal to interact with the patient and HCP respectively.
Therapeutic Portal and Host Portal are web-based applications. The Host Portal is a web-based application that directly interfaces with HCPs. The Therapeutic Portal is a server-based application that supports the creation, storage, and communication of the patient's treatment plan. Using the Therapeutic Portal (via Host Portal), HCPs create a basal treatment plan for patients that include patient-specific basal parameters, such as "initial dose" and "maximum dose" thresholds, and initiates the treatment for the particular patient (called an "episode"). Therapeutic Portal communicates directly with the Host App to initialize the system and to communicate dose recommendations generated by Therapeutic Services.
The Host App is the mobile app that hosts the basal therapy optimization module that interfaces with the patient. A compatible Dexcom CGM mobile app functions as the Host App for the Smart Basal System. The mobile app collects Estimated Glucose Values (EGVs) from the wearable CGM sensor and transmits EGV data to Therapeutic Services. Values for patient-administered basal insulin doses entered in the Host App are communicated to Therapeutic Services as well.
K252818 - Page 2 of 6
{2}
Therapeutic Service is a server-based software component that includes the algorithm for basal insulin dose recommendations. Using the EGV values and insulin dose logs from the Host App, together with the patient's basal treatment plan from the Host Portal (via Therapeutic Portal), Therapeutic Services then generates a daily dose recommendation. This daily dose recommendation is sent to the Host Portal (via Therapeutic Portal) and the Host App (also via Therapeutic Portal) which the patient consults to administer and log the next dose. The daily basal insulin dose recommendation continues until one of the following criteria is met: 1) Therapeutic Services determines the basal optimization is complete; 2) the HCP stops the patient's treatment using the Therapeutic Portal (via Host Portal), 3) the patient stops the treatment via the Host App, or 4) the days in a titration episode for the patient reaches a fixed time limit.
## IV Substantial Equivalence Information:
### A Predicate Device Name(s):
BlueStar CGM insulin dose calculator
### B Predicate 510(k) Number(s):
K222888
### C Comparison with Predicate(s):
| Device & Predicate Device(s): | K252818 | K222888 |
| --- | --- | --- |
| Device Trade Name | Dexcom Smart Basal System | BlueStar CGM insulin dose calculator |
| General Device Characteristic Similarities | | |
| Intended Use/Indications For Use | Intended to calculate insulin doses based on CGM values and/or other relevant information | Same |
| Principle of operation | Algorithmic software device | Same |
| Age range of intended users | 18 years and older | Same |
| Display to user | Calculated doses via mobile application and use a web-based portal for healthcare providers | Same |
K252818 - Page 3 of 6
{3}
| General Device Characteristic Differences | | |
| --- | --- | --- |
| Insulin type | Long-acting insulin | Fast-acting insulin |
| Diabetes type | Type 2 diabetes | Type 1 or Type 2 diabetes |
| Data inputs | CGM values and historic data, initial configuration from HCP, daily basal insulin log | Same and also carbohydrate information |
| Titration episode length | Up to 90 days, unless terminated by the HCP or patient | Continuously as long as the patient is using the same bolus insulin |
V Standards/Guidance Documents Referenced:
ANSI/AAMI/ISO 14971:2019 Medical devices – Applications of risk management to medical devices
ANSI/AAMI/IEC 62304:2006/A1:2016 Medical device software – Software life cycle processes [Including Amendment 1 (2016)]
ANSI/AAMI/IEC 62366-1:2015 + AMDI:2020 (Consolidated Text) Medical devices Part 1: Application of usability engineering to medical devices including Amendment 1
ANSI/AAMI HE75:2009/(R)2018 Human factors engineering – Design of medical devices
VI Performance Characteristics:
Clinical Testing
A prospective, multi-center, single-arm study was conducted at two sites in the US to evaluate the safety and effectiveness of the Dexcom Smart Basal System in adult patients with type 2 diabetes. The study enrolled 14 participants aged 18 years and older who were initiating or optimizing glargine U-100 insulin therapy.
The study was conducted in four phases: Phase 1 involved a 10-day run-in period with CGM training and baseline data collection. Phase 2 consisted of 10-35 days of active titration where HCPs reviewed daily algorithm-generated dose recommendations and communicated them to participants. Phase 3 involved 20 days of continued dosing with the final recommended dose while collecting outcome data. Phase 4 was a single day for post-titration assessments.
Of note, as the system used in the clinical study was an early version of the submitted device, DexBasal, there was no user-facing interface at that time. Instead, the HCP would add the subject to the web portal and specify the treatment plan. Each day, the clinical staff used the HCP web
K252818 - Page 4 of 6
{4}
portal to review the system's recommended dose for glargine U-100, modify the dose if necessary, and communicate the dose to participants. Participants were instructed to log the provided dose into the Dexcom G7 mobile application.
