Diabetes is the fourth leading cause of death in the world and one of the most common endocrine disorders. According to studies, Type 2 diabetes kills thousands of people around the world every year and imposes huge costs on societies in the form of surgeries and other treatment programs, as well as controlling complications and disability. Therefore, predicting and early diagnosis of this disease can greatly help governments and patients.

This dataset is the output of a Chinese research study conducted in 2016. It includes 1304 samples of patients who tested positive for diabetes, and the age of the participants ranges from 21 to 99 years old. The dataset was collected according to the indicators and standards of the World Health Organization, making it a reliable source for building diabetes diagnosis models. Researchers and healthcare professionals can use this dataset to train and test machine learning models to predict and diagnose diabetes in patients.

Features of Dataset: Age Gender BMI SBP (Systolic Blood Pressure) DBP (Diastolic Blood Pressure) FPG (Fasting Plasma Glucose) FFPG (Final Fasting Plasma Glucose) Cholesterol Triglyceride HDL (High-Density Lipoprotein) LDL (Low-Density Lipoprotein) ALT (Alanine Aminotransferase) BUN (Blood urea nitrogen) CCR (Creatinine Clearance) Smoking Status: (1: Current Smoker, 2: Ever Smoker, 3: Never Smoker) Drinking Status: (1: Current Drinker, 2: Ever Drinker, 3: Never Drinker) Family History of Diabetes: (1: Yes, 0: No) Diabetes

More details about dataset: The main dataset, without cleaning, is available at the following link: https://datadryad.org/stash/dataset/doi:10.5061/dryad.ft8750v. The main article corresponding to the dataset can be found at: https://doi.org/10.11.../bmjopen-2018-021768

Description: Welcome to the Diabetes Prediction Dataset, a valuable resource for researchers, data scientists, and medical professionals interested in the field of diabetes risk assessment and prediction. This dataset contains a diverse range of health-related attributes, meticulously collected to aid in the development of predictive models for identifying individuals at risk of diabetes. By sharing this dataset, we aim to foster collaboration and innovation within the data science community, leading to improved early diagnosis and personalized treatment strategies for diabetes.

Columns: 1. Id: Unique identifier for each data entry. 2. Pregnancies: Number of times pregnant. 3. Glucose: Plasma glucose concentration over 2 hours in an oral glucose tolerance test. 4. BloodPressure: Diastolic blood pressure (mm Hg). 5. SkinThickness: Triceps skinfold thickness (mm). 6. Insulin: 2-Hour serum insulin (mu U/ml). 7. BMI: Body mass index (weight in kg / height in m^2). 8. DiabetesPedigreeFunction: Diabetes pedigree function, a genetic score of diabetes. 9. Age: Age in years. 10. Outcome: Binary classification indicating the presence (1) or absence (0) of diabetes.

Utilize this dataset to explore the relationships between various health indicators and the likelihood of diabetes. You can apply machine learning techniques to develop predictive models, feature selection strategies, and data visualization to uncover insights that may contribute to more accurate risk assessments. As you embark on your journey with this dataset, remember that your discoveries could have a profound impact on diabetes prevention and management.

Please ensure that you adhere to ethical guidelines and respect the privacy of individuals represented in this dataset. Proper citation and recognition of this dataset's source are appreciated to promote collaboration and knowledge sharing.

Start your exploration of the Diabetes Prediction Dataset today and contribute to the ongoing efforts to combat diabetes through data-driven insights and innovations.

These datasets provide de-identified insurance data for diabetes. The data is provided by three managed care organizations in Allegheny County (Gateway Health Plan, Highmark Health, and UPMC) and represents their insured population for the 2015 and calendar years. Disclaimer: Users should be cautious of using administrative claims data as a measure of disease prevalence and interpreting trends over time, as data provided were collected for purposes other than surveillance. Limitations of these data include but are not limited to: misclassification, duplicate individuals, exclusion of individuals who did not seek care in past two years and those who are: uninsured, enrolled in plans not represented in the dataset, or were not enrolled in one of the represented plans for at least 90 days.

