EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Enhancing Predictive Intelligence in Diabetes Risk Assessment Using Sensor-Aware Machine Learning with Multivariate Data Integrity Evaluation

Viswanatha Reddy Allugunti, Vijayaramaraju Poospati

International Journal on Advanced Science, Engineering and Information Technology · 25.10.2026 · 16 (5) · 1880-1888

DOI: 10.18517/ijaseit.16.5.14876 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal on Advanced Science, Engineering and Information Technology, online erschienen am 25.10.2026. Verfasst von Viswanatha Reddy Allugunti und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The prediction of diabetes mellitus (DM) risk is an important public health problem and targeting this prediction for the early identification of people who can benefit from preventive interventions. This study proposes a sensor-aware machine learning approach for diabetes risk assessment with introducing the module of Multivariate Data Integrity Evaluation (MDIE). The study relies on a sample of 495 subjects from the PIMA Indians Diabetes Database augmented with three simulated lifestyle-related variables: Average Daily Steps, Sleep Quality Score and Body Fat Percentage. In the MDIE module, Mahalanobis distance is used to detect multivariate anomalies and k-nearest neighbors (k-NN) imputation is used to treat anomalous feature values before classification. The outcome of this data set is tested with Logistic Regression (LR), Support Vector Machine (SVM) with a radial basis function kernel and a Gradient Boosting Machine (GBM). The results reported for the GBM with MDIE accuracy is 94.2% and the area under the receiver operating characteristic curve (AUC-ROC) is 0.97. Additionally, comparative results show that the classifiers were improved in terms of prediction after applying the MDIE processing. Glucose and BMI are prominent reported features, while Average Daily Steps and Body Fat Percentage are simulated lifestyle features. The results suggest that a multivariate approach to data-integrity assessment offers potential for the use of sensor-awareness machine learning in predicting diabetes. The sensor derived variables are however, simulated ones and thus against reality data obtained from wearable-sensors and other independent data should be used to validate generalizability and practical applicability.

Zitieren

APA

Allugunti, V. R., & Poospati, V. (2026). Enhancing Predictive Intelligence in Diabetes Risk Assessment Using Sensor-Aware Machine Learning with Multivariate Data Integrity Evaluation. International Journal on Advanced Science, Engineering and Information Technology, 16(5), 1880-1888. https://doi.org/10.18517/ijaseit.16.5.14876

Vancouver

Allugunti VR, Poospati V. Enhancing Predictive Intelligence in Diabetes Risk Assessment Using Sensor-Aware Machine Learning with Multivariate Data Integrity Evaluation. International Journal on Advanced Science, Engineering and Information Technology. 2026;16(5):1880-1888. doi:10.18517/ijaseit.16.5.14876

BibTeX
@article{Allugunti2026,
  author = {Allugunti, Viswanatha Reddy and Poospati, Vijayaramaraju},
  title = {Enhancing Predictive Intelligence in Diabetes Risk Assessment Using Sensor-Aware Machine Learning with Multivariate Data Integrity Evaluation},
  journal = {International Journal on Advanced Science, Engineering and Information Technology},
  year = {2026},
  volume = {16},
  number = {5},
  pages = {1880--1888},
  doi = {10.18517/ijaseit.16.5.14876}
}
Kostenlos starten →