T1DiabetesGranada: A longitudinal multimodal dataset of type 1 diabetes.

ethical approval

This study was reviewed and approved by the Biomedical Research Ethics Committee of the Province of Granada (CEIm/CEI GRANADA). Protocol Code: K134665CRL, Ethics Portal Code: 0698-N-21. Data has been approved for publication under a data use agreement.

Participant onboarding

Participants had to be patients of the Endocrinology and Nutrition Clinical Unit of the San Cecilio University Hospital in Granada, Spain. They were T1D patients and had to be selected as T1D wearers. freestyle libre Device (Abbott Diabetes Care, Inc., Alameda, California, USA). The study began on January 6, 2018 and ended on March 21, 2022, with 736 patients ultimately enrolled.

Enrollment in this study began when patients were informed that they had been selected for FGM because they met the necessary eligibility criteria. Patients then visited the clinical endocrinology and nutrition department and were explained how the sensor worked, its potential in T1D management, and what to do with the data collected. During this visit, patients were asked to consent to participate in the study and informed that their data may be shared anonymously. A training session followed, during which patients learned how to wear the device, operate it, and upload data to the cloud platform. To upload data, patients had to register in the system and consent to their BGL measurements and demographic information being used for research purposes. They agreed to access, use, and share their anonymized data. If participants wished to withdraw from the study, they had to unenroll from the system or make a specific request. No further data were collected after that point. However, the anonymized collected data was retained. In any case, no participants withdrew from the study.

Data collection

The most commonly used FGM devices during research were: freestyle libre 2the first version, freestyle libre, was also used in some cases at the beginning of the study. Both versions of the device are very similar and the manufacturers are: Abbott Diabetes Care, Inc., Alameda, California, USA^{twenty four}. These devices have a small needle-shaped sensor that measures glucose levels in the interstitial fluid when introduced into the tissue. Each device has a useful lifespan of 14 days, during which time the battery does not need to be charged or otherwise operated. After these 14 days, you will need to replace your device with a new one.

Interstitial fluid glucose measurements are recorded at 15 minute intervals. These measurements are stored in the device’s memory and can store up to 8 hours of data. You must scan him for FGM using a Near Field Communication (NFC) device, either on your cell phone or mobile phone, before these 8 hours have elapsed. FreeStyle Libre Reader. Once the scan is done, the data will be copied from his FGM device to the NFC device. Also, each time a patient performs an NFC scan, her current BGL value is added to the data as an additional measurement point.Then, when the NFC device used to collect measurements is connected to the Internet, the data is rib review cloud platform.

In addition to BGL, demographic and clinical information from patients was also collected in this study. When a patient first visits the Endocrinology and Nutrition Unit, data such as date of birth, history of other illnesses, home address, and contact phone number are collected. For privacy reasons and its relevance to her T1D, the dataset only includes year of birth and additional diagnoses of other diseases. The endocrinologist will take into account each patient’s condition and schedule the following visits (for example, every 3 months or every 6 months) and will take several days before the consultation to measure biochemical parameters. will request a series of biochemical tests. Values ​​of patient biochemical parameters obtained during the study period were also included as part of the dataset.

Data preparation

An independent technician from the Information Technology Service of San Cecilio University Hospital in Granada was appointed to carry out the data anonymization process. They removed information that was sensitive to the patients and irrelevant to the study, such as names, emails, and medical record numbers. In addition, we assigned each patient a unique identifier to ensure that the patient’s identity was not revealed.

Several basic data cleaning tasks were performed on the anonymized data, including removing duplicate rows, removing rows with relevant missing values, and removing irrelevant or empty columns. Additionally, column names were translated from Spanish (local language) to English, and some variable values, such as gender and names of biochemical parameters, were recoded to English. Patients’ diagnoses were associated with codes and descriptions according to the standards of the Spanish Government Ministry of Health.^{twenty five}Therefore, the code is mapped to the equivalent English version of the standard.²⁶ It was conducted. Additionally, the values ​​of some variables have been reformatted. For example, date fields in DD-MM-YYYY format were converted to YYYYY-MM-DD format to optimize tasks such as sorting. Finally, all files comprising the dataset were sorted by patient ID and date (if available).