Diabetes is an opportune disease which has large wealth of data available and has with it huge complications. There is a need for a better and a more accurate approach in the diagnosis of the disease

The data were collected from the Iraqi society, as they data were acquired from the laboratory of Medical City Hospital and (the Specializes Center for Endocrinology and Diabetes-Al-Kindy Teaching Hospital). Patients' files were taken and data extracted from them and entered in to the database to construct the diabetes dataset. The data consist of medical information, laboratory analysis. The data attribute are: The data consist of medical information, laboratory analysis… etc. The data that have been entered initially into the system are: No. of Patient, Sugar Level Blood, Age, Gender, Creatinine ratio(Cr), Body Mass Index (BMI), Urea, Cholesterol (Chol), Fasting lipid profile, including total, LDL, VLDL, Triglycerides(TG) and HDL Cholesterol , HBA1C, Class (the patient's diabetes disease class may be Diabetic, Non-Diabetic, or Predict-Diabetic).

1) Data Introduction • The Diabetes Prediction Dataset is a dataset built for the purpose of predicting diabetes and analyzing related risk factors. It contains various characteristics such as demographics, lifestyle, and clinical measurements, so it can be used to predict a patient's risk of developing diabetes.

2) Data Utilization (1) Diabetes Prediction Dataset has characteristics that: • Key columns (characteristics) include a variety of clinical and lifestyle indicators related to diabetes, including age, gender, body mass index (BMI), blood pressure, blood sugar levels (Glucose), insulin, family history, and physical activity. (2) Diabetes Prediction Dataset can be used to: • Machine Learning/Deep Learning Model Development: It can be used to develop classification models (logistic regression, decision tree, random forest, neural network, etc.) that predict the risk of developing diabetes based on patient characteristics. • Data Analysis and Visualization: It is suitable for correlation analysis, risk factor derivation, Exploratory Data Analysis (EDA) and many other variables such as demographics, clinical figures, lifestyle, and more.

This is a source dataset for a Let's Get Healthy California indicator at "https://letsgethealthy.ca.gov/. This table displays the prevalence of diabetes in California. It contains data for California only. The data are from the California Behavioral Risk Factor Surveillance Survey (BRFSS). The California BRFSS is an annual cross-sectional health-related telephone survey that collects data about California residents regarding their health-related risk behaviors, chronic health conditions, and use of preventive services. The BRFSS is conducted by Public Health Survey Research Program of California State University, Sacramento under contract from CDPH. This prevalence rate does not include pre-diabetes, or gestational diabetes. This is based on the question: "Has a doctor, or nurse or other health professional ever told you that you have diabetes?" The sample size for 2014 was 8,832. NOTE: Denominator data and weighting was taken from the California Department of Finance, not U.S. Census. Values may therefore differ from what has been published in the national BRFSS data tables by the Centers for Disease Control and Prevention (CDC) or other federal agencies.

United States US: Diabetes Prevalence: % of Population Aged 20-79 data was reported at 10.790 % in 2017. United States US: Diabetes Prevalence: % of Population Aged 20-79 data is updated yearly, averaging 10.790 % from Dec 2017 (Median) to 2017, with 1 observations. United States US: Diabetes Prevalence: % of Population Aged 20-79 data remains active status in CEIC and is reported by World Bank. The data is categorized under Global Database’s USA – Table US.World Bank: Health Statistics. Diabetes prevalence refers to the percentage of people ages 20-79 who have type 1 or type 2 diabetes.; ; International Diabetes Federation, Diabetes Atlas.; Weighted average;

This refined dataset is originally based on the "Diabetes Dataset" uploaded by Ahlam Rashid in Mendeley Data. The link to the original dataset is: https://data.mendeley.com/datasets/wj9rwkp9c2/1. The original dataset contains a total of 1000 subjects divided into three classes: diabetic, non-diabetic, and predict-diabetic.

Among the 1000 subjects, 844 are diabetic, 103 are non-diabetic, and 53 are predict-diabetic, resulting in an extreme class imbalance. We found a total of 174 duplicate subjects in the original dataset, which we subsequently removed. After removing the duplicate subjects, there were 690 diabetic, 96 non-diabetic, and 40 predict-diabetic subjects remaining.

From the 826 unique subjects, we selected 128 diabetic, 96 non-diabetic, and 40 predict-diabetic subjects, resulting in a new dataset with a moderate class imbalance.

Decrease the percentage of people with Type 2 diabetes from 11.2% in 2014 to 10.1% by 2019.

