(161 days)
The SweetSpot Diabetes Data Management Service is intended for use in clinical settings by both patients and healthcare professionals, to assist in the review, analysis, and evaluation of blood glucose test results by the clinician to support effective diabetes management. It is intended for use as an accessory to blood glucose meters with data interface capabilities.
The SweetSpot Service allows patients or healthcare professionals to download data from blood glucose meters (BGM) and generates a report from the downloaded data. which is delivered to the healthcare professional for use in patient management. The Service is comprised of three different types of data retrieval stations, a Fetch Utility, a data processing and storage platform, report generation software, and an information delivery service.
The first type of data retrieval station is a Front-Office Kiosk - a dedicated off-the-shelf computer that a patient uses to download the data from their devices (if they download from multiple devices the Service will consolidate that data into a single report). A device-specific cable is required to connect the BGM to the Kiosk for data download. Once the data is downloaded, it is processed into a report and delivered according to the specific clinic's needs and workflow.
The second type of data retrieval station is a Back-Office Kiosk - a dedicated off-theshelf computer that healthcare professionals use to download patient devices and/or monitor reports. This kiosk is also configured for the workflow needs of the clinic.
The third type of data retrieval station is a Back-Office Web Application – a standalone application used by healthcare professionals that does not require a dedicated computer. Each customer uses a specific instance of the web application through a unique URL. This launches a proprietary web application configured for that specific clinic's needs and workflows.
The SweetSpot Fetch Utility retrieves data from various BGM manufacturers' devices and includes device drivers for multiple manufacturers. The Fetch Utility is centrally updated and maintained. When any version of the SweetSpot Services retrieval stations - in any setting - is directed to perform data retrieval, the Service ensures the most upto-date version of the Fetch Utility is used.
The SweetSpot Service is primarily web-based and is delivered using a software-as-aservice (SaaS) model. All data storage and processing takes place on remotely hosted virtualized computing resources on the Internet, often referred to as "cloud computing".
Here's a breakdown of the acceptance criteria and study information for the SweetSpot Diabetes Data Management Service, based on the provided text:
Acceptance Criteria and Reported Device Performance
Note: The provided document is a 510(k) summary, which focuses on demonstrating substantial equivalence to a predicate device rather than presenting detailed acceptance criteria and a formal study report for each specific performance metric with pre-defined thresholds. The "acceptance criteria" are implied by the successful outcomes of the user evaluations and software verification/validation.
| Acceptance Criteria (Implied) | Reported Device Performance |
|---|---|
| Ease of Use (Lay Users - Front-Office Kiosk) | 98% of users thought the Service was easy to use (4.9 average rating on a 1-5 scale, 5 being "Strongly Agree"). |
| Increased Likelihood of Meter Use (Lay Users) | 90% of users thought they would be more likely to bring their meters to appointments if the Service was available (ratings of 4.6 and 4.5). |
| Meter Download Success Rate (Lay Users - Front-Office Kiosk) | 97.6% and 95.3% success rates (96.5% overall success rate). |
| Improvement in Workflow/Efficiency (HCPs - Back-Office Kiosk) | HCPs preferred the consistency, simplicity/ease of use, and speed of the SweetSpot download process to their previous processes; preferred one report format. HCPs thought these features would improve workflow, decrease meter download time, simplify decision-making, reduce training time, and reduce the chance for mistakes in data management. This is a qualitative assessment rather than a quantitative metric. |
| Accuracy of Data Fetch (Lay Users vs. Manufacturer Software) | 100% agreement between download files obtained by lay users (SweetSpot Fetch) and SweetSpot employee (manufacturer's software). |
| Software Functionality (Verification & Validation) | Software verification and validation testing showed that the SweetSpot Service performs as designed. (No specific metrics or thresholds provided). |
Study Details
2. Sample sizes used for the test set and the data provenance:
- Lay User Evaluation (Front-Office Kiosk):
- Sample Size: 88 respondents (users)
- Data Provenance: Prospective. Performed by typical outpatients at two different typical sites of use. The specific country is not mentioned, but given the context of a US FDA submission, it can be inferred to be within the United States.
- HCP User Evaluation (Back-Office Kiosk):
- Sample Size: 11 respondents (healthcare professionals)
- Data Provenance: Prospective. Performed at one typical site of use as part of a larger research study. Country inferred to be the United States.
- Accuracy of Fetch Process:
- Sample Size: Not explicitly stated, but implies a comparison for an unspecified number of downloads/BGMs involving lay users and a SweetSpot employee.
- Data Provenance: Prospective, at one typical site of use. Country inferred to be the United States.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- For "Accuracy of Fetch Process": The ground truth was established by comparing the data downloaded via the SweetSpot Fetch Utility (by lay users) with data downloaded from the same BGM using the BGM manufacturer's own download software (by a SweetSpot employee). This implies the manufacturer's software acts as the de facto "ground truth" reference, operated by an employee who likely had expertise in operating such devices. The specific qualifications of the SweetSpot employee are not detailed beyond "SweetSpot employee."
- For User Evaluations: The "ground truth" was the users' own perceptions and experiences, as captured in surveys, rather than an external expert-established truth.
4. Adjudication method for the test set:
- User Evaluations: No formal adjudication method involving multiple experts is described. User feedback was collected via surveys.
- Accuracy of Fetch Process: The comparison was direct ("100% agreement"). This suggests a binary comparison for each data point and no need for "adjudication" in the sense of resolving discrepancies between multiple expert opinions.
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 is described. The device is a data management service, not an AI for image interpretation or diagnosis that would typically involve "readers" in the context of MRMC studies. The evaluations focused on usability and data integrity, not diagnostic accuracy improvement.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
- The "Accuracy of Fetch Process" study can be considered a form of standalone performance evaluation for the data retrieval component. It assessed the algorithm's ability to accurately retrieve data without human intervention altering the data content itself, comparing it to a trusted reference (manufacturer's software).
- The software verification and validation also represent standalone testing of the algorithm's functionality and performance as designed.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.):
- For "Accuracy of Fetch Process": The ground truth was established by the output of the BGM manufacturer's own download software, considered the authoritative source for the BGM data.
- For User Evaluations: The ground truth was the self-reported user experience and perceptions (e.g., ease of use, likelihood of meter use).
8. The sample size for the training set:
- The document does not mention a "training set" in the context of machine learning or AI. The SweetSpot Service is described as a data management service, implying it facilitates data transfer and presentation, but not necessarily using a machine learning model that would require a dedicated training set. The software was likely developed and refined through standard software development and testing cycles rather than ML model training.
9. How the ground truth for the training set was established:
- As no "training set" is described for an ML model, this question is not applicable based on the provided text.
§ 862.1345 Glucose test system.
(a)
Identification. A glucose test system is a device intended to measure glucose quantitatively in blood and other body fluids. Glucose measurements are used in the diagnosis and treatment of carbohydrate metabolism disorders including diabetes mellitus, neonatal hypoglycemia, and idiopathic hypoglycemia, and of pancreatic islet cell carcinoma.(b)
Classification. Class II (special controls). The device, when it is solely intended for use as a drink to test glucose tolerance, is exempt from the premarket notification procedures in subpart E of part 807 of this chapter subject to the limitations in § 862.9.