Among 401 total dose recommendations generated by the system, HCPs accepted 387 (96.5%) without modification. Fourteen recommendations (3.5%) were adjusted, with 13 of these adjustments due to a temporary API outage that prevented the system from receiving necessary input data, and one adjustment due to a patient logging error (double basal logged) that triggered a 0-unit recommendation that the HCP overrode
The primary objective of the study was to evaluate the safety and effectiveness of the CGM-informed dose calculator for recommending doses of basal insulin using CGM metrics of time in range (TIR) of 70-180 mg/dL during the study compared to the 10-day run-in period. The clinical study CGM metric results are presented in the table below. Of the 14 subjects, over 35 days, 5 subjects reached "completion" with a safe and effective dose (positive termination of the device), 4 were estimated to be within 5 days of completion, 2 hit their HCP's maximum dose limit, 1 hit the algorithm maximum dose limit (1 u/kg), and 2 may or may not have reached completion. There were no reported adverse events due to use of the device. There was also no severe hypoglycemia, DKA, or HHS events. There was 1 psychiatric hospitalization unrelated to use of the device.
Summary of 10-day G7 CGM Metrics – All subjects (N = 14)
| Range (mg/dL) | % Time, Baseline, Mean (SD) | % Time, Post-Titration, Mean (SD) | % Difference, Mean (SD) | % Difference, Median (Q1, Q3) | P-value* |
| --- | --- | --- | --- | --- | --- |
| 70-180 | 30.8 (24.4) | 51.5 (22.1) | 20.7 (23.0) | 16.0 (2.5, 43.7) | 0.0017 |
| < 54 | 0.0 (0.0) | 0.0 (0.0) | 0.0 (0.0) | 0.0 (0.0, 0.0) | - |
| < 70 | 0.0 (0.1) | 0.1 (0.2) | 0.1 (0.2) | 0.0 (0.0, 0.1) | - |
| > 180 | 69.2 (24.5) | 48.4 (22.2) | -20.8 (23.1) | -15.9 (-43.7, -2.5) | 0.0017 |
| > 250 | 34.5 (33.8) | 18.1 (22.7) | -16.4 (26.4) | -8.4 (-20.2, 0.6) | 0.0085 |
*Wilcoxon Signed-Ranks Test was used
For the 3 subjects who initiated basal insulin during this study, average TIR increased from 16.8% to 65.7%, average time above 180 mg/dL decreased from 83.2% to 34%, and the average time below 70 mg/dL was unchanged (0.0%) during the study. For all subjects, overnight (12:00 am – 6:00 am) CGM metrics showed an average TIR increase from 43.7% to 67.0%, average time above 180 mg/dL decreased from 56.2% to 32.7%, and the average time below 70 mg/dL was unchanged (0.1%).
In-Silico Testing
K252818 - Page 5 of 6
{5}
After the clinical study, the sponsor made changes to the dosing algorithm to enhance device performance and robustness, particularly in edge/patient use error cases. The changes were tested in an *in silico* clinical trial re-run validation test to assess how changes made to the algorithm used in the clinical study impact the recommended doses compared to the recommended doses in the clinical study algorithm. Changes made to the clinical study algorithm between use in the clinical study and commercialization do not affect the safety and effectiveness of the device.
The sponsor also performed a simulation to evaluate the impact of CGM sensor bias from the 10-day G7 CGM, used in the clinical study to support this device validation, and the 15-day G7 CGM, representing the intended CGM device for Smart Basal commercialization, on the glycemic safety and effectiveness of dose recommendations generated by this device. The validation considered systemic bias in estimated glucose values, which is a source of input error relevant to the Smart Basal algorithm’s operation. The validation results demonstrate that the Smart Basal algorithm maintains safe and effective dose recommendations when used with the 15-day G7 device.
## Software
Dexcom provided software documentation consistent with FDA Guidance for the Content of Premarket Submissions for Device Software Functions and consistent with software with a major level of concern. Software documentation was acceptable.
## Cybersecurity
Dexcom provided cybersecurity documentation consistent with FDA Guidance Cybersecurity in Medical Devices: Quality System Considerations and Content of Premarket Submissions. Cybersecurity documentation was acceptable.
## Human Factors
A human factors study was conducted to assess if representative users of the Smart Basal System can safely use the system under representative conditions. Study subjects were representative of the device’s intended use population (healthcare providers and patients with type 2 diabetes).
## VII Proposed Labeling:
The labeling supports the finding of substantial equivalence for this device.
## VIII Conclusion:
The submitted information in this premarket notification is complete and supports a substantial equivalence decision.
K252818 - Page 6 of 6
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.