Population-based county-level estimates for prevalence of DC were obtained from the Institute for Health Metrics and Evaluation (IHME) for the years 2004-2012 (16). DC prevalence rate was defined as the propor-tion of people within a county who had previously been diagnosed with diabetes (high fasting plasma glu-cose 126 mg/dL, hemoglobin A1c (HbA1c) of 6.5%, or diabetes diagnosis) but do not currently have high fasting plasma glucose or HbA1c for the period 2004-2012. DC prevalence estimates were calculated using a two-stage approach. The first stage used National Health and Nutrition Examination Survey (NHANES) data to predict high fasting plasma glucose (FPG) levels (≥126 mg/dL) and/or HbA1C levels (≥6.5% [48 mmol/mol]) based on self-reported demographic and behavioral characteristics (16). This model was then applied to Behavioral Risk Factor Surveillance System (BRFSS) data to impute high FPG and/or HbA1C status for each BRFSS respondent (16). The second stage used the imputed BRFSS data to fit a series of small area models, which were used to predict county-level prevalence of diabetes-related outcomes, including DC (16). The EQI was constructed for 2006-2010 for all US counties and is composed of five domains (air, water, built, land, and sociodemographic), each composed of variables to represent the environmental quality of that domain. Domain-specific EQIs were developed using principal components analysis (PCA) to reduce these variables within each domain while the overall EQI was constructed from a second PCA from these individual domains (L. C. Messer et al., 2014). To account for differences in environment across rural and urban counties, the overall and domain-specific EQIs were stratified by rural urban continuum codes (RUCCs) (U.S. Department of Agriculture, 2015). Results are reported as prevalence rate differences (PRD) with 95% confidence intervals (CIs) comparing the highest quintile/worst environmental quality to the lowest quintile/best environmental quality expo-sure metrics. PRDs are representative of the entire period of interest, 2004-2012. Due to availability of DC data and covariate data, not all counties were captured, however, the majority, 3134 of 3142 were utilized in the analysis. This dataset is not publicly accessible because: EPA cannot release personally identifiable information regarding living individuals, according to the Privacy Act and the Freedom of Information Act (FOIA). This dataset contains information about human research subjects. Because there is potential to identify individual participants and disclose personal information, either alone or in combination with other datasets, individual level data are not appropriate to post for public access. Restricted access may be granted to authorized persons by contacting the party listed. It can be accessed through the following means: Human health data are not available publicly. EQI data are available at: https://edg.epa.gov/data/Public/ORD/NHEERL/EQI. Format: Data are stored as csv files. This dataset is associated with the following publication: Jagai, J., A. Krajewski, K. Price, D. Lobdell, and R. Sargis. Diabetes control is associated with environmental quality in the USA. Endocrine Connections. BioScientifica Ltd., Bristol, UK, 10(9): 1018-1026, (2021).

South Africa is experiencing a rapidly growing diabetes epidemic that threatens its healthcare system. Research on the determinants of diabetes in South Africa receives considerable attention due to the lifestyle changes accompanying South Africa’s rapid urbanization since the fall of Apartheid. However, few studies have investigated how segments of the Black South African population, who continue to endure Apartheid’s institutional discriminatory legacy, experience this transition. This paper explores the association between individual and area-level socioeconomic status and diabetes prevalence, awareness, treatment, and control within a sample of Black South Africans aged 45 years or older in three municipalities in KwaZulu-Natal. Cross-sectional data were collected on 3,685 participants from February 2017 to February 2018. Individual-level socioeconomic status was assessed with employment status and educational attainment. Area-level deprivation was measured using the most recent South African Multidimensional Poverty Index scores. Covariates included age, sex, BMI, and hypertension diagnosis. The prevalence of diabetes was 23% (n = 830). Of those, 769 were aware of their diagnosis, 629 were receiving treatment, and 404 had their diabetes controlled. Compared to those with no formal education, Black South Africans with some high school education had increased diabetes prevalence, and those who had completed high school had lower prevalence of treatment receipt. Employment status was negatively associated with diabetes prevalence. Black South Africans living in more deprived wards had lower diabetes prevalence, and those residing in wards that became more deprived from 2001 to 2011 had a higher prevalence diabetes, as well as diabetic control. Results from this study can assist policymakers and practitioners in identifying modifiable risk factors for diabetes among Black South Africans to intervene on. Potential community-based interventions include those focused on patient empowerment and linkages to care. Such interventions should act in concert with policy changes, such as expanding the existing sugar-sweetened beverage tax.

