Salami Safran Olugbenga

Salami Safran Olugbenga

Auto-generated profile, based on publicly available information

Salami Safran Olugbenga

Faculty of Engineering, Universiti Putra Malaysia (UPM), Serdang, Selangor, Malaysia

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
5 August 2025 in IEEE Access
Verified
Journal Article
Open access
SOSalami Safran Olugbenga
Salami Safran Olugbenga
AMAashiq M.N.M.
Aashiq M.N.M.
Siti Anom AhmadSiti Anom Ahmad
Siti Anom Ahmad
MTMaw Pin Tan
Maw Pin Tan
YZYuanyuan Zhao
Yuanyuan Zhao
FRFakhrul Zaman Rokhani
Fakhrul Zaman Rokhani
Cognitive frailty (CF), defined as the co-existence of physical frailty and cognitive impairment without dementia, has two subtypes: reversible cognitive frailty (RCF) and potentially reversible cognitive frailty (PRCF). This study aimed to develop machine learning models for the early prediction of CF by incorporating both RCF and PRCF into a unified framework. Data from 2,173 participants were sourced from the AGELESS and MELoR cohort studies, including clinical, blood, urine, and health status information. Participants were grouped into six CF categories based on the FRAIL scale and MoCA scores from 2020 and 2022. Seven machine learning models, SVM, LR, KNN, RF, CART, LDA, and GNB, were trained using clinical, blood, and urine datasets. Clinical variables outperformed other data types, with SVM, LR, and CART models achieving 95% accuracy and AUC scores of 1.0, while blood- and urine-based models showed lower performance (AUCs of 0.8 and 0.89, respectively). F1 scores and ROC analysis confirmed the robustness of the clinical models, and cross-validation showed consistent performance on unseen data, indicating no overfitting. These findings highlight the superior predictive power of clinical variables for identifying both subtypes of CF, supporting their potential for use in early, accurate diagnosis. Nonetheless, external validation using independent cohorts is recommended to ensure broader applicability.