Dataset Description: Several hundred rural African-American patients were included. The diabetes.csv file contains the raw data of all patients, including those with missing data. This can be used for descriptive statistics. The data dictionary to explain the columns can be found here: here and here

The Diabetes_Classification file was cleaned and manipulated. Any patient without a hemoglobin A1c was excluded. If their hemoglobin A1 c was 6.5 or greater they were labelled with diabetes = yes [column = "glyhb"]. Sixty patients out of 390 were found to be diabetic. A code book of the variables is included in one of the tabs. The goal is to use machine learning (classification algorithm) to predict diabetes based on demographic and laboratory variables. What are the strongest predictors? If you exclude glucose, how strong is the prediction?

SUMMARYThis analysis, designed and executed by Ribble Rivers Trust, identifies areas across England with the greatest levels of diabetes mellitus in persons (aged 17+). Please read the below information to gain a full understanding of what the data shows and how it should be interpreted.ANALYSIS METHODOLOGYThe analysis was carried out using Quality and Outcomes Framework (QOF) data, derived from NHS Digital, relating to diabetes mellitus in persons (aged 17+).This information was recorded at the GP practice level. However, GP catchment areas are not mutually exclusive: they overlap, with some areas covered by 30+ GP practices. Therefore, to increase the clarity and usability of the data, the GP-level statistics were converted into statistics based on Middle Layer Super Output Area (MSOA) census boundaries.The percentage of each MSOA’s population (aged 17+) with diabetes mellitus was estimated. This was achieved by calculating a weighted average based on:The percentage of the MSOA area that was covered by each GP practice’s catchment areaOf the GPs that covered part of that MSOA: the percentage of registered patients that have that illness The estimated percentage of each MSOA’s population with diabetes mellitus was then combined with Office for National Statistics Mid-Year Population Estimates (2019) data for MSOAs, to estimate the number of people in each MSOA with depression, within the relevant age range.Each MSOA was assigned a relative score between 1 and 0 (1 = worst, 0 = best) based on:A) the PERCENTAGE of the population within that MSOA who are estimated to have diabetes mellitusB) the NUMBER of people within that MSOA who are estimated to have diabetes mellitusAn average of scores A & B was taken, and converted to a relative score between 1 and 0 (1= worst, 0 = best). The closer to 1 the score, the greater both the number and percentage of the population in the MSOA that are estimated to have diabetes mellitus, compared to other MSOAs. In other words, those are areas where it’s estimated a large number of people suffer from diabetes mellitus, and where those people make up a large percentage of the population, indicating there is a real issue with diabetes mellitus within the population and the investment of resources to address that issue could have the greatest benefits.LIMITATIONS1. GP data for the financial year 1st April 2018 – 31st March 2019 was used in preference to data for the financial year 1st April 2019 – 31st March 2020, as the onset of the COVID19 pandemic during the latter year could have affected the reporting of medical statistics by GPs. However, for 53 GPs (out of 7670) that did not submit data in 2018/19, data from 2019/20 was used instead. Note also that some GPs (997 out of 7670) did not submit data in either year. This dataset should be viewed in conjunction with the ‘Health and wellbeing statistics (GP-level, England): Missing data and potential outliers’ dataset, to determine areas where data from 2019/20 was used, where one or more GPs did not submit data in either year, or where there were large discrepancies between the 2018/19 and 2019/20 data (differences in statistics that were > mean +/- 1 St.Dev.), which suggests erroneous data in one of those years (it was not feasible for this study to investigate this further), and thus where data should be interpreted with caution. Note also that there are some rural areas (with little or no population) that do not officially fall into any GP catchment area (although this will not affect the results of this analysis if there are no people living in those areas).2. Although all of the obesity/inactivity-related illnesses listed can be caused or exacerbated by inactivity and obesity, it was not possible to distinguish from the data the cause of the illnesses in patients: obesity and inactivity are highly unlikely to be the cause of all cases of each illness. By combining the data with data relating to levels of obesity and inactivity in adults and children (see the ‘Levels of obesity, inactivity and associated illnesses: Summary (England)’ dataset), we can identify where obesity/inactivity could be a contributing factor, and where interventions to reduce obesity and increase activity could be most beneficial for the health of the local population.3. It was not feasible to incorporate ultra-fine-scale geographic distribution of populations that are registered with each GP practice or who live within each MSOA. Populations might be concentrated in certain areas of a GP practice’s catchment area or MSOA and relatively sparse in other areas. Therefore, the dataset should be used to identify general areas where there are high levels of diabetes mellitus, rather than interpreting the boundaries between areas as ‘hard’ boundaries that mark definite divisions between areas with differing levels of diabetes mellitus.TO BE VIEWED IN COMBINATION WITH:This dataset should be viewed alongside the following datasets, which highlight areas of missing data and potential outliers in the data:Health and wellbeing statistics (GP-level, England): Missing data and potential outliersLevels of obesity, inactivity and associated illnesses (England): Missing dataDOWNLOADING THIS DATATo access this data on your desktop GIS, download the ‘Levels of obesity, inactivity and associated illnesses: Summary (England)’ dataset.DATA SOURCESThis dataset was produced using:Quality and Outcomes Framework data: Copyright © 2020, Health and Social Care Information Centre. The Health and Social Care Information Centre is a non-departmental body created by statute, also known as NHS Digital.GP Catchment Outlines. Copyright © 2020, Health and Social Care Information Centre. The Health and Social Care Information Centre is a non-departmental body created by statute, also known as NHS Digital. Data was cleaned by Ribble Rivers Trust before use.COPYRIGHT NOTICEThe reproduction of this data must be accompanied by the following statement:© Ribble Rivers Trust 2021. Analysis carried out using data that is: Copyright © 2020, Health and Social Care Information Centre. The Health and Social Care Information Centre is a non-departmental body created by statute, also known as NHS Digital.CaBA HEALTH & WELLBEING EVIDENCE BASEThis dataset forms part of the wider CaBA Health and Wellbeing Evidence Base.

Find data on pediatric diabetes in Massachusetts. This dataset contains information on the number of cases and prevalence of Type 1 and Type 2 diabetes among students, grades K-8, in Massachusetts.

South Africa ZA: Diabetes Prevalence: % of Population Aged 20-79 data was reported at 5.520 % in 2017. South Africa ZA: Diabetes Prevalence: % of Population Aged 20-79 data is updated yearly, averaging 5.520 % from Dec 2017 (Median) to 2017, with 1 observations. South Africa ZA: Diabetes Prevalence: % of Population Aged 20-79 data remains active status in CEIC and is reported by World Bank. The data is categorized under Global Database’s South Africa – Table ZA.World Bank: Health Statistics. Diabetes prevalence refers to the percentage of people ages 20-79 who have type 1 or type 2 diabetes.; ; International Diabetes Federation, Diabetes Atlas.; Weighted average;

Archived as of 6/26/2025: The datasets will no longer receive updates but the historical data will continue to be available for download. This dataset provides information related to the services of diabetes patients. It contains information about the total number of patients, total number of claims, and dollar amount paid, grouped by recipient zip code. Restricted to claims with service date between 01/2012 to 12/2017. Diabetic patients are identified as diagnosed with the following ICD codes: E110, E112, E114, E115, E116, E118, E119, 25000, 25002, 25010, 25012, 25020, 2522, 25030, 25040, 25040, 25042, 25050, 25052, 25060, 25062, 25070, 25072, 25080, 25082, 25090, 25092, and O241 between 2010 to 2017. Provider is billing provider. Provider with an NPI are only considered. All types of claims except dental are considered. This data is for research purposes and is not intended to be used for reporting. Due to differences in geographic aggregation, time period considerations, and units of analysis, these numbers may differ from those reported by FSSA.

The percentage of people with diabetes who have received nine care processes. Current version updated: Mar-17 Next version due: Mar-18

Population-based county-level estimates for diagnosed (DDP), undiagnosed (UDP), and total diabetes prevalence (TDP) were acquired from the Institute for Health Metrics and Evaluation (IHME) for the years 2004-2012 (Evaluation 2017). Prevalence estimates were calculated using a two-stage approach. The first stage used National Health and Nutrition Examination Survey (NHANES) data to predict high fasting plasma glucose (FPG) levels (≥126 mg/dL) and/or hemoglobin A1C (HbA1C) levels (≥6.5% [48 mmol/mol]) based on self-reported demographic and behavioral characteristics (Dwyer-Lindgren, Mackenbach et al. 2016). This model was then applied to Behavioral Risk Factor Surveillance System (BRFSS) data to impute high FPG and/or A1C status for each BRFSS respondent (Dwyer-Lindgren, Mackenbach et al. 2016). The second stage used the imputed BRFSS data to fit a series of small area models, which were used to predict the county-level prevalence of each of the diabetes-related outcomes (Dwyer-Lindgren, Mackenbach et al. 2016). Diagnosed diabetes was defined as the proportion of adults (age 20+ years) who reported a previous diabetes diagnosis, represented as an age-standardized prevalence percentage. Undiagnosed diabetes was defined as proportion of adults (age 20+ years) who have a high FPG or HbA1C but did not report a previous diagnosis of diabetes. Total diabetes was defined as the proportion of adults (age 20+ years) who reported a previous diabetes diagnosis and/or had a high FPG/HbA1C. The age-standardized diabetes prevalence (%) was used as the outcome. The EQI was constructed for 2000-2005 for all US counties and is composed of five domains (air, water, built, land, and sociodemographic), each composed of variables to represent the environmental quality of that domain. Domain-specific EQIs were developed using principal components analysis (PCA) to reduce these variables within each domain while the overall EQI was constructed from a second PCA from these individual domains (L. C. Messer et al., 2014). To account for differences in environment across rural and urban counties, the overall and domain-specific EQIs were stratified by rural urban continuum codes (RUCCs) (U.S. Department of Agriculture, 2015). This dataset is not publicly accessible because: EPA cannot release personally identifiable information regarding living individuals, according to the Privacy Act and the Freedom of Information Act (FOIA). This dataset contains information about human research subjects. Because there is potential to identify individual participants and disclose personal information, either alone or in combination with other datasets, individual level data are not appropriate to post for public access. Restricted access may be granted to authorized persons by contacting the party listed. It can be accessed through the following means: Human health data are not available publicly. EQI data are available at: https://edg.epa.gov/data/Public/ORD/NHEERL/EQI. Format: Data are stored as csv files. This dataset is associated with the following publication: Jagai, J., A. Krajewski, S. Shaikh, D. Lobdell, and R. Sargis. Association between environmental quality and diabetes in the U.S.A.. Journal of Diabetes Investigation. John Wiley & Sons, Inc., Hoboken, NJ, USA, 11(2): 315-324, (2020).

IntroductionType 2 diabetes (T2D) is a growing public health burden throughout the world. Many people looking for information on how to prevent T2D will search on diabetes websites. Multiple dietary factors have a significant association with T2D risk, such as high intake of added sugars, refined carbohydrates, saturated fat, and red meat or processed meat; and decreased intake of dietary fiber, and fruits/vegetables. Despite this dietary information being available in the scientific literature, it is unclear whether this information is available in gray literature (websites).ObjectiveIn this study, we evaluate the use of specific terms from diabetes websites that are significantly associated with causes/risk factors and preventions for T2D from three term categories: (A) dietary factors, (B) nondietary nongenetic (lifestyle-associated) factors, and (C) genetic (non-modifiable) factors. We also evaluate the effect of website type (business, government, nonprofit) on term usage among websites.MethodsWe used web scraping and coding tools to quantify the use of specific terms from 73 diabetes websites. To determine the effect of term category and website type on the usage of specific terms among 73 websites, a repeated measures general linear model was performed.ResultsWe found that dietary risk factors that are significantly associated with T2D (e.g., sugar, processed carbohydrates, dietary fat, fruits/vegetables, fiber, processed meat/red meat) were mentioned in significantly fewer websites than either nondietary nongenetic factors (e.g., obesity, physical activity, dyslipidemia, blood pressure) or genetic factors (age, family history, ethnicity). Among websites that provided “eat healthy” guidance, one third provided zero dietary factors associated with type 2 diabetes, and only 30% provided more than two specific dietary factors associates with type 2 diabetes. We also observed that mean percent usage of all terms associated with T2D causes/risk factors and preventions was significantly lower among government websites compared to business websites and nonprofit websites.ConclusionDiabetes websites need to increase their usage of dietary factors when discussing causes/risk factors and preventions for T2D; as dietary factors are modifiable and strongly associated with all nondietary nongenetic risk factors, in addition to T2D risk.

T1DiabetesGranada

A longitudinal multi-modal dataset of type 1 diabetes mellitus

Documented by:

Rodriguez-Leon, C., Aviles-Perez, M. D., Banos, O., Quesada-Charneco, M., Lopez-Ibarra, P. J., Villalonga, C., & Munoz-Torres, M. (2023). T1DiabetesGranada: a longitudinal multi-modal dataset of type 1 diabetes mellitus. Scientific Data, 10(1), 916. https://doi.org/10.1038/s41597-023-02737-4

Background

Type 1 diabetes mellitus (T1D) patients face daily difficulties in keeping their blood glucose levels within appropriate ranges. Several techniques and devices, such as flash glucose meters, have been developed to help T1D patients improve their quality of life. Most recently, the data collected via these devices is being used to train advanced artificial intelligence models to characterize the evolution of the disease and support its management. The main problem for the generation of these models is the scarcity of data, as most published works use private or artificially generated datasets. For this reason, this work presents T1DiabetesGranada, a open under specific permission longitudinal dataset that not only provides continuous glucose levels, but also patient demographic and clinical information. The dataset includes 257780 days of measurements over four years from 736 T1D patients from the province of Granada, Spain. This dataset progresses significantly beyond the state of the art as one the longest and largest open datasets of continuous glucose measurements, thus boosting the development of new artificial intelligence models for glucose level characterization and prediction.

Data Records

The data are stored in four comma-separated values (CSV) files which are available in T1DiabetesGranada.zip. These files are described in detail below.

Patient_info.csv

Patient_info.csv is the file containing information about the patients, such as demographic data, start and end dates of blood glucose level measurements and biochemical parameters, number of biochemical parameters or number of diagnostics. This file is composed of 736 records, one for each patient in the dataset, and includes the following variables:

Patient_ID – Unique identifier of the patient. Format: LIB19XXXX.

Sex – Sex of the patient. Values: F (for female), masculine (for male)

Birth_year – Year of birth of the patient. Format: YYYY.

Initial_measurement_date – Date of the first blood glucose level measurement of the patient in the Glucose_measurements.csv file. Format: YYYY-MM-DD.

Final_measurement_date – Date of the last blood glucose level measurement of the patient in the Glucose_measurements.csv file. Format: YYYY-MM-DD.

Number_of_days_with_measures – Number of days with blood glucose level measurements of the patient, extracted from the Glucose_measurements.csv file. Values: ranging from 8 to 1463.

Number_of_measurements – Number of blood glucose level measurements of the patient, extracted from the Glucose_measurements.csv file. Values: ranging from 400 to 137292.

Initial_biochemical_parameters_date – Date of the first biochemical test to measure some biochemical parameter of the patient, extracted from the Biochemical_parameters.csv file. Format: YYYY-MM-DD.

Final_biochemical_parameters_date – Date of the last biochemical test to measure some biochemical parameter of the patient, extracted from the Biochemical_parameters.csv file. Format: YYYY-MM-DD.

Number_of_biochemical_parameters – Number of biochemical parameters measured on the patient, extracted from the Biochemical_parameters.csv file. Values: ranging from 4 to 846.

Number_of_diagnostics – Number of diagnoses realized to the patient, extracted from the Diagnostics.csv file. Values: ranging from 1 to 24.

Glucose_measurements.csv

Glucose_measurements.csv is the file containing the continuous blood glucose level measurements of the patients. The file is composed of more than 22.6 million records that constitute the time series of continuous blood glucose level measurements. It includes the following variables:

Patient_ID – Unique identifier of the patient. Format: LIB19XXXX.

Measurement_date – Date of the blood glucose level measurement. Format: YYYY-MM-DD.

Measurement_time – Time of the blood glucose level measurement. Format: HH:MM:SS.

Measurement – Value of the blood glucose level measurement in mg/dL. Values: ranging from 40 to 500.

Biochemical_parameters.csv

Biochemical_parameters.csv is the file containing data of the biochemical tests performed on patients to measure their biochemical parameters. This file is composed of 87482 records and includes the following variables:

Patient_ID – Unique identifier of the patient. Format: LIB19XXXX.

Reception_date – Date of receipt in the laboratory of the sample to measure the biochemical parameter. Format: YYYY-MM-DD.

Name – Name of the measured biochemical parameter. Values: 'Potassium', 'HDL cholesterol', 'Gammaglutamyl Transferase (GGT)', 'Creatinine', 'Glucose', 'Uric acid', 'Triglycerides', 'Alanine transaminase (GPT)', 'Chlorine', 'Thyrotropin (TSH)', 'Sodium', 'Glycated hemoglobin (Ac)', 'Total cholesterol', 'Albumin (urine)', 'Creatinine (urine)', 'Insulin', 'IA ANTIBODIES'.

Value – Value of the biochemical parameter. Values: ranging from -4.0 to 6446.74.

Diagnostics.csv

Diagnostics.csv is the file containing diagnoses of diabetes mellitus complications or other diseases that patients have in addition to type 1 diabetes mellitus. This file is composed of 1757 records and includes the following variables:

Patient_ID – Unique identifier of the patient. Format: LIB19XXXX.

Code – ICD-9-CM diagnosis code. Values: subset of 594 of the ICD-9-CM codes (https://www.cms.gov/Medicare/Coding/ICD9ProviderDiagnosticCodes/codes).

Description – ICD-9-CM long description. Values: subset of 594 of the ICD-9-CM long description (https://www.cms.gov/Medicare/Coding/ICD9ProviderDiagnosticCodes/codes).

Technical Validation

Blood glucose level measurements are collected using FreeStyle Libre devices, which are widely used for healthcare in patients with T1D. Abbott Diabetes Care, Inc., Alameda, CA, USA, the manufacturer company, has conducted validation studies of these devices concluding that the measurements made by their sensors compare to YSI analyzer devices (Xylem Inc.), the gold standard, yielding results of 99.9% of the time within zones A and B of the consensus error grid. In addition, other studies external to the company concluded that the accuracy of the measurements is adequate.

Moreover, it was also checked in most cases the blood glucose level measurements per patient were continuous (i.e. a sample at least every 15 minutes) in the Glucose_measurements.csv file as they should be.

Usage Notes

For data downloading, it is necessary to be authenticated on the Zenodo platform, accept the Data Usage Agreement and send a request specifying full name, email, and the justification of the data use. This request will be processed by the Secretary of the Department of Computer Engineering, Automatics, and Robotics of the University of Granada and access to the dataset will be granted.

The files that compose the dataset are CSV type files delimited by commas and are available in T1DiabetesGranada.zip. A Jupyter Notebook (Python v. 3.8) with code that may help to a better understanding of the dataset, with graphics and statistics, is available in UsageNotes.zip.

Graphs_and_stats.ipynb

The Jupyter Notebook generates tables, graphs and statistics for a better understanding of the dataset. It has four main sections, one dedicated to each file in the dataset. In addition, it has useful functions such as calculating the patient age, deleting a patient list from a dataset file and leaving only a patient list in a dataset file.

Code Availability

The dataset was generated using some custom code located in CodeAvailability.zip. The code is provided as Jupyter Notebooks created with Python v. 3.8. The code was used to conduct tasks such as data curation and transformation, and variables extraction.

Original_patient_info_curation.ipynb

In the Jupyter Notebook is preprocessed the original file with patient data. Mainly irrelevant rows and columns are removed, and the sex variable is recoded.

Glucose_measurements_curation.ipynb

In the Jupyter Notebook is preprocessed the original file with the continuous glucose level measurements of the patients. Principally rows without information or duplicated rows are removed and the variable with the timestamp is transformed into two new variables, measurement date and measurement time.

Biochemical_parameters_curation.ipynb

In the Jupyter Notebook is preprocessed the original file with patient data of the biochemical tests performed on patients to measure their biochemical parameters. Mainly irrelevant rows and columns are removed and the variable with the name of the measured biochemical parameter is translated.

Diagnostic_curation.ipynb

In the Jupyter Notebook is preprocessed the original file with patient data of the diagnoses of diabetes mellitus complications or other diseases that patients have in addition to T1D.

Get_patient_info_variables.ipynb

In the Jupyter Notebook it is coded the feature extraction process from the files Glucose_measurements.csv, Biochemical_parameters.csv and Diagnostics.csv to complete the file Patient_info.csv. It is divided into six sections, the first three to extract the features from each of the mentioned files and the next three to add the extracted features to the resulting new file.

Data Usage Agreement

The conditions for use are as follows:

You confirm that you will not attempt to re-identify research participants for any reason, including for re-identification theory research.

You commit to keeping the T1DiabetesGranada dataset confidential and secure and will not redistribute data or Zenodo account credentials.

You will require